A brief overview
Vietnamislearlyaedglingmarketeonomy,withthemarket`seeds'wereonlyplanted in late 1980s Over the past 16 years, the national nanial system is stilldeveloping initsinfany Eventodaybasibankingfailitiesarenotfamiliarwiththeommunity.
Withthe reent opening of Vietnam's rstpost-warstok exhange, named asHo Chi
The Minh City Securities Trading Center has opened up new investment opportunities, sparking a renewed interest in financial markets However, after 30 months since the initial stock trade, many investors have come to understand that navigating the stock market presents significant challenges.
In Vietnam, common investment vehicles have traditionally lacked diversity, with gold being a primary choice for many, especially older individuals who remember the hyperinflation period of 1986-1989, during which gold maintained its value Additionally, the US Dollar has become increasingly significant in household savings, along with other hard currencies, such as the Euro, which is emerging as a competitor While land and real estate are not legally recognized as ownership, they are considered relatively safe investments that provide de facto individual ownership rights.
Since the 1990s, economic reforms have significantly improved living standards and savings among the population The rise of businesses dealing with foreign currencies and the challenges posed by exchange rate fluctuations have increased due to the removal of international trade barriers for domestic firms As a result, the importance of hard currencies in household savings has heightened public awareness and interest in understanding exchange rate behaviors and future trends.
This empirical study focuses on the exchange rates of major foreign currencies traded in Vietnam's foreign exchange markets, while also examining the impact of gold prices as an alternative form of currency influenced by global price movements Both currencies and gold are crucial to the population and serve as alternatives to corporate equities for investment Understanding their historical patterns and behaviors will enhance our comprehension of the investment preferences of individuals in Vietnam.
This paper is organized into several sections to provide a comprehensive understanding of Vietnam's financial economy reform, which has established foundational elements for the current financial market The following section will focus on key factors, including local currency, domestic inflation, and gold as a primary store of value, which will be the subjects of our subsequent analysis Section 2 offers an extensive introduction to Vietnam's economy, addressing its limited exposure to the outside world and aiming to familiarize readers with the ongoing reforms Section 3 reviews relevant literature on financial aspects and benefits pertinent to our analysis, presenting key hypotheses that will guide our research, along with a subsection on data treatment and estimation methods specific to this study Section 4 provides detailed empirical data to enhance our understanding of the research objectives in relation to our defined research questions, culminating in a summary of key findings that will lead to necessary inferences regarding the proposed hypotheses Finally, recommendations for Vietnam's exchange rates will be offered.
More words on the Vietnamese nanial reforms
Since the mid-1980s, Vietnam has gradually shifted away from its old centrally planned economy due to significant economic challenges faced in the early 1980s These challenges included low agricultural productivity, declining living standards in both rural and urban areas, and stagnation in industrial growth Despite being an agricultural nation, Vietnam paradoxically had to import food to meet its population's basic needs, a stark contrast to its earlier status as a rice exporter in the early 20th century The economic reforms began with the introduction of Vietnam's post-war foreign direct investment law in 1986, following the sixth Congress of the Vietnam Communist Party This reform was motivated by a new understanding of the economy that acknowledged the role and contributions of non-state economic actors, including private households, privately run companies, and foreign-invested enterprises.
The recognition of the private economy has significantly stimulated both domestic and foreign investment, serving as a key growth engine for the country over the past 16 years of reform Driven by rapid changes in economic structures, the financial system has also undergone a substantial transformation Similar to other reforms in Vietnam, financial restructuring was implemented in a top-down manner, beginning with the separation of commercial lending functions from the former quasi-central bank, the National Bank of Vietnam, which is the predecessor of today's State Bank of Vietnam This shift facilitated the development of a two-tiered banking system, paving the way for the emergence of shareholding banks.
Since the late 1980s and early 1990s, the financial landscape has evolved significantly, leading to a diverse ownership structure in the banking sector Unlike rigid state-run commercial banks, which currently number around five, a variety of non-bank financial companies have emerged, including leasing firms, non-deposit-taking institutions, and investment funds This shift highlights the growing complexity and diversification within the financial industry.
In addition to formal economic units, the informal economy in urban areas has seen a rise in private households offering financial services, such as gold and foreign currency trading shops These informal entities often act as commercial lenders for households that lack access to formal banking resources The resurgence of the informal economy, featuring both state-run and non-state financial participants, has significantly contributed to the revitalization of financial markets in Vietnam While still in its early stages, today's financial markets encompass essential components of a modern market-oriented economy, including interbank foreign exchange and wholesale lending markets, commercial money and capital markets, and the equity market launched in July 2000 Additionally, several futures trading floors have been introduced for commodities such as cashew nuts, cotton, rice, coffee, and aquaculture products.
With the resurgence of the rural financial sector in Vietnam's economy, residents are increasingly focusing on investment opportunities, particularly in US dollars, alongside real estate assets This study aims to explore these emerging assets, examining their volatility and varying price trends that have influenced the investment preferences of the population over a relatively short period.
Domesti urreny and hard urrenies
From a xed to a more exible exhange rate regime
Duringthelosed-doortime,Vietnam'sforeigntradewasmodest Mostofthetradefo- usedon old-timeSoialistountries,namely theformerSovietUnion,EasternEurope ountries (e.g., former East Germany, Czehoslovakia, Poland, Bulgaria), and China.
The Soviet ruble played a critical role as the currency for computation and payment in trade Throughout most of its history, the country maintained a fixed exchange rate regime, where the currency rate was predetermined by the government This system was expected due to the strictly centrally planned economy, which required all goods, including currencies, to have fixed prices However, the fixed regime failed to account for actual output growth, money demand, and purchasing power, as the concepts of supply and demand were replaced by state plans and distributions The need for change arose when the country began trading with non-socialist countries, particularly as the Soviet Union and former socialist nations reformed their economies.
In the late 1980s and early 1990s, Vietnam experienced an economic shakeout as foreign aid diminished and traditional Eastern European markets became inaccessible This shift created a pressing need for new trading partners, prompting Vietnam to reform its trading activities The term "hard currencies" gained prominence in the media, with the US Dollar quickly becoming a familiar currency, replacing the ruble Consequently, the exchange rate regime had no choice but to implement necessary changes to establish a new economic structure and secure its positions in trading, investment, and monetary settlements.
The exchange rate issues, particularly the USD against the VND, have become increasingly relevant for both the business community and the general population Businesses are seeking ways to hedge against currency risks, while individuals focus on preserving and appreciating their savings Since 1995, the challenge of holding foreign currencies, especially USD, has led to significant USD-denominated deposits and under-the-mattress holdings by households and enterprises During periods of USD shortages for debt settlements and trade payments, the government often intervenes by allowing foreign enterprises with export earnings to sell their hard currencies In the late 1990s, exporters were required to convert a substantial portion of their hard currency revenues Exchange rates are influenced by market expectations, government actions, and public sentiment Despite government efforts towards dedollarization in the mid-1990s, a large portion of the population still maintains their savings in USD deposits Additionally, the practice of lending in USD within the banking sector perpetuates the use of this foreign currency in the economy Overall, foreign currencies, particularly the US Dollar, play a crucial role in the economy, with many national savings held in USD Enterprises hold foreign currencies primarily for international payments and to hedge against future fluctuations, particularly appreciation The behavior of foreign currencies is now closely monitored by researchers, professionals, and individual investors alike.
National ination situation
During the command economy era, the Vietnamese government did not formally recognize inflation, a stance common among socialist economies This approach aligns with the prevailing economic doctrine that dismisses the existence of inflation and unemployment as significant issues.
However, the omputationof onsumer prieindex remainedin thepast, although re- eivedverymodestattentionomparedtothingslikeplans,ordersandaomplishment of orders One of the major reasons for use of CPI in this study is data availability.
