Final Exchange rate pass-through to inflation in Vietnam_on Journal.pdf

16
1 EXCHANGE RATE PASS-THROUGH INTO INFLATION IN VIETNAM: AN ASSESSMENT USING VECTOR AUTOREGRESSION APPROACH NGUYEN Dinh Minh Anh * , TRAN Mai Anh ** and VO Tri Thanh *** Published on Vietnam Economic Management Review, 2010 Updated by NGUYEN Dinh Minh Anh 1/2011 Abstract This paper has estimated the pass-through of exchange rate into inflation in Vietnam during M1:2005 - M3:2009 using the Vector Auto-regression (VAR) model. The result shows that the pass-through coefficient is 0.07 after a period of 2 months since the initial shock to exchange rate. This impact is completely removed in the third month. In comparison with such coefficients of some other developing countries, the exchange rate pass-through into inflation in Vietnam is at moderate-sized level. The paper also finds that high inflation in Vietnam in recent years is mainly due to the expansion of money supply. To control inflation, therefore, the State Bank of Vietnam (SBV), first and foremost, needs to manage money supply. Moreover, as the money supply is effectively controlled, an exchange rate arrangement following market determinants will not cause inflation. Also, VND interest rate is one powerful tool to control the inflation. Key words: Exchange rate, Pass-through, Inflation, VAR. JEL Classification Numbers: F31, C32, E52 1. INTRODUCTION Exchange rate is a crucial economic variable to the open economies. The exchange rate can affect the economy through different channels such as trade, prices, and budget. One of its most important impacts is on inflation, which is broadly termed as the exchange rate pass-through (ERPT) into inflation. The higher the pass-through coefficient is the more effective is the exchange rate as a tool for controlling inflation. The exchange rate pass-through effects in various economies can be different. For instance, during Asian 1997-crisis, a devaluation of Won had only marginal impact on the inflation rate in Korea, whereas a devaluation of Rupiah led to a considerably high inflation rate in Indonesia. Since Doimoi (Renovation) the success of Vietnam in attaining rather high economic growth has largely been attributed to macroeconomic stability. In recent years, however, the economy has been facing with pressures of rising inflation, increasing trade imbalance, dollarization and capital inflow fluctuation. In this context, it is very essential to understand the degree and timing of exchange rate pass-through, especially as an inflation targeting policy is adopted. There have been, in fact, some studies of the relationship between exchange rate movement and inflation in Vietnam. Hang (2010) finds that the exchange rate policy could not be against * NGUYEN Dinh Minh Anh, University of Economics and Business (UEB), Vietnam National University (VNU) ** TRAN Mai Anh, University of Economics and Business (UEB), Vietnam National University (VNU) *** VO Tri Thanh, Vice President of Central Institute for Economic Management (CIEM).

Transcript of Final Exchange rate pass-through to inflation in Vietnam_on Journal.pdf

Page 1: Final Exchange rate pass-through to inflation in Vietnam_on Journal.pdf

1

EXCHANGE RATE PASS-THROUGH INTO INFLATION IN VIETNAM:

AN ASSESSMENT USING VECTOR AUTOREGRESSION APPROACH

NGUYEN Dinh Minh Anh*, TRAN Mai Anh

** and VO Tri Thanh

***

Published on Vietnam Economic Management Review, 2010

Updated by NGUYEN Dinh Minh Anh 1/2011

Abstract

This paper has estimated the pass-through of exchange rate into inflation in Vietnam during

M1:2005 - M3:2009 using the Vector Auto-regression (VAR) model. The result shows that the

pass-through coefficient is 0.07 after a period of 2 months since the initial shock to exchange

rate. This impact is completely removed in the third month. In comparison with such coefficients

of some other developing countries, the exchange rate pass-through into inflation in Vietnam is

at moderate-sized level. The paper also finds that high inflation in Vietnam in recent years is

mainly due to the expansion of money supply. To control inflation, therefore, the State Bank of

Vietnam (SBV), first and foremost, needs to manage money supply. Moreover, as the money

supply is effectively controlled, an exchange rate arrangement following market determinants

will not cause inflation. Also, VND interest rate is one powerful tool to control the inflation.

Key words: Exchange rate, Pass-through, Inflation, VAR.

