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

net/publication/282731582

THE FACTORS AFFECTING THE GOLD MARKET IN VIETNAM

Thesis · May 2014

CITATION READS

1 2,718

1 author:

Tram Ngoc Dang

Université de Fribourg
1 PUBLICATION 1 CITATION

SEE PROFILE

All content following this page was uploaded by Tram Ngoc Dang on 11 October 2015.

The user has requested enhancement of the downloaded file.

 THE MINISTRY OF EDUCATION AND TRAINING OF
VIETNAM
THE UNIVERSITY OF ECONOMICS OF HO CHI MINH CITY
……………..……………….

THE DEPARTMENT OF FINANCE

GRADUATION THESIS

THE FACTORS AFFECTING THE GOLD

MARKET IN VIETNAM
Supervisor : Quach Doanh Nghiep, MA
Student : Dang Ngoc Tram
Class : TC06
Course : 2010 - 2014
Student ID : 31101020904

Ho Chi Minh City - 2014

 The factors affecting the gold market in Vietnam
March 29, 2014

TABLE OF CONTENTS
LIST OF TABLES ..................................................................................................................... 4

LIST OF CHARTS ..................................................................................................................... 4

LIST OF PHOTOGRAPH ......................................................................................................... 4

LIST OF APPENDIXED FIGURES......................................................................................... 4

ABSTRACT ................................................................................................................................. 4
CHAPTER 1 ................................................................................................................................. 6
INTRODUCTION ........................................................................................................................ 6
1.1. The rationale of the thesis ............................................................................................... 6
1.2. The aim and scope of the thesis ...................................................................................... 6
1.3. The study methods of the thesis...................................................................................... 7
1.3.1. The pattern ............................................................................................................... 7
1.3.2. The data .................................................................................................................... 7
1.3.3. The testing methods ................................................................................................. 7
1.4. The structure of the thesis ............................................................................................... 7
CHAPTER 2 ................................................................................................................................. 8
LITERATURE REVIEWS .......................................................................................................... 8
2.1. Study approaches ............................................................................................................ 8
2.1.1. The macro-oriented approach .................................................................................. 8
2.1.2. The speculative approach ......................................................................................... 9
2.1.3. The inflation-hedging approach ............................................................................. 10
2.2. The available experimental evidence ........................................................................... 11
2.2.1. Gold and the USD to VND exchange rates ........................................................... 11
2.2.2. Gold and inflation .................................................................................................. 12
2.2.3. Gold and interest rates ........................................................................................... 14
2.2.4. Gold and crude oil prices ....................................................................................... 14
2.3. Conclusions................................................................................................................... 15
CHAPTER 3 ............................................................................................................................... 16
Page 1
 The factors affecting the gold market in Vietnam
March 29, 2014
STUDY METHODS .................................................................................................................. 16
3.1. Data description ............................................................................................................... 16
3.2. Expected value analysis ................................................................................................... 17
3.2.1. Gold and the USD to VND exchange rates ............................................................... 17
3.2.2. Gold and inflation ...................................................................................................... 18
3.2.3. Gold and interest rates ............................................................................................... 18
3.2.4. Gold and crude oil prices ........................................................................................... 18
3.3. Analytical framework ...................................................................................................... 19
3.3.1. Terms and definitions ............................................................................................ 19
3.3.2. Spurious regression ................................................................................................ 21
3.3.3. Cointegration.......................................................................................................... 22
3.4. Testing methods ............................................................................................................ 22
3.4.1. Unit root test .......................................................................................................... 22
3.4.2. Cointegration test ................................................................................................... 23
3.5. The theoretical model ................................................................................................... 24
3.5.1. The Vertor Error Correction Model (VECM)........................................................ 24
3.5.2. Explanations:.......................................................................................................... 25
3.5.3. Implementing the model ........................................................................................ 26
CHAPTER 4 ............................................................................................................................... 27
FINDINGS ................................................................................................................................. 27
4.1. The statistical description of the data .............................................................................. 27
4.2. Correlation matrices ......................................................................................................... 28
4.3. Graph-based data analysis ............................................................................................... 29
4.3.1. Gold and the USD to VND exchange rates ............................................................... 29
4.3.2. Gold and inflation ...................................................................................................... 30
4.3.3. Gold and interest rates ............................................................................................... 31
4.3.4. Gold and crude oil prices ........................................................................................... 32
4.4. Summary .......................................................................................................................... 33
4.5. The results from the VECM model.................................................................................. 33
4.5.1. Stationarity test .......................................................................................................... 33
4.5.2. Cointegration test using the Johansen methodology and the optimal time lag of the
model ................................................................................................................................... 34
Page 2
 The factors affecting the gold market in Vietnam
March 29, 2014
4.5.3. The process of estimating the VECM ....................................................................... 36
4.6. The impulse response function and the variance decomposition .................................... 44
4.6.1. The impulse response function .................................................................................. 44
4.6.2. The variance decomposition ...................................................................................... 46
4.7. Residual’s stationarity test ............................................................................................... 47
4.8. Testing for the stability of the regression model using a unit circle ................................ 48
CHAPTER 5 ............................................................................................................................... 49
CONCLUSIONS ........................................................................................................................ 49
5.1. Summarizing the experimental evidence found in Vietnam............................................ 49
5.2. Applying the study in reality ........................................................................................... 50
5.3. Limitations and recommendations ................................................................................... 50
APPENDICES ............................................................................................................................ 51
REFERENCES ........................................................................................................................... 60

Page 3
 The factors affecting the gold market in Vietnam
March 29, 2014
LIST OF TABLES

Table 1: Some noticeable previously conducted studies ........................................................ 10

Table 2: Research materials .................................................................................................... 18
Table 3: Expected values ........................................................................................................ 20
Table 4: The statistical description of the data ....................................................................... 28
Table 5: Correlation matrices ................................................................................................. 30
Table 6: Cointegration test results .......................................................................................... 35
Table 7: Criterion to the optimal time lag .............................................................................. 36
Table 8: Regression estimation results and the long-run effects of economic variables on gold
prices ....................................................................................................................................... 37
Table 9: The list of long-run variable coefficients ................................................................. 38
Table 10: The list of short-run regression coefficients .......................................................... 40
Table 11: The effects of the USD to VND exchange rates on gold prices ............................. 41
Table 12: The effects of inflation on gold prices .................................................................... 42
Table 13: The effects of interest rates on gold prices ............................................................. 43
Table 14: The effects of crude oil prices on gold prices ........................................................ 44
Table 15: The variance decomposition ................................................................................... 47
Table 16: Residual’s stationarity test results .......................................................................... 48
LIST OF CHARTS

Chart 1: The correlation between gold prices and the USD to VND exchange rates ....... 31
Chart 2: The movements in gold prices and inflation ....................................................... 32
Chart 3: The movements in gold prices and interest rates ................................................ 33
Chart 4: The movements in gold prices and crude oil prices ............................................ 33
LIST OF PHOTOGRAPH

Graph 1: The impulse response function........................................................................... 46

Graph 2: Unit circle test results ......................................................................................... 49

LIST OF APPENDIXED FIGURES

Figure 1: Variables’ stationarity test results ...................................................................... 52

Figure 2: Estimating the short-run correlations between gold prices and exchange rates,
inflation, interest rates, and crude oil prices on the basis of the VECM regression results. 55
Figure 3: Complete variance decomposition ..................................................................... 59
Page 4
 The factors affecting the gold market in Vietnam
March 29, 2014

ABSTRACT
“Until the middle of the 1990s, gold was widely used to pay in transactions in Vietnam
as an alternative to currencies, especially in real estate transactions. It had even been used as a
mean of payment in transactions of home furniture or luxury garments before the time. There
have been no records of how many successful transactions of this type were made or how
much those were worth. In fact, many people in Vietnam still keep their habit of hoarding gold
these days because they consider it as a method of saving, although no interest is included, so
the banks have started to offer their clients gold saving services which include certain interest
rates”.1

The banks’ interest rate adjustments for gold deposits have lead to considerable changes
in gold prices as compared to those prices before 1975. Meanwhile, the existence of the gold
exchange since 2006, its noticeable activities during the 2007 – 2008 period and the financial
crisis in 2008 and 2010 have lead to the constant devaluation of the domestic currency, causing
many people to rush to buy gold as a financial safe haven. Consequently, gold prices in
Vietnam were rising at much higher paces than those in the world. For this reason, researchers
from many countries, including Vietnam, have taken on their quest to discover the factors that
influence gold prices in Vietnam as well as the causes of their movements.

With a great desire to follow their path, the researcher has decided to conduct a study
titled “The factors affecting the gold market in Vietnam”. After several experimental tests,
the researcher managed to identify the relationships between gold prices and the United States
Dollar (USD) to Vietnamese Dong (VND) exchange rates, inflation, interest rates and crude oil
prices. In addition, the researcher has also specified the positive and negative correlations
among the mentioned factors and quantified the volatility of gold prices in accordance to the
changes in these factors. The findings of this study have been found to confirm the previously
presented experimental test results and to be valid for the gold market in Vietnam.

1
http://archive.saga.vn/Cohoigiaothuong/Thitruong1/tthh/14941.saga
Page 5
 The factors affecting the gold market in Vietnam
March 29, 2014
CHAPTER 1

INTRODUCTION
Through this chapter, the researcher would present a brief introduction to the thesis, which
includes the rationale of the thesis, the aims and tasks of the thesis, the methods and
methodology of the thesis, and the structure of the thesis.
1.1. The rationale of the thesis
Vietnam, along with many other economies in the world, is facing many obstacles that hamper
its development, including the raging global recession, the devaluation of the US dollars
(USD), which have always been considered as a dominating currency and stored in the foreign
currency reserves of most countries, the prolonged inflation in Vietnam, the decline of its stock
market, and the massive movements of crude oil prices. These factors, together with the
conventional habit of hoarding gold in Vietnam, have rendered gold as the safest and the most
attractive investment option.

However, the current policies on gold market management have been found to be ineffective,
contrary to the natural tendency in a market economy, causing an imbalance between supply
and demand, and favourable for gold speculation, gold trafficking and illegal activities in gold
trading. As a result, it has become exceptionally difficult for the government to control the
domestic gold market, as an entity of the entire financial market, in order to curb inflation, to
maintain macroeconomic stability. Similarly, investors also find it burdensome to utilize gold
as a safe investment option.

Because of the mentioned reasons, the researcher has decided to conduct a study titled “The
factors affecting the gold market in Vietnam”. The study was based on the researcher’s
knowledge in market economy and in applied econometrics, and serves the purposes of
indentifying the relationships of the factors affecting gold prices in Vietnam and of providing
more detailed information about gold to investors as well as managers.
1.2. The aim and scope of the thesis
Through this thesis, the researcher aims to investigate the relationships between gold prices
and the USD to VND exchange rates, crude oil prices, inflation and interest rates in Vietnam,
in order to find the answers for the following questions:

 What are the factors affecting gold prices in Vietnam?

Page 6
 The factors affecting the gold market in Vietnam
March 29, 2014
 How should the influence of these factors on gold prices be assessed?

Relevant data collected from 2008 until 2014 was used to complete the thesis.
1.3. The study methods of the thesis
1.3.1. The pattern

The researcher applied the Vector Error Correction Model (VECM) and cointegration theory to
find the answers for the presented questions. The VECM was chosen for it would be a useful
method to examine financial time series and especially to observe long-run relationships
among economic variables.

The details of this model would be included in Chapter 3.

1.3.2. The data

The data in use was collected weekly from June 1st, 2008 until February 15th, 2014, and
includes the prices of SJC ten-tael gold pieces in the Vietnamese market, the prices of DIESEL
0.5 oil, the USD to VND exchange rates, the inflation rates and the interest rates in Vietnam.

1.3.3. The testing methods

The researcher applied unit root tests and cointegration tests for this thesis.
1.4. The structure of the thesis
The thesis consists of five following chapters:

Chapter 1: Introduction.

Chapter 2: Literature reviews

Chapter 3: Study methods

Chapter 4: Findings

Chapter 5: Conclusions

Page 7
 The factors affecting the gold market in Vietnam
March 29, 2014
CHAPTER 2

LITERATURE REVIEWS
Many studies were conducted prior to this thesis, the aim of which was to attempt to explain
gold’s price behavior and the elements influencing it. Yet, although this thesis contains several
distinct features, it still had to follow Bernard Dierinck’s basic principle: in order to study the
prices of gold, researchers need to employ macro-oriented, speculative and inflation-
hampering approaches.
2.1. Study approaches
2.1.1. The macro-oriented approach

The macro-oriented approach uses the primary macro variables to explain the movements in
gold prices. These variables include exchange rates, the global income and political shocks,
among others.