This although an onlybe obtained after some painstaking eort is the only ination dataone an have inVietnam.
The transition to a more open world economy in Vietnam brought the reality of inflation to the forefront, particularly noticeable during the period from 1986 to 1992 This era marked significant changes as Vietnam abandoned its long-standing Soviet-style distribution system, including food stamps and ration books This shift was accompanied by substantial layoffs of state budget salary earners Inflation began to surge in 1986, and this upward trend persisted throughout the years that followed.
Vietnam'seonomy Finanialdataisevenaharderobjet,asagreedbyallonernedeonomists. was reorded in late 1986, when ination level inreased nearly 8 times on an annual basis We allthisperiodthe 1986-92 hyperination(gure (?? )shows thesituation.)
Between 1986 and 1992, Vietnam experienced a challenging high inflation period, leading to a significant decline in living standards for the average Vietnamese As purchasing power decreased, incomes struggled to keep pace, culminating in an alarming annualized inflation rate of 774.7% in December.
1986, andthe lowest 17.43%inApril 1990 Mean levelof annualized onsumerprieis
215.73% for the total period, and the median 91.56% per annum We also note that thisperiodsustainedfor about 60 months, nearly one third ofthe total lengthof time sinethebeginningof Vietnam's reform.
The high inflation trend was curtailed when new growth engines, particularly private and foreign-invested sectors, began to drive efficient production Inflows of foreign direct investment (FDI) initiated significant advancements in manufacturing, agricultural production, and service industries during the early 1990s The level of FDI experienced rapid growth, achieving record levels during this period.
Foreign Direct Investment (FDI) in 1996 reached an annual figure of USD 8.6 billion, significantly contributing to economic improvement Both FDI and domestic investments alleviated financial distress and enhanced residents' income levels This influx of capital led to a notable increase in the GDP growth rate, reflecting better production output and purchasing power Positive economic trends helped stabilize consumer prices, which dramatically declined and consistently reached acceptable levels between 1993 and 2002.
Ination levels of seleted ountries: 1993:01-2002:12 are shown in gure (2.2) be- low.
Nonetheless, as generally agreed among eonomists, despite the danger of hyperina- tion,a `reasonable'positive ination levelshould by no meansbe denouned We now understandthat anaordableinationhas several, butimportant,positive impatson
Vietnam's eonomi growth Firsto, some inationary gap is desirable forahieving
GDP growth in an agriculture-based economy can lead to increased prices for agricultural products, thereby enhancing income and purchasing power for the majority of the domestic population As this understanding becomes clearer to economic agents, economists and business professionals may recognize a concerning deflationary trend emerging, typically with a lag of about 12 months following the notable Asian financial crisis of 1997-98.
N.B.:ThemostvaryingCPIlineisVietnam's;ComparativeCPIlevelsareoftheUSA,UK,Japan, andtheEuropeanUnion(shorterline) CPIdatahavebeenmonthlyprielevels,annualizedtopossess integrity; statedinperent p.a Datafor developedeonomies omefromIMFinternational maroe- onomi statistis; datafor Vietnam fromnational soures: General Statistial Oặe, VNEonomi
CPIlevelintheaforesaidhart,inwhihrsttimesineitsreformin1986,theountry wasfaing downwardgeneralprietrend,ausingnew headahe of eonomi downturn and stagnation.
Not unexpetedly,in this period, FDI level in Vietnam dropped drastiallyto around
USD1billionperannum,omparedtothepeakofnearlyUSD9billionin1996 Literally in gures, for the 24-month period (2000-2001), the Vietnamese eonomy experiened
23 months in deation, with the most serious drop in prie is -2.60% (YOY) in July
2000, andtheonlyimprovementseen a yearlater, inJuly2001, +0.09% Average CPI level is -1.114%, with median almost idential -1.104% Given an estimated growth rate of GDP at about 7% during2002, ination also piked up positively,standing at
F ew words on the gold pries
Nowwelookatanotherindiatorofpubliondeneintermsofeonomistability: the prieof gold Goldis doubtless a preiousmetal that arrieslaborvalue and sarity.
Gold is highly valued in Vietnam, not only for its intrinsic worth but also due to its constant demand and the accessibility of information from local gold shops Economic literature often cites gold as a prime example of value retention Additionally, gold remains a common medium for transactions, particularly in real estate, which constitutes a significant portion of Vietnamese assets.
Intheabove graph(2.3), thedailygold priedatawere obtainedfrom marketsoures, reeting a onsensus of major gold trading households in Hanoi and Ho Chi Minh
MeznCo.'s research team has meticulously organized and updated a unique database for the city, showcasing the average daily price for buying and selling transactions This data differs from the concept of "transfer payment," which pertains only to bank account settlements for larger exchanges The unit price reflects these average daily transactions effectively.
VND1,000 per1/10 of aChinesetael of gold(1 tael1.1 grams).
In Vietnam, gold is viewed as a valuable asset, especially during times of uncertainty, as it tends to retain its value better than interest-bearing deposits or dividend-paying stocks Consequently, Vietnamese investors often turn to gold during periods of heightened uncertainty, given its liquidity compared to other financial assets Additionally, gold prices generally appreciate when the value of the Vietnamese Dong (VND) and the US Dollar (USD) depreciate Over the past five years, gold prices in Vietnam have remained relatively stable until early 2023.
In 2002, gold prices surged by 25.34%, reflecting a global trend rather than genuine domestic demand This increase occurred amidst a challenging world economy, particularly due to the slow recovery in the USA and concerns over a potential US-led war against Iraq.
Statistis on daily return from gold prie hanges during 2002 (through 236 obs., see gure(2.3))showsthatthemeanreturnis0.099%,maxgain2.1414%andtheworstloss
-2.2202% Standard deviation is 0.568%, skewness -11.613 and kurtosis 6.4895 This showsa learleptokurtidistribution,withsubstantialmassenteringoninbetweenof meanandmedianreturnvalues Alookattheabovefrequenydistributionof dailyre- turnfailitatesthisknowledge Thedistributionasunveiledbyoutstatistisisnothing like Gaussian, witha fewnotieableoutliers.
The financial markets in Vietnam have not been extensively analyzed within a detailed scientific framework, warranting a thorough discussion before delving into the more technical aspects of our current paper Recognizing the significance of these factors in both the contemporary economy and future developments highlights the necessity of this research, addressing the critical reasons behind it.
The growing importance of tri-monetary systems—gold and emerging markets—has led to an increasing demand from investors for diverse and well-informed investment decisions This necessitates comprehensive research efforts to understand behavioral patterns, encompassing aspects from risk management to asset allocation While extensive financial literature exists in developed countries, there is a notable lack of serious empirical studies focused on Vietnam Most existing papers are rudimentary and fail to address the complexities of the issue Our aim is to enhance the understanding of Vietnam's financial factors while applying global empirical literature to this largely unexplored emerging market, making our rationale both justifiable and self-explanatory.
Thissetion ontinueswitha reviewon relevant literature, whih enlightens ourdire- tion for the study, as well as the hief researh interest in the partiular situation of
Vietnam's foreign exchange market has been the subject of extensive international research, as highlighted in our literature review We summarize key empirical studies that address related issues globally, while also referencing existing research specific to Vietnam Despite the underdevelopment of empirical studies within the nation, this gap further motivates our investigation into long-standing questions that remain unanswered, particularly in light of the growing interest in the economy.