JEL Classification Numbers: F31, C32, E52

1. INTRODUCTION

Exchange rate is a crucial economic variable to the open economies. The exchange rate can

affect the economy through different channels such as trade, prices, and budget. One of its most

important impacts is on inflation, which is broadly termed as the exchange rate pass-through

(ERPT) into inflation. The higher the pass-through coefficient is the more effective is the

exchange rate as a tool for controlling inflation.

The exchange rate pass-through effects in various economies can be different. For instance,

during Asian 1997-crisis, a devaluation of Won had only marginal impact on the inflation rate in

Korea, whereas a devaluation of Rupiah led to a considerably high inflation rate in Indonesia.

Since Doimoi (Renovation) the success of Vietnam in attaining rather high economic growth has

largely been attributed to macroeconomic stability. In recent years, however, the economy has

been facing with pressures of rising inflation, increasing trade imbalance, dollarization and

capital inflow fluctuation. In this context, it is very essential to understand the degree and timing

of exchange rate pass-through, especially as an inflation targeting policy is adopted.

There have been, in fact, some studies of the relationship between exchange rate movement and

inflation in Vietnam. Hang (2010) finds that the exchange rate policy could not be against

* NGUYEN Dinh Minh Anh, University of Economics and Business (UEB), Vietnam National University (VNU)

** TRAN Mai Anh, University of Economics and Business (UEB), Vietnam National University (VNU)

*** VO Tri Thanh, Vice President of Central Institute for Economic Management (CIEM).

Page 2: Final Exchange rate pass-through to inflation in Vietnam_on Journal.pdf

2

inflation unless the money supply and credit growth rate were managed. However, this study

does yet estimate the magnitude and timing of the changes in exchange rate into inflation. Using

Vector Auto-regression (VAR) approach for evaluating the impact of one-time exchange rate

shock to inflation, Minh (2009) shows that, the exchange rate pass-through in Vietnam is at the

medium level as compared to other economies. It does not, however, provide with logical

explanations about the order of variables in Cholesky decomposition. Moreover, there is a doubt

about the study’s conclusion that the changes in aggregate demand do not affect inflation. The

objective of this paper is also to study the exchange rate pass-through issue in Vietnam,

attempting to overcome some drawbacks in previous studies.

The remainder of the paper is organized as follows. Section 2 reviews theoretical background

and econometric techniques for estimating the ERPT. Section 3 describes the model,

methodology and relevant data used for quantitative assessment in the case of Vietnam. It then

presents the estimation of the exchange rate shocks to the domestic prices along the distribution

chain. Finally, Section 4 concludes with a summary of findings, policy recommendations and

suggestions for further research.

2. THEORETICAL BACKGROUND

The definition of the exchange rate pass-through into prices can be somehow understood

differently. Olivei (2002) regards the ERPT as the response of import price in percentage when

the nominal exchange rate changes by 1%. Some other the studies such as Lian (2006) and

Nkunde Mwase (2006) use a broad definition of ERPT, which reflects the changes of the

domestic prices1 in response to 1% - exchange rate shock. This paper follows the later definition.

It is mostly about the exchange rate pass-through into inflation (ERPTIF) meaning the change in

percentage of the consumer price in response to 1% - change of the exchange rate2. Similarly, it

can also be the exchange rate pass-through into the import price (ERPTIP) or the exchange rate

pass-through into the production price (ERPTPP).

According to Nicoleta (2007), the changes in exchange rate can influence the inflation rate

through two channels: direct and indirect ones. The direct channel can be seen through the

exchange rate shock as a devaluation of local currency. This makes the imported consumer

goods and raw materials become more expensive. The later leads to higher production costs and

as a result, higher consumer prices (Figure 2.1).

Figure 2.1: The direct channel of exchange rate

1 The domestic price means the import price, the production price and the consumer price.

2 Sometime in short it also refers just as the exchange rate pass-through (ERPT)

Page 3: Final Exchange rate pass-through to inflation in Vietnam_on Journal.pdf

3

Source: Nicoleta (2007).

The indirect channel supposes that a depreciation of domestic currency makes domestic goods

cheaper and hence, demand for this country’s exports increases. This will trigger an increase in

labor demand, wages and aggregate demand, and as a result, could lead to inflation. This effect,

however, can only happen in the long-run due to rigidity of price in the short term. But the

dollarization phenomenon may magnify the indirect effect. As domestic currency devalues, the

prices of assets (like real estate or luxury items) counted in foreign currencies increase, and this

causes the increases in consumer prices (through income-generating asset effect).