A study done by Dooley, Isard and Taylor (1995) was an example for this approach.
They performed a variety of experimental tests and successfully verified that gold prices could
explain exchange rate movements, in addition to monetary policies and other variables in
standard exchange rate models. The core concept introduced by Dooley, Isard and Taylor in
this study is to view gold as an asset without a country. If the attractiveness of holding net
claims on A is reduced by a shock of any type, the demands for other assets, including gold
and net claims on B, would be increased, causing changes in market-clearing prices.

Sjaastad and Scacciavillani (1996) conducted another noticeable relevant study which
involves their research on the relationship between gold prices and the foreign exchange
market in the 1982 – 1990 period. Their findings suggest that while the rise or fall in value of
the European Euro (EUR) caused considerable impacts on the price of gold, the USD had little
influence over it. According to Sjaastad and Scacciavillani, “movements in the real exchange
rates among the major currencies account for nearly half of the observed variance in the spot
price of gold during that period”.

Other researchers have also presented a number of studies on the relationship between gold
and exchange rates, the common aim of whom is to determine whether or not gold is an
effective tool to manage foreign exchange risks.

Page 8
 The factors affecting the gold market in Vietnam
March 29, 2014
Table 1: Some noticeable previously conducted studies

Order Researchers Researched currencies Implemented Did gold act as an

models inflation hedge?

1 Capie, British pound, US dollar and EGARCH Yes

Mills, and Japanese yen.
Wood
(2005)

2 Wang and Japanese yen. Tail value at risk Yes

Lee (2011) (TVAR)

3 Mark Joy Euro, Japanese yen, Indian rupee, Dynamic Yes, for all
(2011) Taiwan dollar, Australian dollar, Conditional currencies.
Canadian dollar, Danish krone, Correlation
Israeli shekel, Maltese lira, New
Zealand dollar, Norwegian krone,
Singapore dollar, South African
rand, Swedish krona, Swiss franc,
British pound.

4 Ciner, US dollar and British pound. Dynamic Yes.

Lucey and Conditional
Gurdgiev Correlation and
(2013) GARCH

2.1.2. The speculative approach

From this approach, gold is considered as a channel for speculation, and the researchers
employing this approach focus on studying the rationality and the irrationality of the
movements in gold prices.

Diba and Grossman (1984) conducted a study on the possibility of “rational bubbles” in
the relative price of gold and presented three discoveries. First, the relative price of gold,
which is defined as the overall price index of goods and services in gold, is related to market
fundamentals, the nominal interest rate for instance. Second, the process generating differences
Page 9
 The factors affecting the gold market in Vietnam
March 29, 2014
of market fundamentals is stationary. Finally, actual price movements do not involve rational
bubbles.

Pindyck (1993), on the other hand, attempted to develop a present gold price assessment
model for future contracts. The model has been successfully employed in the transactions of
copper, wood and oil. In the case of gold, however, the model has not proven to be effective,
for the convenience yield of gold is different from that of other commodities.

Studies that solely focus on the relationship between gold prices and crude oil prices are rare,
since the price of oil is simply one single variable among many when researchers study the
factors affecting gold prices.

2.1.3. The inflation-hampering approach

From this approach, researchers focus on evincing that gold investments can hedge against
inflation. This means investing in gold will offer investors an insurance for their investments.
Due to this nature, studies of this type often involve long-run and short-run movements of
gold.

Kolluri (1981), who is a noticeable example of those researchers who employ the
inflation-hampering approach, has done some empirical research to identify the importance of
gold in inflation prevention. Kolluri applied two other approaches in order to accomplish this
task. First, he modelized the relations of profits generated by gold investments and those
generated by expected inflation or its variations using the Cochrane-Orcutt iterative process.
The data used by Kolluri for this approach was the prices of gold from 1968 until 1980.
Second, he modelized and observed the profits of shares and bond from 1926 until 1978 before
applying this model for gold. Finally, he concluded that gold was an effective inflation hedge
in the 1968 – 1980 period.

Another researcher, Sherman (1983), applied multiple regression analysis to determine

which factors influence gold prices. With the given R2 = 76.03%, he concluded that tension
index, interest rates in the United States (US) foreign exchange market, gross domestic
products (GDP), excess liquidity and expected inflation are important factors that affect gold
prices. Yet, when Sherman tried to make some adjustments to eliminate serial correlation,
some variables, such as tension index, appeared to lose their statistical significance while some
others, such as expected inflation, became more statistical significant.
Page 10
 The factors affecting the gold market in Vietnam
March 29, 2014
Mahdavi and Zhou (1997), on the other hand, did their research to determine whether or
not gold or any other commodity is an inflation index. They discovered that inflation can be
predicted through error-correction models and that commodities can be used as an effective
inflation index because this index is cointegrated with the commodity price index. From their
point of view, if this inflation index produces unreliable predictions of inflation rates, it is only
because of massive, short-run movements in gold prices.

The study conducted by Ghosh, Levin, MacMillan and Wright (2004), which focused
on the long-run relationship between gold prices and the average prices of commodities in the
United States of America (USA), is also an exceptional work. Through this study, they have
presented an explanation for the short-run differences of the mentioned long-run relationship
using cointegrated regression models, with the data collected from 1976 until 1999.

After the study of Ghosh, Levin, MacMillan and Wright (2004), Levin and Wright
(2006) conducted a study that verified the possibility of the long-run relationship between gold
prices and the average prices of commodities in the USA in the 1975 – 2006 period. They
concluded in their study that in the period, if the average price of commodities in the USA
rised by 1%, the price of gold would also rise by 1%

Lampinen (2007) made an attempt to re-examine the findings previously proposed by

Levin and Wright (2006), which took him an additional year’s time for extended research.
Lampinen’s findings were similar to those introduced by Levin and Wright, except he used 19
dummy variables while Levin and Wright only used 10. The role of these dummy variables is
to help researchers avoid autoregression in the model residuals.
2.2. The available experimental evidence
2.2.1. Gold and the USD to VND exchange rates

Many studies on the relationship between gold prices and exchange rates have been
introduced. Although different studies involves different types of variables, most of them come
to a conclusion that gold prices react against the movements in the value of the US dollars and
the profitability ratio in securities.

Koutsoyiannis (1983) learnt that the price of gold is influenced by the economy of the
USA rather than by that of the world. He confirmed that the US dollar is a global liquidity and

Page 11
 The factors affecting the gold market in Vietnam
March 29, 2014
that both gold prices and crude oil prices are measured in this currency. Hence, the inverse
relation between gold prices and the dollars was discovered.

Smith (2001) did some research on the relationship between gold prices and exchange
rates using the daily-updated and monthly-updated data from 1991 until 2001, and eventually
verified the short-run relationship between them.

Vural (2003) made testing attempts on gold price sensitivity using a variety of variables,
including the USD to EUR exchange rates, the Dow Jones index, crude oil prices, interest
rates, silver prices and copper prices, along with the multiple regression model and the data
collected monthly from 1990 until 2003. His test results confirm a positive correlation between
gold prices and the prices of silver, copper and crude oil, and a negative correlation between
gold prices and the USD to EUR exchange rates, interest rates and the Dow Jones index.

Tully and Lucey (2007) applied the Asymmetric Power Generalized Auto Regressive
Conditional Heteroscedasticity (APGARCH) to examine the effects of certain macro variables
on gold prices and ultimately found an inverse relationship between gold prices and the USD
in the 1984 – 2003 period.

2.2.2. Gold and inflation

One of the factors causing the devaluation of the Vietnamese Dong (VND) and of assets in
general is inflation. Movements in gold prices can quickly lead to movements in inflation due
to value preservation. However, certain conditions, such as monopolistic competition,
transaction costs or unique characteristics of a nation, can render gold ineffective against
inflation. When exchange rates have a closer correlation with gold prices, the resulted
exchange rate shocks seem to have major impacts on domestic prices and asset value as
measured in the domestic currency. Thus, investors often choose to buy large amounts of gold
in order to avoid exchange rate shocks. Moreover, price adjustments may prove to be
unreasonable, because exchange rate shocks hardly change the price of gold. These shocks are
an outcome of either a perfect competitive market or transaction costs created by the
management agencies’ monetary policies (Wang and Lee, 2011).

Page 12
 The factors affecting the gold market in Vietnam
March 29, 2014
2.2.2.1. Gold as an inflation hedge
Levin and Wright (2006) applied the VECM and cointegration of time series from 1976 until
2005 and found an inverse relationship between the movements in gold prices and the
movements in exchange rates in the USA. They also confirmed that gold did act as an inflation
hedge in the long run in a number of countries, including Saudi Arabia, China, India, Indonesia
and Turkey.

Sjaastad (2008) concluded in his study that the superiority of the European monetary
market over the gold market in the 1990s was later replaced by the dominance of the US
dollars. In addition, such gold providers as Australia, South Africa and Russia appeared to
have little to no influence on the price of this commodity. Furthermore, while the previously
conducted studies aimed to contradict gold’s role as an inflation hedge, Sjaastad’s study
proved that hedging inflation is a purpose of hoarding gold.

Öztürk and Açıkalın (2008) successfully found the long-run relationship between gold
prices and the consumer price index (CPI) using the Granger cointegration test and the data
collected monthly from 1995 until 2006.

Topçu (2010) examined the relationships between gold prices and the Dow Jones
indices, the inflation rates in the USA and the M3 money supply during the 1995 – 2009 period
by employing multiple regression analysis. His findings suggest that the profitability ratios of
the Dow Jones index and the US dollars have direct relationships with gold prices. These
relationships, however, will no longer be statistically significant if a negative correlation
between gold prices and crude oil prices as well as inflation is found or if no positive
correlation between them is found.

2.2.2.2. Gold not as an inflation hedge

Many experimental evidences demonstrate that gold has not been an effective hedge against
inflation in numerous countries. The followings are some noticeable studies that introduce
such evidences:

Sherman (1983) conducted a study based on the data collected yearly from 1970 until
1980. Through his study, Sherman verified the positive correlation between gold prices and
expected inflation. This correlation has been proven to be statistically significant.

Page 13
 The factors affecting the gold market in Vietnam
March 29, 2014
Jaffe (1989) did similar research on the basis of the data collected from 1970 until 1980
and found no evidence of using movements in gold prices to predict CPI movements.

Larson and McQueen (1995) found a positive correlation between gold prices and
inflation, but this correlation was not statistically significant.

Mahadavi and Zhou (1997) and Lawrence (2003) also concluded in their studies that
gold prices do not react to CPI movements.

Tkacz (2007), on the other hand, confirmed that movements in gold prices did signal
changes in inflation in 14 countries. The characteristics and institutions of each of these
countries determined whether changes in inflation were positive or negative.

2.2.3. Gold and interest rates

Few studies on the relationship between gold prices and interest rates have been conducted and
the latter appears to have a one-way relationship with the former. In other words, gold prices
are likely to rise when interest rates fall, thus investments that include gold become highly
evaluated. People have a propensity to invest in other commodities, those such as gold, that
every time the nominal interest rates of short-term financial assets are low and contain risks.
They normally hold onto their gold even when holding fees are required. The low nominal
interest rates are a result of increases in gold demand and gold prices.

The studies conducted by Loutsoyiannis (1983) and Fortune (1987) also emphasized the
influence of interest rates on gold prices, whereas Cai et al (2001) concluded in his study that
movements in gold prices are caused by interest rate movements.

Vural (2003) made testing attempts on gold price sensitivity using a variety of variables,
including the USD to EUR exchange rates, the Dow Jones index, crude oil prices, interest
rates, silver prices and copper prices, along with the multiple regression model and the data
collected monthly from 1990 until 2003. His test results confirm a positive correlation between
gold prices and the prices of silver, copper and crude oil, and a negative correlation between
gold prices and the USD to EUR exchange rates, interest rates and the Dow Jones index.

2.2.4. Gold and crude oil prices

Oil prices were relatively stable worldwide until 1973. They constantly fluctuated after 1973
and the influence of oil price shocks on the global economy was becoming increasingly high.

Page 14
 The factors affecting the gold market in Vietnam
March 29, 2014
A number of studies have been conducted to investigate the relationship between crude oil
prices and macroeconomic factors.