Related literature
World literature
The existing literature on the empirical properties of asset prices and returns, including foreign currencies and gold, as well as their relationship with endogenous economic variables like inflation, is extensive In this applied study, identifying relevant frameworks is crucial for achieving an efficient solution, although the task is complex and not straightforward We will review related literature that guides us toward a practical consideration, which may require modifications It is essential to adhere to economic theories that underpin our empirical hypotheses and methodologies Additionally, data challenges are inherent in studies of young emerging markets, regardless of the researcher's preparation and thoughtfulness This section will address these critical issues.
There exists a vast literature on issues relating to exhange rates Many of the re- searhes fous on the veriation of a Riardo propositionof purhasing power parity
In the early 19th century, significant attention was given to the theoretical foundations of exchange rate dynamics within economic settings, while many researchers explored the statistical properties and empirical evidence of exchange rates across different historical regimes The findings have been mixed, with ongoing debates among scholars regarding key definitions, such as the distinction between long-run and short-run dynamics Empirical studies often utilize large datasets spanning extensive periods, while others focus on modeling long-run equilibrium relationships among endogenous variables using short-run data Additionally, the interrelationships between money, bonds, and international asset portfolios are crucial in understanding exchange rate movements This highlights the importance of money as a durable productive asset and nominal bonds in converting money into portfolio assets, as well as the critical linkages between nominal spot exchange rates, equilibrium conditions, and forward exchange rates.
Purchasing Power Parity (PPP) is a condition that ensures equilibrium in open commodity markets It highlights how changes in exchange rates are necessary to establish differential rates of change in the purchasing powers of various currencies The presence or absence of PPP directly impacts the exchange risks associated with international investments, yet this relationship is often overlooked in existing literature.
Clearly, the above postulation is one of the major reasons that neessitate our study.
Analyzing Vietnam's exchange rates poses significant challenges due to the lack of essential macroeconomic data and the unique characteristics of a developing economy This study primarily focuses on capturing the empirical properties of exchange rates through univariate time series analysis We will explore various important models, utilizing available data to test the validity of the Purchasing Power Parity (PPP) theorem against random walks and to model the behavior of real exchange rates The following summary reviews key literature that informs our research questions and establishes a relevant framework for our empirical analysis.
Frenkel (1981:[9℄) ontributes to empirial knowledge about the exhange rates and pries during 1970s, when several ritial hanges took plae In this work, the au- thoronsidersbotheặienyofforeignexhangemarketand exhangerate movement.
Frenkel's main onlusions support the eặieny of foreign exhange markets during
In the 1970s, exchange rates experienced significant volatility due to macroeconomic shifts, with news playing a crucial role in this dynamic Adler and Lehmann (1983) examined the disparity between Purchasing Power Parity (PPP) beliefs and the observed martingale behavior of various currencies from 1964 to 1981, finding limited empirical support for long-run PPP Their research emphasized the need for model adjustments to address potential issues in previous studies, particularly regarding the nonstationarity of nominal exchange rates and Consumer Price Index (CPI) variables This work laid the groundwork for further univariate analyses of Real Exchange Rate (RER) behavior, highlighting multiple adjustment methods, with CPI being a key factor among others.
In light of the enormous interests in disovering data-dependent empirial results on
The Purchasing Power Parity (PPP) theory faces significant challenges in reconciling the high short-run volatility of real exchange rates (RER) with the observed mean reversion rate of 15% per annum Rogo (1996) provides a concise yet comprehensive overview of this doctrine, highlighting the complexities involved in achieving full reconciliation of these economic phenomena.
Despite the persistence of PPP deviations, there are suggestions for qualitative explanations that have sparked further research interests in econometric methodologies, such as multivariate VAR analysis However, Rogo expresses discomfort with his own reasoning regarding this puzzle Choi (1999) investigates real exchange rates, providing evidence of serial correlation in the log-differenced real exchange rates using a monthly forex sample from 1960 to 1993, specifically analyzing USD real exchange rates.
CAD,DEM,JPYandSFR.Thenullhypothesisofrandomwalksinlog-dierenedreal exhange ratesseems to beonrmed withmost urreniesforpost-Bretton Wood pe- riod However,inthelongerrun,theauthorsynthesizedthestatistisasaveriationof
PPP for3/4 urreniesintheonsideredsample Mixedresults have ledto ontinuous debate aboutsupportingevidene fororagainstthe long-standingPPP theorem.
In an earnest eort to improve empirial ndings, alleviating mixed (and somewhat not unambiguous) results on the PPP debate, Fl^ores, Jorion, Preumont and Szafarz
In 1999, a multivariate framework was proposed, emphasizing its advantages over univariate test frameworks and providing strong evidence supporting long-run Purchasing Power Parity (PPP) This framework allows for varying speeds of mean reversion in real exchange rates across different countries, particularly focusing on the US Dollar against major European currencies The article highlights the significance of both theoretical modeling and empirical research, showcasing the extensive literature dedicated to testing these concepts.
Cassel's Purchasing Power Parity (PPP) doctrine highlights the challenge of overcoming the random walk phenomenon, particularly in the context of univariate time series tests Despite mixed empirical results, which have generally supported PPP over long periods—typically around a century of monthly or quarterly data—there remains a significant research interest in refining testing methodologies This ongoing debate between unit-root proponents and PPP advocates has spurred the development of new estimators aimed at achieving more conclusive and persuasive outcomes.
Related literature in Vietnam
The exchange rate issue in Vietnam has become prominent since the onset of economic reforms, particularly highlighted by the 1997-1999 Asian financial crisis, which raised concerns about the future stability of regional currencies A key factor in this context is the foreign exchange rates, which are often viewed as operational matters and briefly discussed in professional reports However, general and qualitative assessments rarely tackle specific issues or offer a rigorous analytical framework, resulting in a lack of comprehensive understanding and leaving foreign exchange rate research largely unaddressed.
There is a lack of comprehensive research on exchange rates within Vietnam's economic literature However, various independent studies have highlighted the dynamics of the foreign exchange market during the reform period A notable study by Vuong et al (2001) analyzes the statistical properties of the US Dollar exchange rate, focusing on short-term forecasts using low-order ARIMA models Additionally, Vuong and Ngo (2002) examine the existence of a parallel exchange rate system in Vietnam as a potential buffer against the impacts of the regional financial crisis on the national financial system, ultimately advocating for the development of free-market exchange rate systems.
In a related study, Vuong (2003) examines the long-run and short-run dynamics of price-based exchange rate considerations Insights from participants at the Solvay seminar in May 2003 highlight the gaps in this research area Despite the lack of systematic literature on this topic, recent interest from various authors indicates a growing need for further investigation.
Motivation for the study
This dissertation explores the evolution and changes in Vietnam's financial markets during the reform period It emphasizes the significance of understanding the relationships between various variables in these emerging markets, alongside a general analysis of market operations Specifically, the focus is on the development of Vietnam's nascent market economy.
This ontext gives rise to our researh, of ourse, beause it provides us with several onrete motivationsasfollows:
Since the initiation of economic reforms in Vietnam in 1986, the foundations of a market economy have been established However, a more thorough examination of various aspects is necessary to substantiate this claim.
We have a good motivation to look at the foregin exhange market, one of the rstmarketsexistinginVietnam,andoperatingativelyduetointernationaltrade liberalization.
Understanding exchange rates and related issues is crucial for the expanding business community The absence of systematic and quantitative research on this topic has hindered market development, transactions, and financial instruments that could have operated effectively For instance, the introduction of currency options has been delayed, despite a long-standing demand from enterprises.
3 The eonomy of Vietnam used to undergo a vibrant and painful hyperination hapter The nationwas praised for its eortto urb ination problems inmid-
In the 1990s, the national problem was closely linked to exchange rate issues, with behaviors influenced by both factors To establish a true relationship between them, empirical research is necessary This challenges the classic Purchasing Power Parity (PPP) hypothesis, which has critical implications If we cannot provide evidence supporting the validity of PPP, we risk rejecting the policy implications that central banks and relevant authorities may have used to stabilize this vital market, as evaluations regarding currency valuation become increasingly complex and overwhelming.