There are many different important factors – both macro and micro – determining the pass-

through of exchange rate (Box 2.1). Following the classification by An (2006), the micro factors

are: 1) pricing-to-market and mark-up adjustments; 2) market segmentation features such as

transportation and distribution costs, non-tariff barriers and the role of multinational

corporations; 3) the degree of returns to scale; and 4) the elasticity of demand for imported

goods. Macro factors include: 1) the level of inflation and the perceived persistence of exchange

rate swings; 2) the monetary policy environment; and 3) the size and openness of the economy.

Box 2.1: Factors affects the pass-through of exchange rate into inflation

Micro factors

1) Krugman (1987) analyses the pricing-to-market phenomenon, according to which foreign suppliers,

wishing to keep constant market shares, accept smaller profit margins when the importing country's

currency depreciates. Pricing-to-market thus implies a lower pass-through.

2) Burstein et al (2001) find that the local distribution costs (such as wholesaling and retailing) represent

up to 40% of the final retail price of any commodity. As these costs are less dependent on exchange rate

developments, they may consequently lower the pass-through even for internationally tradable goods.

3) Yang (1997) and Olivei (2002) study the degree of returns to scale, concluding that the rate of

exchange rate pass-through is inversely related to the elasticity of the marginal cost with respect to output.

If the marginal costs decrease with output, higher demand stimulated by price decreases resulting from

exchange rate appreciation should lead to further cost and thus price reductions, implying a higher rate of

pass-through.

4) Foreign suppliers are likely to adjust their prices according to the perceived demand elasticity in the

import country. The higher the elasticity of demand to price changes, the less likely are firms to pass

through the whole exchange rate shock (Yang 1997).

Macro factors

1) Taylor (2000) argues that the inflationary environment and the perceived persistence of shocks are

decisive determinants of pass-through rates. More precisely, firms are less likely to adjust their prices if

the exchange rate changes or inflation are expected to be volatile and temporary (a point also stressed by

Mann (1986) and empirically supported by McCarthy (2000)).

2) The connection between inflation and pass-through levels implies that monetary policies should also

affect the transmission of exchange rate movements to domestic prices. Gagnon and Ihrig (2004) find that

countries with credible and anti-inflationary monetary policies generally exhibit lower pass-through

levels.

3) Country openness, proxied by the import share in total production, also affects pass-through rates.

Intuitively, the more open the country is to international trade, the greater the exchange rate pass-through

to consumer prices should be. Moreover, according to McCarthy (2000), a small country should

Page 4: Final Exchange rate pass-through to inflation in Vietnam_on Journal.pdf

4

experience higher pass-through levels than a large country. This is because the fall in demand in a large

country in reaction to domestic price increases resulting from exchange rate depreciation reduces world

demand and hence depresses world prices.

Source: Heidi Cigan et al (2008).

In measuring ERPTIF, two techniques have been commonly used in a number of studies. The

first technique known as the standard single-equation regression technique is used in the studies

by Olivei (2002), Campa and Goldberg (2005), Campa, Goldberg and González-Mínguez

(2005), and Otani, Shiratsuka and Shirota (2005). They apply the OLS to evaluate the pass-

through, with polynomial distributed lags to capture the dynamic response of traded goods prices

to exchange rate changes. But this method has a disadvantage paying no attention to the time

series properties of the data, as most macroeconomic series and asset prices such as exchange

rates, economics growth or inflation are non-stationary. Therefore, the assumptions of the OLS

estimation are violated, leading to the problems of spurious regression. Moreover, the estimation

could suffer from inconsistency problems due to the endogenous determination of exchange rates

and prices.

The second technique is named as VAR. McCarthy (2000) is among the first researches

employed the VAR framework to estimate the ERPT. The VAR models have several advantages

compared to the single-equation-based methods. First, they could solve endogeneity problem

inherent in the single-equation-based methods. Moreover, the estimated impulse response

functions trace the effects of a shock to one endogenous variable on other variables through the

structure of VAR, which allows us to assess not only pass-through within a specific period, but

also its dynamics through time. VAR approach, therefore, is an effective measure of the degree

and timing of pass-through parameters. Some typical studies using VAR models are of Hahn

(2003) and Faruqee (2006) for the cases of developed countries, especially in European, Ito and

Sato (2006) for East Asian countries, Belaisch (2003) for Brazil, and Leigh and Rossi (2002) for

Turkey. In this paper, we also use a VAR model to estimate the ERPTIF in Vietnam. The

empirical evidence is shown in Section 3 after a brief examination of the relationship between

exchange rate movement and inflation in Vietnam since 1990s.