One of the first and most noticeable studies were those conducted by Hamilton (1983),
Burbidge and Harrison (1984), Gisser and Goodwin (1986), Loungani (1986), and Mork
(1989), which focused on the connections between crude oil prices and macroeconomic
factors. Recent studies are often based on the time series of nations, such as Guo (2005) or
Breitenfellner and Crespo (2008), or on cross-sectional data, such as Cunado and Gracia (2003,
2005) or Jimenez and Marcelo (2005).

Among the most recent studies on this subject, there are Basher and Sadorsky (2006),
Park and Ratti (2008), Kilian and Park (2009), and Narayan (2010). Park and Ratti (2008) used
the vector autoregression (VAR) model to study the correlation between oil price shocks and
profitability ratios in security in the USA and 13 European countries.

A study titled “Oil and gold: correlation or causation?”, which was conducted by Le
Thai Ha and Youngho Chang based on the data collected monthly from January 1986 until
April 2011, aims to examine the relationship between the prices of gold and oil, the two
strategic commodities of many countries. This relationship was examined through inflation and
through gold and oil’s interactions with the dollar index. After using several different
representative variables of oil prices, Le Thai Ha and Youngho Chang realized the
asymmetricity and nonlinearity of crude oil prices’ influence on gold prices. Their study
findings also verify a long-run relationship between gold prices and crude oil prices.

Another noticeable study is the one conducted by Cengiz Toraman, Çağatay Başarır and
Mehmet Fatih Bayramoğlu, titled “Determination of Factors Affecting the Price of Gold: A
Study of MGARCH Model”. The primary aim of this study is to determine the factors
affecting gold prices using relevant data of crude oil prices, the USD exchange rates, inflation
and effective interest rates in the USA, which was collected monthly from June 1992 until
March 2010. Their findings suggest that gold prices have a close negative correlation with the
USD exchange rates and a positive correlation between with crude oil prices.
2.3. Conclusions
Overall, in most studies, gold has been considered as a tool to manage exchange rate risks or as
an inflation hedge. On this basis, the researcher would investigate the relationships between
gold prices in Vietnam and the USD to VND exchange rates, inflation, interest rates and crude
oil prices.
Page 15
 The factors affecting the gold market in Vietnam
March 29, 2014
CHAPTER 3

STUDY METHODS
This chapter consists of three sections, namely the data description, the expected value analysis
and the analytical framework of the VECM. The researcher would present the theoretical
models that assisted her analysis, the testing methods used in those models and the solutions
for non-stationary time series in this chapter.
2.4. 3.1. Data description
The data used for this thesis was collected in Vietnam and includes the prices of ten-tael SJC
gold, the prices of crude oil in the world, the USD to VND exchange rates, inflation and basic
rates of interest. Since the data of gold prices and exchange rates was collected daily while the
data of the other variables was collected monthly, the researcher decided to convert all data
into monthly data and only use the data collected from June 1st, 2008 until February 15th, 2014
for the thesis. Afterward, the researcher employed frequency conversions to convert this data
into weekly data. In order to avoid market noise, the researcher used the data collected in the
first day of each month rather than the average monthly data, as Forrest Capie, Terence C.Mills
and Geofrey Wood (2004) did in their study. Additionally, the missing observatory data meant
to be conducted during off days was replaced by the data collected through the nearest
observations.

Why did the researcher choose to analyze weekly data?

According to Smith (2001) and Forrest Capie, Terence C.Mills and Geofrey Wood (2004),
weekly data is most suitable for this type of research for the following reasons:

- Exchange rates and gold prices are highly fluctuative and unexpected elements may
cause sudden rises of gold prices. These rises are temporary and unrelated to any other
variable in the model, and gold prices will return to normal afterwards. This, as a result,
could prevent researchers from discovering the true relationships between gold prices
and the other variables in the model.

- Monthly data, on the other hand, may not indicate any significant movements of the
factors affecting gold prices because they only have short-run effects on gold.

Page 16
 The factors affecting the gold market in Vietnam
March 29, 2014
In order to eliminate the abnormal, sudden movements often considered as outliers, the
researcher took the natural logarithms (ln) of all variables in the model. This is a very popular
method and has been applied by numerous researchers, including Forrest Capie, Terence
C.Mills and Geofrey Wood (2004) and Juan C. Reboredo (2013), among others.

Table 2: Research materials

For gold prices http://vangvietnam.vn/Articles/Default.aspx?CateID=192

For crude oil prices http://www.quyettoan.vn/dutoan/Thong-cao-bao-chi-gia-xang-dau

For inflation rates http://www.gdpinflation.com/2013/08/inflation-rate-in-vietnam-

from-2000-to.html

For the USD to VND exchange http://www.fxstreet.com.vn/rates/exchange-rates/

rates

For interest rates http://www.sbv.gov.vn/portal/faces/vi/vim/vipages_cstt/laisuat/bang

laisuat

2.5. 3.2. Expected value analysis

Based on the experimental evidence introduced in chapter 2, the researcher would present
some analyses and a table indicating the expected values of all variables in the model before
implementing tests.

3.2.1. Gold and the USD to VND exchange rates

According to Sjaastad and Scacciavillani (1996), movements in the EUR have great effects on
gold prices. Capie, Mills and Wood (2005), Wang and Lee (2011), Mark Joy (2011), and
Ciner, Lucey and Gurdgiev (2013) made the same conclusion stating that the price of gold
decreases when the price of the USD increases.

However, these studies were conducted in countries that quotes foreign exchange rates
indirectly, those such as the USA or England, whereas in Vietnam, these rates are directly
quoted (exchange rates are quoted in foreign currencies and are priced in the VND). Hence, the
researcher expected a positive correlation between gold prices and the USD to VND exchange
rates, in other words, when the price of the USD rises while that of the VND falls, investors

Page 17
 The factors affecting the gold market in Vietnam
March 29, 2014
are more likely to buy gold, causing its demands to exceed its supplies, which consequently
leads to rises in gold prices.

3.2.2. Gold and inflation

One of the factors causing the devaluation of the Vietnamese Dong (VND) and of assets in
general is inflation. Movements in gold prices can quickly lead to movements in inflation due
to value preservation. Many researchers, including Kolluri (1981), Sherman (1983), Mahdavi
and Zhou (1997), Ghosh, Levin, Macmillan and Wright (2004), Sjaastad (2008), and Öztürk
and Açıkalın (2008), agreed that gold acts as an inflation hedge. Some other researchers, those
such as Sherman(1983), Jaffe(1989), and Cecchetti et al (2000), disagreed with it. Researchers
such as Tkacz (2007) and Christe and David et al (2000) did not make clear conclusions.
Therefore, the researcher of this thesis expected a correlation, either positive or negative,
between gold prices and inflation in Vietnam.

3.2.3. Gold and interest rates

The influence of interest rates on gold prices have been concluded in the studies conducted by
Loutsoyiannis (1983), Fortune (1987), Cai et al (2001) and Vural (2003). They found that there
is a causality between gold prices and interest rates, and that interest rates have a direct
relationship with gold prices. In other words, when nominal interest rate falls, investors will
seek for more profitable investment options, those such as gold, causing gold prices to rise.
Based on this theory, the researcher of this thesis expected a negative correlation between gold
prices and interest rates.

3.2.4. Gold and crude oil prices

Researchers such as Sadorsky (2006), Park and Ratti (2008), Kilian and Park (2009), Narayan
(2010) and Cengiz Toraman, Çağatay Başarır and Mehmet Fatih Bayramoğlu conducted
studies on the gold market of many countries and found a positive correlation between gold
prices and crude oil prices. The same correlation was found by Le Thai Ha and Youngho
Chang in their study, “Oil and gold: correlation or causation?”. As a result, the researcher of
this thesis also expected a positive correlation between gold prices and crude oil prices in
Vietnam.

Page 18
 The factors affecting the gold market in Vietnam
March 29, 2014
Table 3: Expected values

Variable Unit Notation Notation after Expected

before calculating value (+/-)
calculating differences
differences

Gold prices VND per 10 lngold dlngold +/-

taels

The USD to VND VND lnusd-vnd dlnusd_vnd +

exchange rates

Inflation % per week lninflation dlninflation +/-

Interest rates % per week lninterestrate dlninterestrate -

Crude oil prices VND lnoil dlnoil +

3.3. Analytical framework
3.3.1. Terms and definitions

a. The stationarity of time series2

Stationary time series

A stationary time series is one that:

 Has a tendency to return to its average state and its variables fluctuate around a certain
average value over the long run.

 Has statistical properties remain constant over time.

 Has an autocorrelation schema where coefficients decrease as time lag increases.

In statistical language, the characteristics of a stationary time series Yt can be describe as

follow:

 E(Yt) is a constant in all time periods t:

E(Yt) = (9)

 Var(Yt) is a constant in all time periods t:

2
“A guidebook on Eviews”, Bui Duong Hai, 2011.
Page 19
 The factors affecting the gold market in Vietnam
March 29, 2014
Var(Yt) = E(Yt-μ)2= (10)

 Cov(Yt , Yt-k) is a constant in all time periods t and k is not equal to 0:

Cov(Yt , Yt-k) = = E[(Yt-μ) (Yt-k-μ)] (11)

symbol represents covariance at time lag k (between Yt and Yt-k)

The characteristics of a stationary time series are very important in a model. According to
Gujarati (2003), if the time series is non-stationary, its behavior can only be investigated in one
specific time period. Because of this, analytical results cannot be used for generalizing a theory
or for forecasting future movements. Another reason is that regression models for non-
stationary time series often lead to Spurious regressions, which cause analytical results to lose
there statistical significance.

Non-stationary time series

Most time series are in fact non-stationary. Some economic variables, especially in finance, are
believed to follow a random walk.

Random walks are classified into two types: (i) random walks with constants and (ii) without
constants.

(i) Random walks with constants

Time series Yt will be considered a random walk if:

(1)

Which can be generalized as:

∑ (2)

Then:

∑ (3)

And:

Var ( =t (4)

Page 20
 The factors affecting the gold market in Vietnam
March 29, 2014
In the end, although the average value of Y ramains constant, when t increases, its variance
also rises. This violates the conditions of a stationary time series.

(ii) Random walks without constants

Equation (1) can be rewritten as:

∑ (5)

Then:

∑ (6)

Var ( =t (7)

In the end, both the average value and the variance of Y change over time. This violates the
assumptions of a stationary time series.

b. Solutions for non-stationary time series

Calculating the difference of a non-stationary time series, especially its first differences, is a
popular solution. According to Gujarati (2003), the first differences of most non-stationary
time series are stationary time series. This difference can be presented as follow:

Equations (1) and (5) can be rewritten as:

∑ (8)

3.3.2. Spurious regression

a. Definition
A Spurious regression can be defined as a phenomenon that appears when a regression
of two irrelevant and non-stationary time series with a high R2 indicates the agreement or
conflict between them. Although the value of R2 may be high, the results of this regression
model may not be economically significant. Since the ordinary least squares (OLS) may not be
constant over time, statistical tests implemented based on the model are probably incorrect and
insignificant.

Granger and Newbold (1974) applied a A Monte Carlo simulation to generate a large number
of Yt and Xt time series, the roots of which can be presented as follow:
Page 21
 The factors affecting the gold market in Vietnam
March 29, 2014
(9)

(10)

with eyt and ext generated randomly.

Yt and Xt are independent time series, therefore their regressions always produce incorrect
results. However, after Granger and Newbold tested a regression model which had a massive
number of different values of Yt, they rejected the assumption that Ho is a coefficient of Xt
when Xt is equal to 0 in 75% of the cases. They also found that the values of R2 in their
regression equations were very high while the values of Durbin- Watson2 were low.

b. Explanations for Spurious regressions

According to Asteriou, Spurious regressions are a result of the fact that X and Y are two non-
stationary time series, hence their linear combination (the residual of a regression equation)
may also be non-stationary. If it is really non-stationary, it violates the basic assumptions of
the OLS. In this case, residuals such as this are supposed to be extending over time, yet the
values of R2 in the OLS are still high because the OLS tends to select parameters that would
lead to the minimum residual sum of squares.

3.3.3. Cointegration

Cointegration was first mentioned by Granger before other researchers conducted

thorough studies on it. Economic time series usually make it difficult for researchers to
conduct studies in empirical econometrics due to Spurious regressions. Calculating the
difference of a time series until it becomes stationary is a posible solution for this. However,
this is yet to be an optimal option because it will eliminate the valuable long-run statistics of
the time series.