This setion is to stress the importaneof methodoly and data inthis type of applied eonometris, by providingneessary materialsforbuildingthework.
Eonometri methods
The eonometri hypothesis
We have explored the richness of panel literature in examining the long-run relationship between various endogenous and exogenous variables and the nominal exchange rates of a domestic currency Understanding this long-run equilibrium relationship is crucial for policymakers, as it significantly influences their perspectives and decision-making processes In the following section, we will establish relevant paradigms for hypothesis testing.
The long-standing establishment of PPP dotrine of exhange rate equilibrium rela- tionshipan be postulatedbythesimpleequation:
? are are domesti and foreign urrenies, respetively; S denotes the ex- urreny (e.g in Vietnam's ase VND 15,500 per USD 1.0); and K a onstant that helpsdenethePPP relationshipinthe theoretisystem.
By denition, we superimpose thatthe relationship P =KSP
? holds inthe long run, andthisimpliesthatKisestablishedastheonstant Conventionally,inthelogarithmi form,eq.(1) isusuallyre-dened as: p t
(2) where small letters denote natural logarithms of previous variables We also have a re-arrangement of the above equality to have another meaningful relationship for the
PPP equilibriumwhenexhangerates areallowed to move byeonomi variables: s t
(3) where, representssome onstant thatestablishesthe equilibrium.
Johansen proedure (1991:[10℄), for analyzing the long-run relationship between sta- tionary(ornon-stationary)variablesisrepresentedbythefollowingunrestritedvetor autoregression(VAR)system:
(4) where X t is an (p1) vetor of stohasti variables, integrated of order one or less
) , thatis a white-noiseproess vetor with non-diagonal ovariane matrix; and a deterministi term, represented by a (p1) vetor; ặ isa (p1) oeặient vetor on lineartrend terms.
The VAR system an be transformed into a vetor error-orretion model (VECM) generallyrepresentedbyequation ([? ℄)
Naturally,I(2)andhigher-orderareexluded. the (pp) oeặient matrix an be deomposed into and , both being (r) matries Therelationis formulatedintheequation (6):
The (pr) matrix serves as the E-C coefficients matrix, measuring how variables within the system adjust to establish long-run behavioral equilibrium Additionally, this matrix indicates co-integrating vectors, which represent the system's long-run equilibria, provided that convergence occurs A useful guideline for modeling is that when series are differenced stationary and a co-integration relationship is confirmed, employing a Vector Error Correction Model (VECM) can be beneficial.
Two majorapproahes intreating VAR systems areinitiated in pioneeringtheories of
Engle-Granger(1987:[6℄)andJohansen(1988) InJohansen(1988,1991), theapproah reommends a foremost fous on short-run dynamis of exhange rates by regressing
Co-integrating equationsand long-runsystemequilibrium:
Asthenominal exchange rates and individual prices in both domestic and foreign countries exhibit nonstationarity, making the search for potential first-order co-integration a viable approach Co-integrating vectors in exchange dynamics, derived from the interaction of real exchange rate components, provide valuable insights into long-run equilibrium and the finite duration of convergence to the expected real exchange rate.
The re-parameterized equation (5) represents a vector error correction model (VECM), indicating the presence of integration through its rank If the rank is either full or zero, it signifies that no co-integrating equation exists in the long-run relationship Consequently, the system's reversion can be decomposed into a (p) matrix, where the columns are linearly independent co-integrating vectors This (p) matrix is interpreted as the adjustment matrix, which reflects the speed at which the system adjusts to past shocks in order to reduce deviations from the co-integrating relationships.
It is well known that a o-integration test represents a test for weaker version of PPP beausethesupportingargumentforPPPonlyrequiresthefollowingstohastivariable z t dened byeq.(8): z t
An informative test for co-integration in Vietnam is feasible due to limited data availability This study proposes two well-known tests based on the Engle-Granger (EG) and Johansen paradigms Specifically, the Johansen paradigm utilizes a likelihood ratio rank test to evaluate the given equation.
(9) assuggested byJohansen (1988, 1991) Finally,the traestatisti forthehypothesisif all theseriesintheVAR systemare stationary (i.e r=p),is asfollows:
The trae test statistiin eq.(10) appliesto examine the null hypothesis that at least r o-integrating relationships in X t system exist against an alternative of (r+1) o- integrating equations.
In establishing the form of desired VECM for PPP tests for dierent exhange rates, we rewritethe followingdynamis: p t
The other ECM equations are also in the similar form Thus, our vetor of stohas- ti variables is:X t
? t ) While in onduting these test onduts, we use the informationalriteria AIC/SC,forwhih smallervaluesarepreferred:
We also note that the non-standard ritial values are taken from Osterwald- Lenum
In 1992, findings were reported that slightly differ from those in Johansen and Juselius (1990) Alternatively, the Engle-Granger framework can be utilized to test the co-integrating relationship between non-stationary stochastic variables by examining the error correction (EC) terms in spurious regression This test can be conducted using either the Augmented Dickey-Fuller (ADF) or Phillips-Perron (PP) methods For a typical dynamic two-variable system, we can consider the following: y_t.
Our ECMan simplybe thefollowingdynamial system: y t
The system unveils thefollowing impositions If y t and x t are o-integrated, standard
VAR in dierene seletion is misspeied Thus, usingthis approah, after verifying theI(1)speiationforvariablesof themodel,wesimplyregress one variable against theothers(inourPPPtest,inludingp t
? t ) toexaminethestationarityofresidualse- ries Residualsstationarityan betestedusingeitherADForPP regressions,referring to modied ritialvaluesprovidedinPhillipsand Ouliaris(1988:[4℄).
Unit root testsand long-runequilibriumrelationship:
Critics of time series analysis emphasize the importance of identifying the presence of a unit root This study will focus on several key considerations, including the examination of regression variables within a system, which can be misleading when dealing with non-stationary variables Although the t-test statistics may offer insights, the acceptance or rejection of a unit root in our exchange rate data—whether in levels or changes—will clarify the characteristics of the time series In many cases, these conclusions can help elucidate the equilibrium relationship between the variables in question, such as the nominal exchange rate.
Below we look at the ADF analysis from another view, in whih the modied ADF paradigmproves usefulto ouronsideration of theexhangerate long-run equilibrium onditions,too.
Along-runequilibriumonditionofRERanbetestedintwoforms First,realexhange rate,havingaountedfordiereneof relative pries(loalandforeign),exhibitsrela- tive (weaker-form)PPP, shouldtheADF regression supportalternative of stationarity asina normal ADF regression, forinstane,the eq.(?? ) Now, onsiderthe regression
When this ADF onsideration fails to rejet H
To analyze long-run equilibrium possibilities, we must utilize structural regression techniques such as vector autoregression (VAR) or cointegrating vector models The application of Augmented Dickey-Fuller (ADF) regression can yield valuable insights, including the half-lives of real exchange rates (RER) and the speed of convergence to the long-run equilibrium mean RER, as discussed in various studies (Culver and Papell, 1999; Salehizadeh and Taylor, 1999; Christev and Noorbakhsh, 2000; Sarno, 2000; Sarno and Taylor, 1998; Lothian and Taylor, 1996) Additionally, for a typical variation of the Purchasing Power Parity (PPP) theorem, the time trend component is often removed to ensure economic consistency Consequently, our model will primarily include a constant term for both technical and economic reasons, along with lagged dependent variables and lag-differenced series.
On the dataset and eonomi v ariables
Relative prie ratio
Belowarethe graphs(4.2.1) of dierent prieratios,omputedasdierenes inlogs of exhange rates, stated per VND Relative prieratios (levels) arefor theentireperiod of1986:01-2002:12
The prie ratio is dened simply as the log-dierene between the two nations' CPI levelsindiatedbytheindex(nottheperentageperiodigrowth rates,asusuallypub- lished on oặial media.): R P
= ln(P i ) , and P i is the index of onsumer prie (weighted basket) for the nation i The above graph suppliesmonthlyCPI data.
Consumer prie index - CPI
CPI is reonstruted foreasier readings (without aeting the statisti omputations) as in the gure below By reonstruting, we have startingvalue of CPI is 100(%) in
Consumer Price Index (CPI) data in Vietnam has been available on a monthly basis since the 1990s, marking a significant development in economic statistics However, this data is often released later than other indicators, such as foreign exchange rates, which can be observed through market transactions Notably, when comparing CPI levels across developed economies over a span of approximately 17 years, there are distinct differences in the sales figures represented on the vertical axes of the graphs.
In analyzing the Consumer Price Index (CPI) growth among Japan, the UK, and Vietnam, it is evident that while all three countries have experienced growth, Japan's CPI has increased at a slower pace, with the UK showing the highest growth rate Notably, even in the UK, the price index has not doubled from its initial value of 100 after 17 years In stark contrast, Vietnam's CPI demonstrates a significant increase, characterized by a steep curve indicating a rapid inflation rate during the early reform years It wasn't until the late 1990s that Vietnam achieved a stabilized CPI, highlighting the substantial differences in inflation trends among these nations.
Asian nations faed the serious onsequenes of the regional (spread) nanial risis
(whih led many ountries to devalue their urrenies, and eonomies plunged into a reessionand deationstage).
Hyperination didnot solelyhappen inVietnam Several other Eastern European na- tionsalsoexperienedthesituation,movingoutoftheentrallyplannedeonomymodel.
Hyperinflation has been viewed as a significant disruption to the traditional price-planning mechanisms in former command economies The rapid price increases during the early reform period reflect a response to economic uncertainty and the necessary adjustments in demand and supply, particularly in economies where demand had long outstripped supply without effective solutions to restore balance Although state planning ostensibly replaced other market tools, it ultimately failed to address the underlying supply shortages in the economy.
In the context of the theonomic theory, the Consumer Price Index (CPI) significantly influences foreign exchange rates, particularly when a nation like Vietnam opens its economy to international interactions This article will explore the impact of hyperinflation on stabilizing the Purchasing Power Parity (PPP) relationship in the long run, specifically in the case of Vietnam, supported by empirical evidence.
Real exhange rates - RER
This study analyzes real effective exchange rates, calculated as the number of Vietnamese Dong (VND) per unit of key foreign currencies, specifically focusing on the USD, GBP, JPY, and the emerging EUR Unlike many other studies that use a common base currency such as USD, EUR, or GBP, our approach employs a unique methodology for time series analysis of exchange rates We believe this perspective is essential, as the VND is a low-value currency, which necessitates a distinct examination of hard currencies in relation to it.
The graphs in the appendix reveal the behavior patterns of real exchange rates in relation to bilateral relative prices, specifically those based on the Consumer Price Index (CPI) These graphs demonstrate similar trends across various exchange rate series of foreign currencies.
VNDvalues The majordierenes areaused by thedierenesof prie(measuredin naturallogarithm ofnational index),whih renderthedistane fromthe RERand the foreignprie(log) seriesdierent among theforeignurrenies inquestion Ingeneral, however, they tendto behave insome onsensus.
Bilateral relations between Vietnam and other countries should emphasize the adjustment of real exchange rates Our analysis utilizes scatter plots to illustrate data points that represent clusters of pairwise coordinates in the Consumer Price Index (CPI), which govern the time series of the real exchange rate (RER).
The scatter plots comparing consumer price movements between the US and UK show a significant divergence, with data points forming a nearly straight line that equalizes the plane quarter This notable difference highlights the impact of inflation on bilateral exchange rate movements, which will be explored in our subsequent analysis The bilateral pairwise US-UK Consumer Price Index (CPI) scatter diagram serves as an essential reference for understanding this disparity.
(100) Datasampleperiod: 1986:01-2002:12 ChartsunveiloverallpitureofCPI move- mentsoftheUSAinomparison withonernedountries.
We have disussed about the data used in this study After relevant transformation, they arereadyforbeingtested underourempirialframeworks.
5.1 An analysis of variables and the relevane of data used in the ointegration analysis
The nonstationary harater of the variables
Univariate time series in financial markets often exhibit non-stationary characteristics Numerous studies have documented evidence of non-stationarity in exchange rates and consumer prices measured as indices Table 1 presents the results of our unit root tests on univariate time series, both in levels and first differences, focusing on a sample period from January 1986 to December 2002, which encompasses the entire reform period.
RER ADF PP RER ADF PP
NominalER ADF PP NominalER ADF PP
CPI ADF PP CPI ADF PP
?indiates the null hypothesisof unitroot rejeted at 1(5)%level.
RER, nominal ER and CP index are providedin natural logarithms.
RER: ADF ritial values for US, UK, JP at 1(5)% level: -4.0070 (-
1(5)%level: -4.0061(-3.4329);for EU:-4.0158(-3.4376) Nominal ER:
At the 1% and 5% significance levels, the ADF critical values for the US, UK, and Japan are -4.0070 and -3.4333, respectively, while for the EU, they are -3.4727 and -2.8798 The PP critical values for the US, UK, and Japan are -4.0061 and -3.4329 at the same significance levels, and for the EU, the values are -3.4708 and -2.8789 Additionally, the CPI ADF critical values for the US, UK, Japan, and EU are -4.0070 and -3.4333, while the PP critical values for the same regions are -4.0061 and -3.4329, with the EU also showing a value of -4.0061.
(-3.4329). priesinlevelsarenon-stationary,giventheoptimallagtrunation,aordingminimiz- ing AIC and SC.Both ADF and PP test statistis annot rejet the null of unit root existene inunivariateseriesinlevels.
Vietnam oặially reognized other hard urrenies than the old Soviet rubble sine
June-1989 For this, trade and ommerein new hard urrenies beome more ative, andtheexhangeratesreetbetterommerialnatureoftheexhangeratesdynamis.
During the recent evolution of financial markets, the distinction between black market and official exchange rates has become negligible, making official market data relevant for understanding exchange rates within Vietnam's broader economic context The changes in the Real Exchange Rate (RER) of four key currencies from June 1989 to December 2002 indicate a lack of stationarity in RER variables, as illustrated in Figure 8.
RER timeseriesforUSD, JPY, GBP,and EUR 1989:01-2002:12 diatesomewhatsimilarbehavior,observingpeaks and valleys Nettingeet of pries logarithms has partially wiped out divergent behaviors of nominal exhange rates of these urrenies.
The sub-sample unit root tests indicate that the null hypothesis of non-stationarity cannot be rejected for all exchange rate series in levels, even at the 10% significance level, as the ADF and PP test statistics fall within the acceptance region of the empirical distribution However, the tests decisively reject the unit root hypothesis for all first-order differenced time series at the 1% level Additionally, the correlogram of the Real Exchange Rate (RER) and Consumer Price Index (CPI) for the US demonstrates a clear pattern of non-stationarity.
ACFdoesnotdieoutafterquiteanumberoflags Forinstane,atlag24,theautoor- relationoeặient is stillsigniant, with ratherlarge (positive) magnitude, as shown
I(1) At this point, we are ready for further examination of ointegration and other relationshipinequilibrium.