A linear combination of two distinct non-stationary time series is frequently expected to

be also non-stationary. Yet, in certain cases, simulstaneous movements of these time series are
expected and their linear combination becomes stationary. When this happens, these two time
series are considered to be cointegrated.
3.4. Testing methods
3.4.1. Unit root test

A unit root test is widely used for testing the stationarity of a time series.
Page 22
 The factors affecting the gold market in Vietnam
March 29, 2014
Suppose we have an autoregression equation:

với (11)

And assumptions:

According to Dickey and Fuller (1976), predicted value t of coefficient depends on tau

statistics τ (τ = ). This testing method is also called the

Dickey-Fuller (DF) test.

To implement a test with H0, tau statistics τ would be compared with the DF test results. If the
absolute value of τ is greater than the expected significance α, H0 would be rejected, otherwise
H0 would be accepted. However, comparisons such as this often depend on p-values. H0 is
accepted if p-values ≥ α and rejected if p-values < α.

3.4.2. Cointegration test

Testing for cointegration using the Johansen methodology

The aim of the test: To determine the number of linear combinations of some stationary
variables and some non-stationary ones. This will indicate the total number of balanced long-
run relationships.

In this cointegration equation, the average value of variable Yit may not equal 0. Therefore, the
cointegration equation may contain intercept coefficients and trends.

There are two test statistics in the Johansen testing:

a. Trace Statistics
Assumptions:

H0: There are at most r cointegrated relationships (r = 0, 1, 2, …, k-1)

H1: There are m cointegrated relationships

The result can be presented as follow:

Page 23
 The factors affecting the gold market in Vietnam
March 29, 2014

represents different values in descending order.

b. Maximum Eigenvalue
Assumptions:

H0: There are at most r cointegrated relationships (r = 0, 1, 2, …, k-1)

H1: There are m cointegrated relationships

The result can be presented as follow:

LRmax ( /r +1) = - n. log(1− ) = LRtr ( +1⁄k)−LRtr( ⁄k)

3.5. The theoretical model
The aim of the thesis is to examine the relationships between gold prices with the USD to
VND exchange rates, inflation, interest rates and crude oil prices in Vietnam. The researcher
employed the VECM and cointegration theory on the basis of the data collected from June
2008 until February 2014.

 Why did the researcher choose the VECM?

While the OLS and VAR models could only help researchers examine short-run relationships
among economic variables, the VECM could be an effective tool for the long-run relationships
among them. In addition, the VECM could allow regressions of non-stationary time series,
which could not be done using the OLS model, and help researchers avoid Spurious
regressions.

3.5.1. The Vertor Error Correction Model (VECM)

During an examination of a non-stationary time series of economic variables, researchers need

to be cautious when they verify the relationships among them. These relationships depend
mostly on the stationarity of each variable. If two different non-stationary variables are two
fifth degree integrated time series, a balanced long-run relationship between them is possible.
This phenomenon is called cointegration. In terms of econometrics, if two or more variables
are cointegrated, they have the same degree of integration (time series I(p)) and their linear
combination is a stationary time series (I(0)).
Page 24
 The factors affecting the gold market in Vietnam
March 29, 2014
Time series {y1,t} và {y2,t} are cointegrated if: {y1,t} and {y2,t} are both p degree integrated
time series: I(p)

We have a combination (a1,a2) not simultaneously equal to 0 for which the sum of a1*y1,t +
a2*y2,t is a stationary time series: I(0).

The VECM is one of many models applied for observing such cointegrations among time
series. Suppose time series y1,t and y2,t are two first degree integrated time series, their
relationship can be described with a reduced form of the VAR model as follow:

Yt= A0+ A1*Yt-1+ A2*Yt-2+ Et (*)

For which:

[ ] is a variable matrix

[ ] is a coefficient matrix of variables Yt-1 and Yt-2

[ ] is a confusion matrix of an equation

Equation (*) can be rewritten as follow:

ΔYt= A0+ (-I+A1+A2)*Yt-1 - A2*ΔYt-1+ Et (**)

With the assumption that Et is a stationary time series, we can see that the left side of equation
(**) is a stationary time series and each item on its right side is also a stationary time series.
Thus, either (-I+A1A2)*Yt-1 or a linear combination of y1,t and y2,t has to be a stationary time
series. This confirms the cointegration between y1,t and y2,t. When a VAR model is presented
in the form of equation (**), it is called a VECM. As a result, equation (**) can be rewritten
into the following system of equations:

y1,t – y1,t-1= a1,0+ α1*(y1,t-1+β*y2,t-1) + ɛ1,t

y2,t - y2,t-1 = a2,0 +α2*(y1,t-1+β*y2,t-1) + ɛ2,t

3.5.2. Explanations:

Coefficient β represents the long-run relationship between y1,t and y2,t, coefficients α1 and α2
represent the short-run adjustment mechanism.

Page 25
 The factors affecting the gold market in Vietnam
March 29, 2014
3.5.3. Implementing the model

- Step 1: Examine the stationarity of each time series

- Step 2: Test for cointegration using the Johansen methodology

- Step 3: Establish a testing model to determine the short-run and long-run relationships
among time series

o Case 1: If no integration is found, the VAR model will be applied to determine

the short-run and long-run relationships among time series

o Case 2: If an integration is found, the VECM model will be applied to determine

the short-run and long-run relationships among time series

- Step 4: Test for residuals in the VECM

- Step 5: Implement an impulse response analysis and a variance decomposition analysis

- Step 6: Examine the stationarity of residuals in the VECM

Page 26
 The factors affecting the gold market in Vietnam
March 29, 2014
CHAPTER 4

FINDINGS
In this chapter, the researcher would present the results of the test for the stationarity of time
series, of the optimal time lag test and of the cointegration test. In addition, the researcher
would also conduct a process of estimating the VECM and of assessing its appropriateness.
2.6. 4.1. The statistical description of the data
Table 4: The statistical description of the data

lngold lnusd_vnd lninflation lninterestrate lnoil

Average 15.79379 8.395681 -3.813411 -6.394362 8.270542

value

Median 15.86320 8.373382 -3.896525 -6.391367 8.327848

Maximum 16.30373 8.708104 -2.649582 -5.818860 8.769854

value

Minimum 14.87672 7.831816 -5.680792 -6.791635 7.496600

value

Standard 0.351231 0.148378 0.640242 0.194571 0.285241

deviation

Deviation -0.513740 -0.485257 -0.163103 0.526700 -0.507768

Kurtosis 2.180442 2.774986 2.688589 3.966463 2.2932276

Jarque - Bera 21.30450 12.24117 2.508434 25.20566 18.87953

Probability 0.000024 0.002197 0.285299 0.000003 0.00079

Total 4674.961 2485.122 -1128.770 -1892.731 2448.080

Total 36.39208 6.494746 120.9234 11.16808 24.00199

deviation

Observations 296 296 296 296 296

Page 27
 The factors affecting the gold market in Vietnam
March 29, 2014

The data used for this thesis includes 5 time series, each of which involves 296
observations. In addtition, the data was collected weekly from June 1, 2008 until February 15,
2014. With 296 observations, all data used for this thesis would be at 10% significance level.
Table 4 indicates the statistical description of the variables used in this thesis.

The average value and the median of variable lngold are 15.79% and 15.86%,
respectively. Gold prices in Vietnam were generally more fluctuative than those in other
countries during the making of many studies, such as the ones conducted by Forrest Capie,
Terence C.Mills and Geofrey Wood (2004) in the USA (6.25%) or by Capie, Mills, and Wood
(2005) in England (5.968%). With the the Jarque–Bera test = 21.34 and is greater than all
critical values, it fits all significance levels. Additionally, the deviation and the kurtosis are -
0.514 and -0.82, respectively3. Therefore, prices of gold are not normally distributed.

The average value and the median of variable lngold are 15.79% and 15.86%,
respectively. Gold prices in Vietnam were generally more fluctuative than those in other
countries during the making of many studies, such as the ones conducted by Forrest Capie,
Terence C.Mills and Geofrey Wood (2004) in the USA (6.25%) or by Capie, Mills, and Wood
(2005) in England (5.968%). With the the Jarque–Bera test = 21.34 and is greater than all
critical values, it fits all significance levels3. Additionally, the deviation and the kurtosis are -
0.514 and -0.824, respectively. Therefore, gold prices are not normally distributed.

This was also the conclusion in the research done by Ghosh, Levin, Macmillan and
Wright (2004) and Vural (2003), among others. Similarly, variables such as lnusd_vnd,
lninterestrate and lnoil are also not normally distributed, regardless of their significance levels.
Only lninflation is normally distributed at 10% significance level.
2.7. 4.2. Correlation matrices
Table 5 demonstates the matrices of the correlations among variables in the VECM, most of
which are generally at 10% significance level. It can be clearly seen that the correlations
between lngold and lnoil as well as lnusd_vnd match the expectations of the researcher.
3
The critical values at 10%, 5% and 1% significance levels are 4.61, 5.99 and 9.21, respectively. With the assumption that
H0 is normally distributed, H0 can be rejected when Jarque-Bera > critical values. This suggests that the data is not normally
distributed.
4
A variable is considered as normally distributed when its deviation and its excess kurtosis (KE = kurtosis – 3) are equal to
0.
Page 28
 The factors affecting the gold market in Vietnam
March 29, 2014
However, only lnoil and lnusd_vnd are statistically significant. In contrast, lninterestrate does
not match its expected value as presented in table 3, and lninflation has a negative correlation
with lngold.

Table 5: Correlation matrices

Covariance Analysis: Ordinary

Date: 03/28/14 Time: 11:32

Sample (adjusted): 6/01/2008 1/26/2014

Included observations: 296 after adjustments

Balanced sample (listwise missing value deletion)

Correlation

Probability LNOIL LNINTERESTRATE LNINFLATION LNUSD_VND LNGOLD

LNOIL 1.000000

-----

LNINTERESTRATE 0.537234 1.000000

0.0000 -----

LNINFLATION 0.150933 0.617940 1.000000

0.0093 0.0000 -----

LNUSD_VND 0.815023 0.449196 0.075389 1.000000

0.0000 0.0000 0.1959 -----

LNGOLD 0.848555 0.162992 -0.016346 0.820668 1.000000

0.0000 0.0049 0.7794 0.0000 -----

2.8. 4.3. Graph-based data analysis

4.3.1. Gold and the USD to VND exchange rates

Chart 1: The correlation between gold prices and the USD to VND exchange rates

Page 29
 The factors affecting the gold market in Vietnam
March 29, 2014
Sources: The researcher’s synthesization

It can be seen from the chart that gold prices did not have an obvious correlation with the the
USD to VND exchange rates during the 2003 – 2008 period. Due to Vietnam’s floating
exchange rate regime, the USD to VND exchange rates were relatively stable. Moreover, even
before and after this period, people tended to buy the US dollars as much as they tended to buy
gold. According to an article, in Ho Chi Minh city alone, there was a rising tendency of
hoarding gold in 2004 as it grew by 15.6%. The growth of the USD was the second highest in
the year as it reached 13.9%.

Yet, after observing the movements in gold prices and in exchange rates in the 2008 – 2014
period, the researcher saw a positive correlation between them, although this correlation may
be very weak.

4.3.2. Gold and inflation

During the 2008 – 2014 period, the correlation between gold prices and inflation repeatedly
changed from positive to negative and from negative to positive.

In shorter periods during these years, specifically from June 2008 until July 2009 and from
July 2011 until July 2012, there was an obvious negative correlation between gold prices and
inflation. However, this correlation changed to positive during the period between January
2010 and June 2011.

Chart 2: The movements in gold prices and inflation

Page 30
 The factors affecting the gold market in Vietnam
March 29, 2014
Sources: The researcher’s synthesization

4.3.3. Gold and interest rates

Interest rates in Vietnam are under the management of its government and is kept between -1
and 1. Because of this, interest rates during the research time were marginally fluctuative.
However, as seen from chart 3, interest rates in the country during the 2008 – 2014 period
were likely to rise (or fall), when gold prices rised (or fell). However, during the period
between February and July of 2011, these variables had opposite movements. Hence, the
correlation between gold prices and interest rates is positive and matches the reseacher’s
expectations.