RER k ADF AIC(SC) PP RER k ADF AIC(SC) PP
: Nullhypothesisofunitroot rejetedat1(5)% level ADFregression spei- ationisbasedonlag trunation(k)that minimizesAIC/SC.RatesforRERinnatural logarithms RER levels: ADF ritial values for US, UK, JP at 1(5)% level: -3.4715(-
2.8792); for EU: -3.4835(-2.8845); PP ritial values for US, UK, JP at 1(5)% level: -
3.4715(-2.8792);for EU:-3.4831(-2.8844);RERrst-orderdierene: ADFritialvalues for US, UK, JP at 1(5)% level: -3.4715(-2.8792); for EU: -3.4839(-2.8847); PP ritial valuesforUS,UK,JPat1(5)%level: -3.4715(-2.8792);forEU:-3.4835(-2.8845).
Table 2: ADFand PP testson RERs
Evidene of the ointegration and PPP
Co-integrated relations among s t ; p t ; and p ? t
Engle-Granger (1987) work indiates that althoughnon-stationary, these types of eo- nomi variable an be onstruted into a linearlyombined system that is stationary.
The results of the Spei test outcomes can indicate the degree to which a long-run equilibrium is achievable based on short-run data sets This analysis is supported by findings from both the Engle-Granger and Johansen cointegration tests.
Co-integration underEngle-Grangertwo-stageanalysis:
Engle-Granger (1987) method proposes a framework for bivariate relationship that presents along-run equilibrium,in whih ase ouronsideration of exhange ratesan be implemented The trivariate system among s t
? t will be grouped into the followingeq.(13): s t
The last equation indiates that nominal exhange rates are not xed (for instane in government'spriepegpoliy);thatis,allowedtoutuate A visualhekforpossibly interrelateds t and d pp
? isprovidedingure[13℄ oftheappendix Thissituationistrue forVNDduringthesub-sampleperiod1989:06-2002:12 Resultsof regressionbased on thelastequation aresummarizedinthetable (3) asfollows.
The test statistic for the error-correction term \( u_{i,t} \) does not conform to the standard distribution typically used in the ADF and PP tests Therefore, our statistical inference relies on the critical values provided by Phillips-Ouliaris (1988) for multiple variables, specifically in our bivariate model where \( m = 2 \).
WehaveobtainedempirialresultsfromimplementingtheE-Gframeworkthatindiate o-integration betweenrelativepriedierene and spot exhange ratesforalldata se- riesinquestion Statistisare summarizedintable (4) Next,estimations of trivariate systems isprovidedin thetable (5). sUSD st = +1:2892 +1:3310dpp
Note: Standarderror(s.e.) isreportedinparentheses;t-Stat insquare brakets.
Table 3: E-G rst-stage bivariateOLS regression
N.E.R Equation ADF-Stat k AIC(SC)
Note:Standarderror(s.e.) isreportedinparentheses;t-Stat insquarebrakets;
? denotesthe nullhypothesisofnonstationarityrejetedat1%level,omparabletoPhillips-Ouliaris.v.,for m=2 Tests areondutedforperiod1986:01-2002:12 u^ t denoteserror-orretiontermfrom therst-stageE-Gregression
Table 4: E-G seond-stage bivariateregression usingtheADF paradigm s
Note: Standarderror(s.e.) isreportedinparentheses;t-Stat. insquare brakets.
Table5: E-G rst-stage trivariateOLS regression
The system separates domesti and foreign pries to reet the relative PPP relation more learly (Standarderrors inthissituationare biased;reportedforompleteness.)
In the next tabulated summary (6), we provide test outomes using the E-G seond- stage trivariateregression basedon theaugmented Dikey-Fuller method.
In our analysis of the original Engle-Granger bivariate and extended trivariate regressions, we primarily reject the null hypothesis of non-stationarity in the error-correction terms derived from empirical data at a conventional 5% significance level The optimal lag length for the ADF tests on the error correction term (ECT) is determined to be 1, which minimizes both AIC and SC criteria Our findings on the cointegration of spot exchange rates and relative price differences, as well as domestic and foreign prices under the Engle-Granger framework, confirm a long-run equilibrium among the variables Although our short-run data does not confirm the non-stationarity of real exchange rate (RER) levels, it reveals that the endogenous variables in the system are interrelated and interact over time, leading to a stationary behavior in the long run This supports a relatively weak version of the Purchasing Power Parity (PPP) theorem Additionally, this two-stage cointegration test may help address the low power issues associated with ADF and PP unit root test frameworks.
The challenge of low power in testing for non-stationarity is widely recognized Previous research by Engle and Granger (1987) in their two-stage regression demonstrated that non-stationary prices and relative price ratios tend to correct errors towards a future equilibrium In this study, we apply a modified Augmented Dickey-Fuller (ADF) test to provide empirical values that indicate this tendency.
The majormotivation forthismodied paradigm is desribed inthe following In our originaltheory,strong-form PPP implies(3.1) An augmentedDikey-Fullerregression
In the testing framework, as wellas shown by theory, ặt should not be present in the
(ADF test statistiomparableto Phillips-Ouliarissimulatedritial values)
N.E.R Equation ADF-Stat k Min.AIC(SC)
Note: Standarderror(s.e.) isreportedinparentheses;t-Stat insquarebrakets;
? denotesthe nullhypothesis of nonstationarityrejeted at 1%level, ompara- ble to Phillips-Ouliaris.v., for m=3 Tests are onduted for period 1986:01-
2002:12 u^ t denotes error-orretiontermfrom therst-stage E-G regression
Table 6: E-G seond-stage trivariateregression usingtheADF paradigm analysisto better reettheerror-orretion mehanism Clearly,thenewonstrution reads intheeq (?? ) below: s t
? t+i +u t where s t is spot exhange rate (in logarithms);R ER real exhangerate; and dpp
? rel- ative prieratio Leadtermin relative prieratiohas beenadded aswellasits lagged values up to order k Our empirial results are in the following table, examining the modelfor i=1 inalllead and lagterms desribedintheeq.(??). s t
Ljung-Boxstatistiforserialorrelationof upto 6 and 12 lags,respetively.
Table 7: AmodiedADF analysis fortheexhange rates
The resultsintable (7)unveila patternof theexhange rate dynamisthat all i s are signiantat1%levelandnegative Theregressionissatisfatorywithresidualsexhibit white-noiseproperty; no serialorrelation afterwards In termsof absolute magnitude, theyappeartohavebeenloseto oneanother Absolutevaluesofinallases arewell belowthe unity and
0 bearopposite signs,that is,negative in andpositive with
(orretlysigned) Foralltestedurrenies,theirmagnitudes,however,do notseem
The new dynamis for all exhange rates indiatethat oeặient of rst-order lagged
The regression analysis reveals that the exchange rate (ER) is statistically significant and has a negative coefficient, indicating its importance in confirming the earlier E-G analysis This negative relationship suggests that large deviations from Purchasing Power Parity (PPP) in the previous period will contribute to reducing divergence in the subsequent period This condition is essential for the exchange rate to eventually stabilize towards equilibrium in the future.
5.2.3 Co-integration test under Johansen's VAR paradigm:
We now report some empirial from max eigenvalue approah of VAR(2) analysis for searhingforo-integratedequationsamongstvariablesversusstationarityharater of thedataseriesusedforthesystem Werepeatthepreviousnotethattheentirereform periodof Vietnamanbeseenasonsistingof twomajorsub-periods,whiharebefore
1989:06andafter Themainreasonsofthisisduetotheliberalizingofthetradeaount sine beginning of 1989, where international trading started beoming a more liberal eldof ativitiesand privateforesouldjoin themarket eặiently Therefore,wean testforPPPseparatelybetweenthetwosub-samplesforageneralunderstanding This isequivalent to testing fora dummyofstrutural break.