Chart 1: The movements in gold prices and interest rates

Sources: The researcher’s synthesization

Page 31
 The factors affecting the gold market in Vietnam
March 29, 2014

4.3.4. Gold and crude oil prices

Chart 2: The movements in gold prices and crude oil prices

Sources: The researcher’s synthesization

During the period between June 2008 and February 2014, the movements in crude oil prices
were in the opposite direction to the movements in gold prices. There were exceptions,
however, such as the period between January and June of 2011.

Page 32
 The factors affecting the gold market in Vietnam
March 29, 2014
2.9. 4.4. Summary
The following conclusions were made after the graph-based data analysis:

 Except for lninflation, all variables in this research are not normally distributed.
 The observatory results of such variables as lnoil and lnusd_vnd match the researcher’s
expectations as well as the previously introduced empirical theories.
 lninflation and lninterestrate still have unclear ralationships with lngold.
 The researcher needed to implement the VECM in order to generate the necessary
quantitative evidence before drawing any further conclusions.
2.10. 4.5. The results from the VECM model
4.5.1. Stationarity test

The researcher decided to implement an augmented Dickey–Fuller test (ADF) in order to

examine the stationarity of the time series used in this thesis. The test results suggest that all
time series have a zero degree nonstationary and yet a first degree stationarity at 10%
significance level. The table below demonstates the ADF test results.

Variable Notation ADF test on time ADF test on first

series degree differences

p - value p - value

Gold prices Lngold 0.9895 0.0000

USD to VND Lnusd_vnd 0.9670 0.0000

exchange rates

Inflation Lninflation 0.6980 0.0012

Interest rates Lninterestrate 0.8428 0.0000

Crude oil prices Lnoil 0.8626 0.0000

Page 33
 The factors affecting the gold market in Vietnam
March 29, 2014
4.5.2. Cointegration test using the Johansen methodology and the optimal time lag of
the model

a. Cointegration test using the Johansen methodology

Table 6: Cointegration test results

Date: 03/28/14 Time: 15:33

Sample (adjusted): 7/06/2008 1/26/2014
Included observations: 291 after adjustments
Trend assumption: Linear deterministic trend
Series: LNOIL LNINTERESTRATE LNINFLATION
LNUSD_VND LNGOLD
Lags interval (in first differences): 1 to 4
Unrestricted Cointegration Rank Test (Trace)

Hypothesized Trace 0.05

No. of CE(s) Eigenvalue Statistic Critical Value Prob.**

None * 0.293216 142.0671 69.81889 0.0000

At most 1 0.065581 41.08111 47.85613 0.1861
At most 2 0.036527 21.34253 29.79707 0.3366
At most 3 0.026005 10.51404 15.49471 0.2432
At most 4 0.009734 2.846502 3.841466 0.0916

Trace test indicates 1 cointegrating eqn(s) at the 0.05 level

* denotes rejection of the hypothesis at the 0.05 level
Unrestricted Cointegration Rank Test (Maximum Eigenvalue)

Hypothesized Max-Eigen 0.05

No. of CE(s) Eigenvalue Statistic Critical Value Prob.**

None * 0.293216 100.9860 33.87687 0.0000

At most 1 0.065581 19.73858 27.58434 0.3595
At most 2 0.036527 10.82849 21.13162 0.6644
At most 3 0.026005 7.667543 14.26460 0.4135
At most 4 0.009734 2.846502 3.841466 0.0916

Max-eigenvalue test indicates 1 cointegrating eqn(s) at the 0.05 level

* denotes rejection of the hypothesis at the 0.05 level
**MacKinnon-Haug-Michelis (1999) p-values
Page 34
 The factors affecting the gold market in Vietnam
March 29, 2014
We have an assumption that H0 has at most r cointegrations with the variables in the model.
The data used in this test is at 5% significance level.

The results generated from both methods of this test (Trace Statistics and Maximum
Eigenvalue) verify a cointegration among the variables in the model. On this basis, the
researcher would later begin the process of estimating the VECM.

b. The optimal time lag of the model

Before starting the process of estimating the VECM, its optimal time lag has to be selected.

Table 7: Criterion to the optimal time lag

VAR Lag Order Selection Criteria

Endogenous variables: DLNOIL DLNINTERESTRATE DLNINFLATION
DLNUSD_VND DLNGOLD
Exogenous variables: C
Date: 03/28/14 Time: 15:48
Sample: 6/01/2008 2/09/2014
Included observations: 291

Lag LogL LR FPE AIC SC HQ

0 3246.011 NA 1.46e-16 -22.27499 -22.21188 -22.24971

1 3655.791 802.6605 1.04e-17 -24.91952 -24.54083* -24.76782*
2 3668.798 25.03061 1.12e-17 -24.83710 -24.14283 -24.55897
3 3677.807 17.02740 1.26e-17 -24.72719 -23.71734 -24.32264
4 3777.194 184.4311* 7.54e-18* -25.23845* -23.91302 -24.70748

* indicates lag order selected by the criterion

LR: sequential modified LR test statistic (each test at 5% level)
FPE: Final prediction error
AIC: Akaike information criterion
SC: Schwarz information criterion
HQ: Hannan-Quinn information criterion

Page 35
 The factors affecting the gold market in Vietnam
March 29, 2014
As seen from table 7, three different information criterion were computed for different time
lags, each at 5% level of Likelihood Ratio (LR), Final Predict Error (FPE), and Akaike
Information Criteria (AIC). The results show that a fourh degree time lag is the optimal time
lag of the model, which equals to four weeks.

4.5.3. The process of estimating the VECM

a. Long-run relationships between gold prices and other variables

Table 8: Regression estimation results and the long-run effects of economic variables on
gold prices

Vector Error Correction Estimates

Date: 03/28/14 Time: 16:26
Sample (adjusted): 7/06/2008 1/26/2014
Included observations: 291 after adjustments
Standard errors in ( ) & t-statistics in [ ]

Cointegrating Eq: CointEq1

LNGOLD(-1) 1.000000
LNOIL(-1) 2.600741
(0.41009)
[ 6.34192]
LNINTERESTRATE(-1) -2.583472
(0.50048)
[-5.16195]

LNINFLATION(-1) 0.144729
(0.10288)
[ 1.40679]

LNUSD_VND(-1) -9.823394
(0.88652)
[-11.0809]

C 29.21724

Page 36
 The factors affecting the gold market in Vietnam
March 29, 2014

The VECM regression results suggest that there are long-run relationships between the
logarithm of gold (lngold) and the logarithms of the other variables (lnusd_vnd, lninterestrate
and lnoil) at 10% significance level.

With H0: a statistically significant regression coefficient;

Which means H0 : βj = 0

At 10% significance level, this is equivalent to:

t n-k0.05 = 1.6772

H0 is rejected when |t – value| t n-k0.05.

Table 9: The list of long-run variable coefficients

Lngold

Variable
Long-run coefficient t - value t n-k0.05

C -29.21724 -

Lnusd_vnd 9.823394 -11.0809

Lninflation -0.144729 1.40679 1.6772

Lninterestrate 2.583472 -5.16195

Lnoil -2.600741 6.34192

Page 37
 The factors affecting the gold market in Vietnam
March 29, 2014
Conclusions: At 10% significance level,

 The regression coefficient of lnusd_vnd is equal to 9.823394 and is statistically

significant (t – value= -11.0809). This suggests that in the long run, the average price of
gold in Vietnam would increase by 9.823394% when the USD to VND exchange rate
increased by 1%. This result matches the researcher’s expectations.
 The regression coefficient of lninflation is equal to -0.144729, but it is still
statistically significant (t – value= 1.40679). This suggests that inflation had no influence
on gold prices in the long run. However, there was an influence as such in the short run.
 The regression coefficient of lninterestrate is equal to 2.583472 and is statistically
significant (t – value= -5.16195). This suggests that in the long run, the average price of
gold in Vietnam would increase by 2.583472% when the rate of interest increased by 1%.
This result contradicts with the researcher’s expectations.
 The regression coefficient of lnoil is equal to -2.600741 and is statistically significant
(t – value= 6.34192). This suggests that in the long run, the average price of gold in
Vietnam would decrease by 2.600741% when the price of crude oil increased by 1%. This
result matches the researcher’s expectations.

b . Short-run relationships between gold prices and other variables

Coefficient R – square = 0.503691 indicates that the model reflects 50.3691% of its total data,
or, in other words, the model produces 50.3691% accurate estimations on its data.

Coefficient Adjusted R - square = 0.464945 indicates that the model’s regression variables
reflects 46.4945% of its total data after its expected variables have been adjusted.

Page 38
 The factors affecting the gold market in Vietnam
March 29, 2014
Table 10: The list of short-run regression coefficients

Coefficient Std.Error t - statistic t n-k0.05

C 0.000757 0.00323 0.23470

D(Lngold (-1)) 0.232816 0.24335 0.95673
D(Lngold(-2)) -0.0059716 0.27172 -0.21977
D(Lngold(-3)) -0.219794 0.27181 -0.80862
D(Lngold(-4)) 0.114576 0.24275 -0.80862
D(Lnusd_vnd(-1)) 0.650903 0.43628 1.49192
D(Lnusd_vnd(-2)) 0.655432 0.50383 1.30091
D(Lnusd_vnd(-3)) 0.724649 0.50609 1.43185
D(Lnusd_vnd(-4)) -0.153565 0.41155 -0.37314
D(Lninflation(-1)) 0.018601 0.12636 0.14721
D(Lninflation(-2)) -0.020347 0.154556 -0.13164
1.6772
D(Lninflation(-3)) -0.036991 0.15460 -0.23928
D(Llamphat(-4)) 0.057490 0.12659 0.45413
D(Lninterestrate(-1)) -0.312765 0.36759 -0.85086
D(Lninterestrate(-2)) -0.119202 0.42518 -0.28036
D(Lninterestrate (-3)) -0.073524 0.42518 -0.28036
D(Lninterestrate (-4)) 0.038288 0.32139 0.11913
D(Ldau (-1)) -0.116647 0.22879 0.47200
D(Lnoil (-2)) -0.104805 0.25639 -0.40878
D(Lnoil(-3)) -0.109140 0.25643 -0.40878
D(Lnoil(-4)) -0.062660 0.22708 -0.42562
ECM(-1) 0.063376 0.00644 9.83558
R-squared 0.503691
Adj. R-squared 0.464945

Page 39
 The factors affecting the gold market in Vietnam
March 29, 2014
a. The influence of the USD to VND exchange rates on gold prices
Table 11: The effects of the USD to VND exchange rates on gold prices

Variable Coefficient Std. Error t-Statistic t n-k0.05

Lnusd_vnd -9.823394 0.88652 -11.0809

D(Lnusd_vnd(-1)) 0.650903 0.43628 1.49192 1.6772

D(Lnusd_vnd(-2)) 0.655432 0.50383 1.30091

D(Lnusd_vnd(-3)) 0.724649 0.50609 1.43185

D(Lnusd_vnd(-4)) -0.153565 0.41155 -0.37314

Table 12 demonstrates the influence of the USD to VND exchange rates on gold prices, both in
the short run (at a four-week time lag) and in the long run. In the short run, these exchange
rates had no influence on gold prices because all regression coefficients generated in the next
four weeks are not statistically significant. Yet, these coefficients are statistically significant in
the long run (t - value = -11.0809), suggesting that the USD to VND exchange rates did have
an influence on gold prices but only after a period of at least four weeks.

Page 40
 The factors affecting the gold market in Vietnam
March 29, 2014
b. The influence of inflation on gold prices
Table 12: The effects of inflation on gold prices

Variable Coefficient Std. Error t-Statistic t n-k0.05

Lninflation -0.144729 0.10288 1.40679

D(Lninflation(-1)) 0.018601 0.12636 0.14721 1.6772

D(Lninflation(-2)) -0.020347 0.154556 -0.13164

D(Lninflation(-3)) -0.036991 0.15460 -0.23928

D(Lninflation(-4)) 0.057490 0.12659 0.45413

Table 13 shows the influence of inflation on gold prices, both in the short run and in the long
run. In the long run, inflation had no influence on gold prices for none of the generated
regression coefficients are statistically significant at 10% level. Similarly, no influence as such
was found in the short run at any of the four time lags. This can be explained by the fact that
all regression coefficients had a (t – value) smaller than 1.6772. On this basis, it can be
concluded that gold did not act as an inflation hedge in Vietnam.5

5
Similar conclusions can be found in studies previously conducted by Sherman (1983), Jaffe (1989) and Cecchetti et al
(2000), among others.
Page 41
 The factors affecting the gold market in Vietnam
March 29, 2014
c. The influence of interest rates on gold prices
Table 13: The effects of interest rates on gold prices

Variable Coefficient Std. Error t-Statistic t n-k0.05

Lninterestrate 2.583472 0.50048 -5.16195

D(Lninterestrate(- 1.6772
-0.312765 0.36759 -0.85086
1))

D(Lninterestrate(-
-0.119202 0.42518 -0.28036
2))

D(Lninterestrate (-
-0.073524 0.42518 -0.28036
3))

D(Lninterestrate (-
0.038288 0.32139 0.11913
4))

Table 14 indicates the influence of interest rates on gold prices, both in the short run (at a four-
week time lag) and in the long run. In the short run, interest rates had no influence on gold
prices because none of the generated regression coefficients are statistically significant at 10%
level. In the long run, however, these coefficients are statistically significant (t - value = -
5.16195 > 1.6772). This suggests that interest rates did have an influence on gold prices but
only after a period of at least four weeks. In addition, their correlation was positive, which
means the price of gold would increase by 2.58% when interest rates increased by 1%.

Page 42
 The factors affecting the gold market in Vietnam
March 29, 2014
d. The influence of crude oil prices on gold prices
Table 14: The effects of crude oil prices on gold prices

Variable Coefficient Std. Error t-Statistic t n-k0.05

Lnoil -2.600741 0.41009 6.34192

D(Lnoil (-1)) -0.116647 0.22879 0.47200 1.6772

D(Lnoil (-2)) -0.104805 0.25639 -0.40878

D(Lnoil(-3)) -0.109140 0.25643 -0.40878

D(Lnoil(-4)) -0.062660 0.22708 -0.42562

Table 15 describes the influence of crude oil prices on gold prices, both in the short run (at a
four-week time lag) and in the long run. In the short run, crude oil prices had no influence on
gold prices because none of the generated regression coefficients are statistically significant.
Still, such an influence existed in the long run as the generated regression coefficients in this
case are statistically significant (t - value = 6.34192). In addition, the correlation between gold
prices and crude oil prices was negative.

e. The Error Correction Model (ECM)

As seen from table 11, coefficient ECM (-1) prepresents the error correction model of time
series and equals to 0.063376, with t – value = 9.83558 (statistically significant). This suggests

Page 43
 The factors affecting the gold market in Vietnam
March 29, 2014
that the price of gold would return to its normal value one week after it fell or rised and the
error correction rate would be 6.3376%.

2.11. 4.6. The impulse response function and the variance decomposition
4.6.1. The impulse response function

Conclusions:

• Most time series, especially the exchange rates – gold prices, interest rates – gold prices
and crude oil prices – gold prices pairs, had a tendency to move towards their normal values in
the long run.

• The price of crude oil had a negative correlation with the price of gold in the fifth week
after any movement in oil prices, or at a fifth degree time lag. This correlation lasted for
another five months before the price of gold returned to its normal value.

• Interest rates had a positive correlation with gold prices within one week after their
movements. This correlation lasted for approximately four weeks before an identical
correlation appeared in the twelfth week, which lasted for the next three weeks.

• The USD to VND exchange rates had a positive correlation with gold prices within one
week after their movements. This correlation lasted for four weeks before turning into a
negative one, which lasted for another five weeks, then it turned back to positive in the
eleventh week.

Page 44
 The factors affecting the gold market in Vietnam
March 29, 2014
Graph 1: The impulse response function

Response to Cholesky One S.D. Innovations

ResponseofofLNVANG
Response lngold to
to LNVANG
lngold Responseofof
Response lnoil to LNVANG
LNDAU lngold
.06 .06

.04
.04

.02
.02
.00

.00
-.02

-.02 -.04
10 20 30 40 50 60 10 20 30 40 50 60

Response of
Response of lninterestrate
LNLAISUAT to toLNVANG
lngold Response
ResponseofofLNLAMPHAT
lninflation to
to LNVANG
lngold
.06 .06

.04
.04

.02
.02
.00

.00
-.02

-.04 -.02
10 20 30 40 50 60 10 20 30 40 50 60

Response of LNUSD_VND to LNVANG

Response of lnusd_vnd to lngold
.06

.04

.02

.00

-.02

-.04
10 20 30 40 50 60

Page 45
 The factors affecting the gold market in Vietnam
March 29, 2014
Table 15: The variance decomposition

Period S.E. LNGOLD LNOIL LNINTERESTRATE LNINFLATION LNUSD_VND

1 0.053103 100.0000 0.000000 0.000000 0.000000 0.000000

2 0.073629 99.26352 0.171515 0.553231 0.011001 0.000735
3 0.085686 97.34910 0.608039 2.005106 0.024753 0.013003
4 0.092819 94.11703 1.325875 4.427329 0.029854 0.099914
5 0.097377 86.51717 3.084466 10.26424 0.038433 0.095693
6 0.103726 76.81403 5.290583 17.55656 0.039423 0.299408
7 0.112433 67.80124 7.275313 23.89534 0.036494 0.991612
8 0.121829 60.18511 8.858525 28.52577 0.034350 2.396238
9 0.128920 54.85914 10.01299 31.09717 0.041997 3.988699
…….
50 0.213204 32.21242 15.89846 41.85117 1.534890 8.503061
51 0.214485 32.08148 15.97777 41.88291 1.556013 8.501831
52 0.215792 31.91730 16.05153 41.95895 1.574871 8.497351
53 0.217148 31.70049 16.12155 42.09108 1.590293 8.496591
54 0.218566 31.42985 16.18904 42.27423 1.601973 8.504907
55 0.220039 31.11951 16.25543 42.48914 1.610508 8.525411
56 0.221539 30.79245 16.32267 42.70955 1.617176 8.558161
57 0.223027 30.47336 16.39245 42.91083 1.623520 8.599850
58 0.224470 30.18290 16.46518 43.07634 1.630899 8.644687
59 0.225850 29.93384 16.53958 43.20010 1.640155 8.686324
60 0.227169 29.72916 16.61330 43.28612 1.651482 8.719939

4.6.2. The variance decomposition

Conclusions:

Aside from gold, which always has the greatest influence on its own, interest rates were found
to have had the most significant influence on gold prices in the long run as they were
responsible for 23.89% to 43.29% of the movements in gold prices. The next most significant
factor was crude oil prices as they caused 8.86% to 16.61% of the movements in gold prices.

Page 46
 The factors affecting the gold market in Vietnam
March 29, 2014
The USD to VND exchange rates were slightly less influential as they lead to 2.39% to 8.71%
of the cases where gold prices fluctuated. Inflation, on the other hand, had little to no influence
on gold prices as only 0% to 1.65% of the cases where gold prices fluctuated were caused by
this factor.

2.12. 4.7. Residual’s stationarity test

A residual’s stationarity test serves as a solution for Spurious regressions, which have been
introduced in chapter 3.

Table 16: Residual’s stationarity test results

Group unit root test: Summary

Series: RESID01, RESID02, RESID03, RESID04, RESID05
Date: 03/29/14 Time: 00:57
Sample: 6/01/2008 2/15/2014
Exogenous variables: Individual effects
Automatic selection of maximum lags
Automatic selection of lags based on SIC: 14
Newey-West bandwidth selection using Bartlett kernel
Balanced observations for each test

Cross-
Method Statistic Prob.** sections Obs
Null: Unit root (assumes common unit root process)
Levin, Lin & Chu t* 12.2011 1.0000 5 1380

Null: Unit root (assumes individual unit root process)

Im, Pesaran and Shin W-stat -7.32414 0.0000 5 1380
ADF - Fisher Chi-square 79.8205 0.0000 5 1380
PP - Fisher Chi-square 667.404 0.0000 5 1450

** Probabilities for Fisher tests are computed using an asymptotic Chi

-square distribution. All other tests assume asymptotic normality.

Page 47
 The factors affecting the gold market in Vietnam
March 29, 2014
From the results demonstrated in table 16, which were generated in both the Phillips – Person
test (PP) and the ADF test, it can be clearly seen that p – value = 0, which means the residual
of the model is a stationary time series at 10% significance level.

2.13. 4.8. Testing for the stability of the regression model using a unit circle
Graph 2: Unit circle test results

Inverse Roots of AR Characteristic Polynomial

1.5

1.0

0.5

0.0

-0.5

-1.0

-1.5
-1.5 -1.0 -0.5 0.0 0.5 1.0 1.5

The test results suggests that all observations lied inside the unit circle. On this basis, it can be
concluded that the researcher applied a stable regression model.

Page 48
 The factors affecting the gold market in Vietnam
March 29, 2014
CHAPTER 5

CONCLUSIONS
2.14. 5.1. Summarizing the experimental evidence found in Vietnam
After all necessary test results were generated and all experimental evidences were found in
the Vietnamese gold market during the period between June 2008 and February 2014, the
researcher could finally answer the following questions in order to fulfill the aim of this thesis:

 What factors affect gold prices in Vietnam?

According to the findings, gold prices in Vietnam are affected by three factors, namely the
USD to VND exchange rates, interest rates and crude oil prices. Among these factors, interest
rates have the most significant influence on gold prices. The only factor that does not affect
gold prices is inflation. These conclusions match those presented in the study conducted by
Larson and McQueen (1995)6.

 How should the influence of these factors on gold prices be assessed?

The influence of the mentioned factors on gold prices is only statistically significant in the
long run at 10% level:

 Gold prices in Vietnam will rise by 9.823394% when the USD to VND exchange rates
rise by 1%.
 Gold prices in Vietnam will rise by 2.583472% when interest rates rise by 1%.
 Gold prices in Vietnam will rise by 2.600741% when crude oil prices rise by 1%.
 Gold prices in Vietnam will remain unchanged when movements in inflation occur.

The results regarding the USD to VND exchange rates and crude oil prices match both the
researcher’s expectations and the findings presented in previous studies but the results
regarding interest rates do not.7

The reason for this is the fact that in the short run, people are more likely to make deposits than
investing in gold when interest rates increase. However, a constant increase in interest rates in
the long run will cause the interest expense to rise. Consequently, businesses, especially

6
Larson and McQueen (1995) found a positive correlation between gold prices and inflation, yet this correlation was not
statistically significant.
7
Details can be found in chapter 2.
Page 49
 The factors affecting the gold market in Vietnam
March 29, 2014
manufacturing companies, will have to bear higher heads-up costs, which will then lead to an
increase in consumer prices and the inflation rate. The rising inflation rate will result in the
devaluation of the VND, which will be a reason for people to lose faith in this currency and to
invest in such a safe haven as gold. Eventually, higher demands for gold will cause its price to
rise.
2.15. 5.2. Applying the study in reality
Apart from its applicability to investors, the thesis can also be useful for government agencies
in Vietnam because they need to observe the movements in the USD to VND exchange rates,
inflation, crude oil prices and interest rates in order to manage the gold market in Vietnam and
even to boost the entire economy of the country.
2.16. 5.3. Limitations and recommendations
The researcher would like to review some limitations of the thesis as well as to propose some
recommendations for future studies:

- The data collection process was the most challenging stage during the making of the
thesis due to the inconsistency of time series. Therefore, the generated test results
were less subjective. In the future, researchers need to seek a better, more effective
method of collecting data.
- The researcher only examined the relationships between gold prices and four factors,
whereas the actual number of factors affecting them is much higher.
- The explanations for the correlations between gold prices and other economic
variables are not yet perfect.
- The only model implemented by the researcher was the VECM, which is commonly
used in Vietnam. The use of more, better models may generate more reliable results
in future research.

Page 50
 The factors affecting the gold market in Vietnam
March 29, 2014
APPENDICES
Figure 1: Variables’ stationarity test results

Null Hypothesis: LNGOLD has a unit root

Exogenous: None
Lag Length: 15 (Automatic based on SIC, MAXLAG=15)

t-Statistic Prob.*

Augmented Dickey-Fuller test statistic 2.003765 0.9895

Test critical
values: 1% level -2.573247
5% level -1.941961
10% level -1.615942

*MacKinnon (1996) one-sided p-values.

Null Hypothesis: D(LNGOLD) has a unit root

Exogenous: None
Lag Length: 14 (Automatic based on SIC, MAXLAG=15)

t-Statistic Prob.*

Augmented Dickey-Fuller test statistic -6.845247 0.0000

Test critical
values: 1% level -2.573247
5% level -1.941961
10% level -1.615942

*MacKinnon (1996) one-sided p-values.