The analysis presented in Table 8 supports the hypothesis that the trivariate systems are cointegrated with reduced rank Specifically, for three out of four exchange rates tested, we accept the hypothesis of a single cointegrating relationship among the three variables at the 5% significance level In the case of the EUR, the cointegration equation holds true at both the 5% and 1% significance levels Additionally, Table 8 provides the normalized cointegrating vector and the adjustment coefficient matrix.
The integration equations indicate that the considered variables tend to combine effectively, forming a stationary linear combination that supports the mean-reverting nature of real exchange rates Additionally, our impulse response analysis of spot exchange rates, based on a one standard deviation shock over a future period of 20 months (utilizing Monte Carlo simulation in EViews with 10,000 replications), reveals that the effects of innovations in spot exchange rates on future real exchange rates diminish over time and do so relatively quickly.
Note: *(**) denotes null hypothesisrejeted at1(5)% level Co-integrating o- eặient (beta)and adjustment oeặient (alpha)vetors are presentedfor one o-integration equation Standard errorsreportedinparentheses ()denotes rst-orderdiereneofdatainlogarithms VARmodelsseletedforlaglengthof
Table 8: Resultsfrom Johansentest foro-integrating equations
Co-integrating endogenous variables of the Real Exchange Rate (RER) involves using a Vector Error Correction Model (VECM), which can be represented through vector autoregressive equations or a system of autoregressive distributed lags This model incorporates lagged variables to analyze the relationships among them We present a representation of vector error correction regression utilizing Johansen's multivariate analysis, focusing on the first estimation applicable to flexible exchange rates Here, changes in spot exchange rates are placed on the left-hand side (LHS), with the Error Correction Term (ECT) followed by lagged changes in the values of endogenous variables The lag structure is determined based on the Akaike Information Criterion (AIC) for model selection, resulting in three lags for all data series.
Each equation within a VECM dynamic system represents an Error Correction Model (ECM), which incorporates restrictions derived from the decomposition Although these restrictions are not unique, they play a crucial role in shaping the structure of the equations.
, where represents a (r1) vetor of adjustment oeặients, and (r1) vetor of oeặient building the long-run equi- libriumbetweenthe endogenousvariablesofthe sytem.
The error correction model (ECM) for the first equation in the vector error correction model (VECM) incorporates an error correction term (ECT) that plays a crucial role in analyzing currency-based exchange rates We will examine the coefficient associated with the ECT, noting that a negative value indicates the presence of an error correction mechanism This mechanism suggests that any past shocks to the system will diminish over time, allowing the exchange rate to gradually return to its mean value This behavior is essential for maintaining the long-run equilibrium condition of real exchange rates, which is the focus of our current study.
Co-integration test under Johansen's V AR paradigm
We now report some empirial from max eigenvalue approah of VAR(2) analysis for searhingforo-integratedequationsamongstvariablesversusstationarityharater of thedataseriesusedforthesystem Werepeatthepreviousnotethattheentirereform periodof Vietnamanbeseenasonsistingof twomajorsub-periods,whiharebefore
1989:06andafter Themainreasonsofthisisduetotheliberalizingofthetradeaount sine beginning of 1989, where international trading started beoming a more liberal eldof ativitiesand privateforesouldjoin themarket eặiently Therefore,wean testforPPPseparatelybetweenthetwosub-samplesforageneralunderstanding This isequivalent to testing fora dummyofstrutural break.
The analysis presented in Table 8 confirms the hypothesis that the trivariate systems are cointegrated with reduced rank Specifically, for three out of four exchange rates tested at the 5% significance level, we accept the hypothesis of a single cointegrating relationship among the three variables In the case of the EUR, the cointegration equation holds true at both the 5% and 1% levels Additionally, the normalized cointegrating vector and adjustment coefficient matrix are included in Table 8.
The integration equations indicate that the considered variables tend to combine well with leads in value, resulting in a stationary linear combination that supports the mean-reverting nature of real exchange rates Additionally, our impulse response analysis of spot exchange rates, conducted over a future period of 20 months through Monte Carlo simulation in EViews with 10,000 replications, reveals that the effects of spot exchange rate innovations on future real exchange rates diminish over time and do so relatively quickly.
Note: *(**) denotes null hypothesisrejeted at1(5)% level Co-integrating o- eặient (beta)and adjustment oeặient (alpha)vetors are presentedfor one o-integration equation Standard errorsreportedinparentheses ()denotes rst-orderdiereneofdatainlogarithms VARmodelsseletedforlaglengthof
Table 8: Resultsfrom Johansentest foro-integrating equations
VECM representation
Co-integrating endogenous variables of the Real Exchange Rate (RER) utilizes the Vector Error Correction Model (VECM), which can be represented through a system of autoregressive distributed lags This model incorporates lagged variables to analyze relationships effectively We present the results of Johansen's multivariate analysis, focusing on the first estimation relevant for flexible exchange rates In this context, changes in spot exchange rates are analyzed on the left-hand side, while the Error Correction Term (ECT) is followed by lagged changes in the values of endogenous variables The lag structure is determined using the Akaike Information Criterion (AIC), identifying three lags for all data series.
Each equation in a VECM dynamic system represents an Error Correction Model (ECM), which incorporates restrictions derived from the decomposition process While these restrictions are not unique, they play a crucial role in shaping the structure of the equations involved.
, where represents a (r1) vetor of adjustment oeặients, and (r1) vetor of oeặient building the long-run equi- libriumbetweenthe endogenousvariablesofthe sytem.
The error correction model (ECM) for the first equation in the vector error correction model (VECM) will include an error correction term (ECT) for a currency-based exchange rate We will analyze the sign of the coefficient associated with the ECT; a negative sign indicates that the ECM functions as an error correction mechanism, allowing past shocks to the system to diminish over time and gradually return the series to its mean value This behavior is essential for maintaining the long-run equilibrium condition of real exchange rates, which is the focus of our current study.
The study explores the dynamics of Purchasing Power Parity (PPP) deviations and the associated error-correction mechanism It highlights that a negative adjustment coefficient indicates the presence of a first-order autoregressive feedback, which mitigates past deviations from PPP theory This short-run error-correction dynamic accounts for the mean-reverting tendency of the long-term relationship toward future equilibrium Additionally, the appendix provides matrix equations that detail the findings from both Vector Error Correction Model (VECM) dynamics and long-run relationships based on a single integrating equation hypothesis, which has been confirmed in all test cases.
Restrited oeặients long-run relations
We nowreportfurtherstatistis thatimposestandard restritionforECTof thelong- run relationship It is known that the oeặient vetor of ECT in the PPP theorem mustsatisfy:
Therefore, the restrition for absolute PPP (strong-form) beomes:
Below is an outome table (9) analyzing test statistis for these restrited oeặients underJohansenproedure.
Whatwendfromtheabovetable(9)resultsisquitemotivating Giventherestrition oftheointegratingvetor,forallases,weanstillrejetthenullofnoointegration.
=(11 1) This is equivalent to oursupport to the holding of strit PPP in the long runthrough the restritedointegratingrelationbeausetheLRteststatistisdonotrejet thebinding restritionsat anyonventionallevels,assumingthenumberofointegratingrelationis one.
Half-lives and onvergene to long-run equilibria
Computed HLs
In thetable (10) below, we summarizesome values of half-lives(HL) fordierent real exhangeratesunderawidelyusedDFregression TheDFframeworkdoesnotsuggest atimetrend fortheregression,however,dueto apossibletimetrendontrol(Balassa-
FX Rank Lag Restrited LRteststat p-Val d.f.
0 ointegrating relationships st=+ặt+ist
The half-lives derived from D-F regressions indicate the time required for the value to decay to 50% of its original magnitude It is evident that the half-life (HL) is influenced by the point estimates of the variable i, as defined in equation (20) below.