Null Hypothesis: LNUSD_VND has a unit root

Exogenous: None
Lag Length: 15 (Automatic based on SIC, MAXLAG=15)

t-Statistic Prob.*

Augmented Dickey-Fuller test statistic 1.498353 0.9670

Test critical
values: 1% level -2.573247
5% level -1.941961
10% level -1.615942

*MacKinnon (1996) one-sided p-values.

Page 51
 The factors affecting the gold market in Vietnam
March 29, 2014

Null Hypothesis: D(LNUSD_VND) has a unit root

Exogenous: None
Lag Length: 14 (Automatic based on SIC, MAXLAG=15)

t-Statistic Prob.*

Augmented Dickey-Fuller test statistic -8.925585 0.0000

Test critical
values: 1% level -2.573247
5% level -1.941961
10% level -1.615942

*MacKinnon (1996) one-sided p-values.

Null Hypothesis: LNINFLATION has a unit root

Exogenous: None
Lag Length: 15 (Automatic based on SIC, MAXLAG=15)

t-Statistic Prob.*

Augmented Dickey-Fuller test statistic 0.050039 0.6980

Test critical
values: 1% level -2.573247
5% level -1.941961
10% level -1.615942

*MacKinnon (1996) one-sided p-values.

Null Hypothesis: D(LNINFLATION) has a unit root

Exogenous: None
Lag Length: 14 (Automatic based on SIC, MAXLAG=15)

t-Statistic Prob.*

Augmented Dickey-Fuller test statistic -3.256423 0.0012

Test critical
values: 1% level -2.573247
5% level -1.941961
10% level -1.615942

Page 52
 The factors affecting the gold market in Vietnam
March 29, 2014
*MacKinnon (1996) one-sided p-values.

Null Hypothesis: LNINTERESTRATE has a unit root

Exogenous: None
Lag Length: 14 (Automatic based on SIC, MAXLAG=15)

t-Statistic Prob.*

Augmented Dickey-Fuller test statistic 0.587890 0.8428

Test critical
values: 1% level -2.573217
5% level -1.941957
10% level -1.615945

*MacKinnon (1996) one-sided p-values.

Null Hypothesis: D(LNINTERESTRATE) has a unit

root
Exogenous: None
Lag Length: 13 (Automatic based on SIC, MAXLAG=15)

t-Statistic Prob.*

Augmented Dickey-Fuller test statistic -5.203603 0.0000

Test critical
values: 1% level -2.573217
5% level -1.941957
10% level -1.615945

*MacKinnon (1996) one-sided p-values.

Null Hypothesis: LNOIL has a unit root

Exogenous: None
Lag Length: 15 (Automatic based on SIC, MAXLAG=15)

t-Statistic Prob.*

Augmented Dickey-Fuller test statistic 0.682329 0.8626

Test critical
values: 1% level -2.573247
5% level -1.941961
10% level -1.615942

*MacKinnon (1996) one-sided p-values.

Page 53
 The factors affecting the gold market in Vietnam
March 29, 2014

Null Hypothesis: D(LNOIL) has a unit root

Exogenous: None
Lag Length: 14 (Automatic based on SIC, MAXLAG=15)

t-Statistic Prob.*

Augmented Dickey-Fuller test statistic -6.101096 0.0000

Test critical
values: 1% level -2.573247
5% level -1.941961
10% level -1.615942

*MacKinnon (1996) one-sided p-values.

Figure 2: Estimating the short-run correlations between gold prices and exchange rates,
inflation, interest rates, and crude oil prices on the basis of the VECM regression results

Error Correction: D(LNGOLD) D(LNOIL)D(LNINTERESTRATE)D(LNINFLATION)D(LNUSD_VND)

CointEq1 0.063376 0.065401 0.066348 0.070491 0.063958

(0.00644) (0.00661) (0.00632) (0.00769) (0.00623)
[ 9.83558] [ 9.90094] [ 10.5029] [ 9.16867] [ 10.2681]

D(LNGOLD(-1)) 0.232816 -0.285519 -0.222849 -0.175647 -0.206943

(0.24335) (0.24946) (0.23857) (0.29035) (0.23523)
[ 0.95673] [-1.14454] [-0.93411] [-0.60495] [-0.87973]

D(LNGOLD(-2)) -0.059716 -0.280801 -0.251514 -0.262024 -0.246229

(0.27172) (0.27856) (0.26639) (0.32421) (0.26267)
[-0.21977] [-1.00806] [-0.94415] [-0.80819] [-0.93742]

D(LNGOLD(-3)) -0.219794 -0.323186 -0.304401 -0.298786 -0.304017

(0.27181) (0.27865) (0.26648) (0.32432) (0.26275)
[-0.80862] [-1.15985] [-1.14231] [-0.92128] [-1.15704]

D(LNGOLD(-4)) 0.114576 0.442408 0.473256 0.462871 0.475362

Page 54
 The factors affecting the gold market in Vietnam
March 29, 2014
(0.24275) (0.24885) (0.23798) (0.28964) (0.23466)
[ 0.47200] [ 1.77782] [ 1.98863] [ 1.59812] [ 2.02579]

D(LNOIL(-1)) -0.116647 0.385489 -0.068805 -0.103001 -0.069869

(0.22879) (0.23454) (0.22430) (0.27299) (0.22117)
[-0.50984] [ 1.64358] [-0.30675] [-0.37731] [-0.31591]

D(LNOIL(-2)) -0.104805 0.119166 -0.099037 -0.101114 -0.091498

(0.25639) (0.26283) (0.25135) (0.30591) (0.24784)
[-0.40878] [ 0.45339] [-0.39401] [-0.33054] [-0.36918]

D(LNOIL(-3)) -0.109140 0.021782 -0.106255 -0.133688 -0.106070

(0.25643) (0.26287) (0.25139) (0.30596) (0.24788)
[-0.42562] [ 0.08286] [-0.42267] [-0.43695] [-0.42791]

D(LNOIL(-4)) -0.062660 -0.415134 0.023125 0.026482 -0.001974

(0.22708) (0.23279) (0.22262) (0.27094) (0.21951)
[-0.27594] [-1.78332] [ 0.10387] [ 0.09774] [-0.00899]

D(LNINTERESTRATE(-
1)) -0.312765 -0.236396 0.234615 -0.408898 -0.335852
(0.36759) (0.37683) (0.36037) (0.43859) (0.35533)
[-0.85086] [-0.62734] [ 0.65104] [-0.93230] [-0.94517]

D(LNINTERESTRATE(-
2)) -0.119202 -0.113397 0.127232 -0.128253 -0.116792
(0.42518) (0.43587) (0.41684) (0.50731) (0.41101)
[-0.28036] [-0.26016] [ 0.30523] [-0.25281] [-0.28416]

D(LNINTERESTRATE(-
3)) -0.073524 -0.089564 0.086190 -0.036100 -0.072080
(0.42613) (0.43685) (0.41777) (0.50845) (0.41193)
[-0.17254] [-0.20503] [ 0.20631] [-0.07100] [-0.17498]

D(LNINTERESTRATE(-
4)) 0.038288 0.217916 -0.223631 0.163435 0.057262
(0.32139) (0.32947) (0.31508) (0.38347) (0.31067)
[ 0.11913] [ 0.66142] [-0.70976] [ 0.42620] [ 0.18432]

Page 55
 The factors affecting the gold market in Vietnam
March 29, 2014

D(LNINFLATION(-1)) 0.018601 0.002421 0.014435 0.692434 -0.004541

(0.12636) (0.12954) (0.12388) (0.15077) (0.12215)
[ 0.14721] [ 0.01869] [ 0.11652] [ 4.59261] [-0.03718]

D(LNINFLATION(-2)) -0.020347 -0.030474 -0.019385 0.170945 -0.025338

(0.15456) (0.15845) (0.15153) (0.18442) (0.14941)
[-0.13164] [-0.19233] [-0.12793] [ 0.92695] [-0.16959]

D(LNINFLATION(-3)) -0.036991 -0.041742 -0.028201 -0.016570 -0.032636

(0.15460) (0.15848) (0.15156) (0.18446) (0.14944)
[-0.23928] [-0.26338] [-0.18607] [-0.08983] [-0.21838]

D(LNINFLATION(-4)) 0.057490 0.055138 0.043699 -0.043451 0.056873

(0.12659) (0.12978) (0.12411) (0.15105) (0.12237)
[ 0.45413] [ 0.42488] [ 0.35210] [-0.28767] [ 0.46475]

D(LNUSD_VND(-1)) 0.650903 0.611037 0.524102 0.461748 1.098222

(0.43628) (0.44725) (0.42772) (0.52056) (0.42174)
[ 1.49192] [ 1.36620] [ 1.22534] [ 0.88702] [ 2.60400]

D(LNUSD_VND(-2)) 0.655432 0.664193 0.605347 0.698076 0.836012

(0.50383) (0.51649) (0.49394) (0.60115) (0.48704)
[ 1.30091] [ 1.28597] [ 1.22555] [ 1.16124] [ 1.71653]

D(LNUSD_VND(-3)) 0.724649 0.725230 0.648141 0.800432 0.802045

(0.50609) (0.51882) (0.49616) (0.60385) (0.48923)
[ 1.43185] [ 1.39786] [ 1.30632] [ 1.32554] [ 1.63942]

D(LNUSD_VND(-4)) -0.153565 -0.282176 -0.270941 -0.572044 -0.558000

(0.41155) (0.42189) (0.40347) (0.49104) (0.39783)
[-0.37314] [-0.66883] [-0.67153] [-1.16496] [-1.40261]

C 0.000757 0.000522 -0.001054 -0.001894 -0.000505

(0.00323) (0.00331) (0.00316) (0.00385) (0.00312)
[ 0.23470] [ 0.15790] [-0.33341] [-0.49206] [-0.16188]

R-squared 0.503691 0.507290 0.524032 0.535446 0.512675

Page 56
 The factors affecting the gold market in Vietnam
March 29, 2014
Adj. R-squared 0.464945 0.468826 0.486875 0.499180 0.474631
Sum sq. resids 0.758572 0.797190 0.729086 1.079931 0.708850
S.E. equation 0.053103 0.054438 0.052061 0.063361 0.051334
F-statistic 13.00003 13.18857 14.10307 14.76429 13.47588
Log likelihood 452.7615 445.5367 458.5301 401.3689 462.6256
Akaike AIC -2.960560 -2.910905 -3.000207 -2.607346 -3.028355
Schwarz SC -2.682852 -2.633197 -2.722499 -2.329638 -2.750647
Mean dependent 0.002826 0.001990 -0.000707 -0.004971 0.001484
S.D. dependent 0.072598 0.074694 0.072678 0.089533 0.070822

Determinant resid covariance (dof adj.) 3.83E-18

Determinant resid covariance 2.59E-18
Log likelihood 3827.687
Akaike information criterion -25.51675
Schwarz criterion -24.06510

Figure 3: Complete variance decomposition

LNINTEREST LNINFLATIO
Period S.E. LNGOLD LNOIL RATE N LNUSD_VND

1 0.053103 100.0000 0.000000 0.000000 0.000000 0.000000

2 0.073629 99.26352 0.171515 0.553231 0.011001 0.000735
3 0.085686 97.34910 0.608039 2.005106 0.024753 0.013003
4 0.092819 94.11703 1.325875 4.427329 0.029854 0.099914
5 0.097377 86.51717 3.084466 10.26424 0.038433 0.095693
6 0.103726 76.81403 5.290583 17.55656 0.039423 0.299408
7 0.112433 67.80124 7.275313 23.89534 0.036494 0.991612
8 0.121829 60.18511 8.858525 28.52577 0.034350 2.396238
9 0.128920 54.85914 10.01299 31.09717 0.041997 3.988699
10 0.133259 51.38068 10.86861 32.27066 0.071960 5.408091
11 0.136025 49.83565 11.39249 32.31446 0.141440 6.315965
12 0.138491 50.11277 11.54577 31.51895 0.264738 6.557782
13 0.141365 51.45621 11.41027 30.31909 0.422975 6.391466
14 0.144214 52.88479 11.20582 29.17971 0.586742 6.142933

Page 57
 The factors affecting the gold market in Vietnam
March 29, 2014
15 0.146454 53.71455 11.11437 28.47900 0.729044 5.963050
16 0.148193 53.64100 11.18485 28.52075 0.828985 5.824419
17 0.150130 52.56491 11.37023 29.48036 0.880907 5.703602
18 0.152972 50.63807 11.58486 31.21272 0.889437 5.674917
19 0.156760 48.25361 11.78751 33.23905 0.870610 5.849228
20 0.160835 45.89859 12.00172 35.01215 0.845377 6.242166
21 0.164417 43.93313 12.25440 36.23288 0.830198 6.749386
22 0.167121 42.54590 12.53887 36.86099 0.835177 7.219066
23 0.169076 41.78118 12.81248 37.00472 0.864552 7.537071
24 0.170662 41.56197 13.02718 36.82210 0.916486 7.672262
25 0.172162 41.70648 13.16994 36.47172 0.983720 7.668146
26 0.173646 41.97107 13.26704 36.11079 1.055605 7.595498
27 0.175093 42.11886 13.35713 35.89563 1.121352 7.507024
28 0.176560 41.97912 13.46165 35.95825 1.172694 7.428285
29 0.178216 41.48435 13.57656 36.35804 1.205281 7.375761
30 0.180208 40.67830 13.68916 37.04169 1.219651 7.371202
31 0.182515 39.68827 13.79777 37.85819 1.221030 7.434738
32 0.184936 38.66957 13.91441 38.63166 1.217335 7.567023
33 0.187217 37.75481 14.05211 39.23473 1.216346 7.742003
34 0.189194 37.03028 14.21111 39.61736 1.223522 7.917726
35 0.190862 36.53304 14.37704 39.79284 1.241153 8.055923
36 0.192317 36.25243 14.53115 39.81058 1.268537 8.137304
37 0.193670 36.13310 14.66233 39.73716 1.302716 8.164687
38 0.194992 36.08726 14.77184 39.64641 1.339498 8.154986
39 0.196320 36.01888 14.86830 39.61033 1.374499 8.127990
40 0.197695 35.85179 14.95932 39.68495 1.404044 8.099898
41 0.199171 35.54969 15.04755 39.89380 1.425907 8.083047
42 0.200783 35.12076 15.13298 40.21906 1.439810 8.087394
43 0.202521 34.60722 15.21718 40.60869 1.447459 8.119453
44 0.204313 34.06681 15.30461 40.99758 1.451941 8.179057
45 0.206065 33.55524 15.39968 41.33079 1.456677 8.257616
46 0.207705 33.11480 15.50294 41.57733 1.464411 8.340514
47 0.209211 32.76862 15.61012 41.73194 1.476590 8.412727

Page 58
 The factors affecting the gold market in Vietnam
March 29, 2014
48 0.210603 32.51830 15.71460 41.80951 1.493254 8.464343
49 0.211920 32.34452 15.81122 41.83790 1.513293 8.493065
50 0.213204 32.21242 15.89846 41.85117 1.534890 8.503061
51 0.214485 32.08148 15.97777 41.88291 1.556013 8.501831
52 0.215792 31.91730 16.05153 41.95895 1.574871 8.497351
53 0.217148 31.70049 16.12155 42.09108 1.590293 8.496591
54 0.218566 31.42985 16.18904 42.27423 1.601973 8.504907
55 0.220039 31.11951 16.25543 42.48914 1.610508 8.525411
56 0.221539 30.79245 16.32267 42.70955 1.617176 8.558161
57 0.223027 30.47336 16.39245 42.91083 1.623520 8.599850
58 0.224470 30.18290 16.46518 43.07634 1.630899 8.644687
59 0.225850 29.93384 16.53958 43.20010 1.640155 8.686324
60 0.227169 29.72916 16.61330 43.28612 1.651482 8.719939

Page 59
 The factors affecting the gold market in Vietnam
March 29, 2014
REFERENCES
Vietnamese materials:
1. A guidebook on Eviews, Bui Duong Hai, The National Economics University of
Hanoi.
2. How to use Eviews, Phung Thanh Binh.
3. Econometrics course book.
4. Documents on SFR’s theoretical models.
5. Documents on Fulbright’s basic econometrics.
Foreign materials:
1. Enders, W., 2004, Applied Econometric Time Series, Wiley.
2. Engle, R.F. and Granger, C.W.J., 1987, Cointegration and Error-Correction:
Representation, Estimation and Testing, Econometrica
3. Ghosh, D.P., E.J. Levin, P. Macmillan and R.E. Wright and, 2004, Gold as an Inflation
Hedge?, Studies in Economics and Finance.
4. Ismail, Z., Yahya, A., Shabri, A., 2009, Forecasting gold prices using multiple linear
regression method, American Journal of Applied Sciences 6.
5. Jaeger, H., 2005, A tutorial on training recurrent neural networks, covering BPTT,
RTRL, EKF and the “echo state network” approach
6. Kolluri, B.R., 1981, Gold as a hedge against inflation: an empirical investigation.
7. Lampinen, A., 2007, Gold Investments and short- and long-run price determinants of
the price of gold.
8. Levin, E.J., Wright R.E., 2006, Short run and long run determinants of the gold price.
World Gold Council, London
9. Mahdavi, S., Zhou, S., 1997, Gold and commodity prices as leading indicators of
inflation: tests of long run relationships and predictive performance.
10. Gold as an Exchange Rate Hedge:Empirical Study in Malaysia Nur Hafidzah Idris,
Nasuhar Ab Aziz, Shafinar Ismail, Aqilah Nadiah Md. Sahiq, Syukurriah Idrus, Zarinah
Abu Yazid, Noor Azzura Mohamed.
11. Gold as a hedge against the dollar, Forrest Capie ,Terence C. Mills,Geoffrey Wood .
12. Using Eviews for Principles of ECONOMETRICS, Third Edition, Wiilam E.Griffiths,
R.Cater Hill, Guay C.Lim.
Page 60
 The factors affecting the gold market in Vietnam
March 29, 2014
13. Bjørnland, H.C. (2009). Oil Price Shocks and Stock Market Booms in an Oil
ExportingCountry. Scottish Journal of Political Economy, Scottish Economic Society,
56(2), pp.232-254
14. “OIL AND GOLD: CORRELATION OR CAUSATION?.”Thai-Ha Le, Youngho
Chang
15. Zhang, Y.J. and Y.M. Wei: The Crude Oil Market and the Gold Market: Evidence for
16. Cointegration, Causality and Price Discovery, Resources Policy. 35(3), 168-77, (2010)
17. Cai, J., Cheung, Y. -L., & Wong, M. C. S.: What moves the gold market? Journal of
Futures
18. Baker, S. A., and van Tassel, R.C., 1985, Forecasting the Price of Gold: A
Fundamentalist
19. Approach, Atlantic Economic Journal, vol. 13, pp. 43-51
20. Diba, B. and Grossman, H., 1984, Rational Bubbles in the Price of Gold, NBER
Working
21. Paper: 1300. Cambridge, MA, National Bureau of Economic Research
22. Dickey, D.A. and Fuller, W.A., 1979, Distribution of the Estimators for Autoregressive
23. Time Series with a Unit Root, Journal of the American Statistical Association, vol. 74.
24. Dooley, M.P., Isard, P. and Taylor, M.P., 1995, Exchange Rates, Country-specific
Shocks and
25. Gold, Applied Financial Economics, vol. 5, pp. 121-129
26. Engle, R.F. and Granger, C.W.J., 1987, Cointegration and Error-Correction:
Representation,
27. Estimation and Testing, Econometrica, vol. 55, pp. 251-276.
28. Ghosh, D.P., E.J. Levin, P. Macmillan and R.E. Wright and, 2004, Gold as an
InflationHedge?, Studies in Economics and Finance, vol. 22, no. 1, pp. 1-25
29. Godfrey, L., 1978a, Testing against General Autoreggressive and Moving Average
ErrorModels when the Regressors include Lagged Dependent Variables, Econometrica,
vol.46, pp. 1293-1301
30. Short-run and Long-run Determinants of the Price of Gold
31. Godfrey, L., 1978b, Testing for Higher Order Serial Correlation in Regression
Equations

Page 61
 The factors affecting the gold market in Vietnam
March 29, 2014
32. When Regressors include Lagged Dependent Variables, Econometrica, vol. 46, pp.
1303-1310
33. Harmston, S., 1998, Gold as a Store of Value, London, World Gold Council
34. Hendry, D.F., 1986, Econometric Modelling with Cointegrated Variables: An
Overview,
35. Oxford Bulletin of Economics and Statistics, vol. 46, pp. 201-212
36. Jarrett, P., 2005, Coping with the Inevitable Adjustment in the US Current Account,
37. Economics Department Working Paper no. 467, OECD, Paris
38. Johansen, S., 1988, Statistical Analysis of Cointegration Vectors, Journal of Economic
39. Dynamics and Control, vol. 12, pp. 231-54
40. Johansen, S., 1991, Estimation and Hypothesis Testing of Cointegration in Gaussian
Vector
41. Autoreggression Models, Economometrica, vol. 59, pp. 1551-80
42. Kinney, C.F., 1999, “The Individual Investor and Inflation-Indexed Bonds: Tax Issues
and
43. Indexed Bonds, F.J. Fabozzi and Associates, New Hope, Penn
44. Kaufmann, T. and Winters, R., 1989, The Price of Gold: A Simple Model, Resources
Policy, vol. 19, pp. 309-318
45. Koutsoyiannis, A., 1983 A Short-Run Pricing Model for a Speculative Asset, Tested
with
46. Data from the Gold Bullion Market, Applied Economics, vol. 15, pp. 563-581
47. Kolluri, B.R., 1981, Gold as a Hedge against Inflation: An Empirical Investigation,
Quarterly
48. Review of Economics and Business, vol. 21, pp. 13-24
49. Laurent, R.D., 1994, Is There a Role for Gold in Monetary Policy?, Economic
Perspectives
50. (Federal Reserve Bank of Chicago), vol. XVIII: pp. 2-14
51. Lawrence, C., 2003, Why is Gold Different from Other Assets? An Empirical
Investigation,
52. London, World Gold Council
53. Levin, E.J, A. Abhyankar, A., and D. Ghosh, 1994, Does the Gold Market Reveal Real

Page 62
 The factors affecting the gold market in Vietnam
March 29, 2014
54. Interest Rates?, Manchester School (supplement), vol. 62, pp. 93-103
55. Mahdavi, S. and S. Zhou, 1997, Gold and Commodity Prices as Leading Indicators of
56. Inflation: Tests of Long-Run Relationship and Predictive Performance, Journal
ofEconomics and Business, vol. 49, pp. 475-489
57. Moore, G., 1990, Gold Prices and a Leading Index of Inflation, Challenge, vol 33, pp.
52-56
58. Muscatelli, V.A. and S. Hurn, 1992, Cointegration and Dynamic Time Series
Models,Journal of Economic Surveys, vol. 6, pp. 1-43
59. Working Paper no. 91-120. Washington, International Monetary Fund
60. O’Connell, R., 2005, What Sets the Precious Metals Apart from Other Commodities?,
61. London, World Gold Council
62. Paulos, G.J., 2001, The Long Wave Gold Cycle, London, Gold-Eagle
63. Pesaran, M. H. and B. Pesaran, B., 1997, Working with Microfit 4.0:
InteractiveEconometric Analysis, Oxford, Oxford University Press
64. Ramsey, J.B., 1969, Tests for Specification Errors in Classical Linear Last Squares
Regression
65. Analysis, Journal of the Royal Statistical Society, B, pp. 350-71
66. Sjaastad, L.A. and F. Scacciallani, 1996, The Price of Gold and the Exchange Rate,
Journal of Money and Finance, vol. 15, pp. 879-897
67. Sherman, E., 1986, Gold Investment: Theory and Application, New York, Prentice Hall
68. Sherman, E. J., 1982, New Gold Model Explains Variations, Commodity Journal,vol.
17, pp.16-20
Websites:
1. http://idoc.vn/tai-lieu/phan-tich-du-lieu-va-hoi-qui-thu-tuc-co-ban-cua-phan-mem-
eviews-3.html
2. http://sfb649.wiwi.hu-berlin.de/fedc_homepage/xplore/tutorials/xegbohtmlnode37.html
3. http://www.marketoracle.co.uk/Article43656.html
4. http://sfb649.wiwi.hu-berlin.de/fedc_homepage/xplore/tutorials/xegbohtmlnode37.html
5. http://www.fxstreet.com/analysis/oil-trading-alert/2014/03/17/
6. http://forums.eviews.com/viewtopic.php?f=7&t=3799

Page 63

View publication stats