(20) whereby,shoks oftheregression deaymonotonially(ashasbeenveriedinprevious examinationof autoorrelationfuntion).
The test results indicate that the half-life of exchange rates varies between 0.85 and 1.67 years across all three cases summarized in the table Specifically, in the case of the DF regression without first-order differencing, the shortest half-life is observed at 1.45 years for the Euro (EUR), while the longest half-life is noted for other currencies.
1.67years(GBP).Thisspeedofmeanreversionisquitefastomparedtothoseobtained by previous studies, where HL is typially ranging from 3 to 5 years Our estimated half-liveslearlyindiatethattheirmeanvaluesarenotonlynite,butalso fairlysmall inmagnitude,omparedtomanypreviouslyreportedresults Forexample,Murrayand
Papell(2002) reportedH-L ofexhange ratesdenominatedinUSDollar inarangelike
0.89 (FFR), to 11.2 years (JPY), after bootstrapping an annualdata sampleof period
Bootstrapping HLs
Ouranalysis above is basedon asample periodof 1986:01-2002:12 Formanyprevious studieson long-run equilibriumof real exhange rates, data samplesusually spanover deades, oreven enturies,inwhihresearhers ndevidene supportingPPP validity.
In thisstudy, thedata sample is limiteddue largelyto a shortperiodof reform (sine
1986) It willthen be worthwhilethatwe ondut a repliationofmanymore samples basedontheatualoneto observeanysignianthangesinourpreviousnoteof nite
The repliation is done witha bootstrapping algorithm exeuted on the Mathematia environment(V4.1)
The article discusses the use of bootstrap algorithms in the Mathematia symbolic package, specifically the bootstrappakage.m developed by Enis Siniksaran from Istanbul It highlights the distribution of stochastic variables and reports various percentile values of the simulated mean Additionally, it references the work of Murray and Papell (2001), emphasizing that bootstrap methods for estimating bounds on lagged variable coefficients can yield asymptotically valid results.
Table 11: Bootstrapped half-livesfor4 RER timeseries
Intheresult table (11)above,we notethatformostases, bootstrapped meanof half- livesisdraggedon TheexeptionisforGBP 750-repliationsamplewhereH-Lislittle smallerthantheobservedvalueinD-F regression Thehange, althoughsmall,inom- putedHLisbeauseabsolutevaluesofmostautoregressiveparametersgetabitsmaller
In 2001, the University introduced running well on version 4.1, allowing data to be loaded directly from the Mathematica front-end as time series for regression analysis The tool utilizes bootstrapping and addresses both pair regression and residual series For optimal evaluation of different data sets, users are advised to terminate the current Mathematica kernel session after completing all runs on a specific data set Notably, an increase in H-L following treatment does not significantly alter the overall results.
The frequency distributions of the autoregressive parameters, presented in the Appendix, reveal that they are not normally distributed but rather skewed to the right, indicating a concentration of values Although the speed of correction towards Purchasing Power Parity (PPP) is relatively slow, it demonstrates a faster pace of adjustment when compared to results from other regions around the world.
This paper examines the significance of exchange rates in transitioning economies, specifically focusing on Vietnam's shift towards a market economy Our analysis indicates evidence of long-run equilibrium through a single cointegrating vector among endogenous variables that influence real exchange rates This finding supports the weaker form of relative Purchasing Power Parity (PPP), suggesting that the Error Correction Term (ECT) can be linearly combined into a stationary process, which helps to reduce deviations from PPP in the long run However, we are unable to reject the restricted cointegrating vector for a strict form of PPP.
The study examines the convergence speed of currencies to Purchasing Power Parity (PPP) and finds a relatively rapid adjustment compared to previous research It reveals that, based on both observed and bootstrapped Hurst-Loeve (H-L) statistics, the time for deviations to correct is typically only two-thirds of the durations reported by other researchers for developed countries This quicker adjustment may be attributed to the post-hyperinflation period, during which the adjustment process is expected to accelerate.
The analysis indicates that during the early stages of the reform, there was a significant deviation in currency values, particularly an overvaluation of the Vietnamese Dong (VND) against foreign currencies This trend is not unusual, as the VND is permitted to fluctuate against other currencies; however, such adjustments have typically been cautious and require a considerable amount of time to take effect.
Nonetheless,its orretionover timedidhappenand beamenegligibledeviationsvery reently Beyond the empirial disussionwe providehereunder a few pointsas addi- tionalonludingremarksof thisstudy.
On the long-run equilibrium relationship and Law of One Prie (LOP) 55
The long-run equilibrium
The evidence presented in the previous section supports the reduced rank hypothesis We will examine the case of a single cointegrating equation that applies to the four dynamical systems governing our exchange rate behaviors These cointegrating equations illustrate long-run equilibrium relationships among the three variables included in the systems, providing empirical descriptions of their interactions.
In these long-run equilibrium relations, the dynamis assume that there exist error- orretionmodelssuhthatdeviationsfromthePPPwillbeadjustedbytheoeặients establishedinthe(31) matrix Westressthattheoeặients
1 inallourempiris arry a negative sign, apparently showing a redution of deviations over time These redutionshelpstabilizingthesystemsinthelongrun,henesupportingtheequilibrium relations.
On the Law of One Prie
The postulated Law of One Price (LOP) highlights the significant economic reforms in Vietnam that have transitioned from a Soviet-style centrally planned model to a more open economy Since these reforms began, Vietnam has faced unprecedented challenges such as hyperinflation and currency depreciation The establishment of economic equilibrium is crucial, as it reflects the positive impact of open-door policies that have facilitated international trade and economic transactions This foundational change is essential for future quantum economic restructuring Theoretically, when transportation costs are excluded, identical goods in different countries should have the same price when expressed in a common currency.
The arbitragethat helps bridgethe diereneis usuallydone via internationaltrades.
Hene,we would state that theoutome of ourestablishment herewouldnot be likely withoutsubstantialeonomihanges ouringwithintheVietnam'seonomyoverthe reentreform period.
Poliy onern: assessing reent exhange rate misalignments
Inlinewith ouranalysisof thelong-run equilibrium,wherewe establishthat itisvery likelythatreal exhangeratedynamiswillbindstohastivariablestoastationary re- lationshipwithnitespeedofonvergene,itispossibletoexplorebehavioralproessof
17yearsoftheeonomy,severalimportanteventsmayrepresentshoksthatoulddrive realexhangeratesfromdeterministitheoretiequilibriumlevels Eonomistsandpol- iymakers shouldalso bewaryofsigniantdeparture(inmagnitude)fromonsidered equilibrium.
This article briefly explores three commonly used benchmarks for detecting deviations from hypothesized equilibrium real exchange rates (ERER) The first method involves regressing real exchange rate (RER) data against a constant, representing a fixed ERER The second method also uses a constant but incorporates a time trend component, resulting in a linear time trend ERER Finally, the third approach is discussed, providing a comprehensive analysis of these benchmarks.
Our comprehensive analysis of Real Effective Exchange Rate (RER) misalignment is detailed in Appendix Table (13) of the study, highlighting several key observations Notably, the same original RER data can yield different interpretations across three models While one model may indicate an undervaluation of the USD against the VND, another could suggest an overvaluation We determine consensus in two ways: if any two of the three models indicate overvaluation, we proceed with caution in interpreting the results.
If all three agree on over- versus undervaluation, we aept the onlusion R ER
USD shows that USD was undervalued against VND signiantly in early years of reform.
Theaverage undervaluationofUSDisabout60%in1987;20%1988 However,in1989, the situation hanged and USD beame more or less overvalued by 15%; over 20% in
1990;15%1991;