3. EMPIRICAL EVIDENCE

Figure 3.1 shows the changes in exchange rate and inflation in Vietnam during 1992-2009.

Figure 3.1: Nominal exchange rate and inflation in Vietnam (1992-2009)

Source: Authors’ own calculation from IFS and GSO.

-10

0

10

20

30

40

0

5000

10000

15000

20000

19

92

19

93

19

94

19

95

19

96

19

97

19

98

19

99

20

00

20

01

20

02

20

03

20

04

20

05

20

06

20

07

20

08

20

09

Exchange rate (VND/USD) Inflation rate (%)

Page 5: Final Exchange rate pass-through to inflation in Vietnam_on Journal.pdf

5

In the early 1990s when the inflation rate was considerably high (up to 35% in 1992), Vietnam

had efforts to control inflation using different policies, including adopting a relatively rigid

exchange rate regime. This option has theoretical foundation that keeping exchange rate stable

can improve trust in domestic currency, forcing the government to control budget deficit and

credit growth and thus, reduce inflation and stabilize the macroeconomic situation. In fact,

inflation during 1992 - 1996 was managed quite effectively. However, it is not the case in the

later periods. During 1997-2003 when Vietnam experienced the Asian crisis and the recovery

after, Vietnamese Dong (VND) was devalued continuously against USD. Yet, high inflation did

not occur, even the economy fell into the period of deflation in 2000-2001. Also, the application

of a relatively rigid exchange rate policy from 2004-2008 did not help the economy control

inflation. Inflation rate increased over the years and then skyrocketed to about 23% in 2008.

A look at the exchange rate and inflation in Vietnam indicates that their relationship is

complicated and the impact of the changes in exchange rate on inflation should be determined by

many other macro-factors. It is worth, therefore, having an empirical evidence of the magnitude

as well as timing of the ERPT in recent years.

3.1. Empirical framework

Based on arguments in Section 2, we set up a VAR model including 7 endogenous variables

(OPGAP;DLOG(CPI);DLOG(VNDM);DLOG(FCD);DLOG(RVND);DLOG(RUSD);DLOG(ER)

) and one exogenous (DLOG(OIL)); where OPGAP is output gap, CPI is consumer price index,

VNDM and FCD are monetary aggregates, RVND and RUSD are deposit interest rates, ER is

exchange rate, and OIL is oil price. All variables are of monthly time series form M1:2005 until

M3:2009 and described in the Table 3.1. They are, excluding OPGAP, RUSD, RVND, adjusted

seasonally. LOG is the natural logarithm and D represents the first difference operator. Table 3.1: The variables used for empirical estimation

- OPGAP: By definition, the output gap is the difference between real output and potential output. An

excess of real output over potential output implies that the economy is growing over its long-run capacity

or, in other words, over the full employment capacity. Therefore the output gap reflects the excess

demand in the economy (which could be positive or negative). The GDP is only available in quarter, so

industrial output, which can be extracted directly from GSO, is used as a proxy for output. Then the

OPGAP is calculated monthly using Hodrick-Prescott method.

- CPI: CPI is a monthly consumer price index (Base year 2005 = 100) taken from IFS database.

- M2: Broad money M2 includes narrow money M1, quasi-money, and bond and money market

instruments taken from IFS database. Data on foreign currency deposits is a component of quasi-money

and also extracted from IFS. Broad money is divided into VND monetary aggregate (VNDM) and foreign

currency deposits (FCD). We separate the broad money into two because the movements of FCD and

VNDM are so different, which is a result of significant degree of dollarization in Vietnam. Because the

data on the amount of foreign currencies circulating as cash outside the banking system is not available,

the amount of foreign currency deposits is used as a proxy for dollarization (Luc 2008).

- RVND and RUSD: The 3-month VND deposit rate (RVND) is used as a proxy for interest rate and

sourced from IFS database. The 3-month USD deposit rate (RUSD) is taken from BIDV, which

represents for the domestic US dollar rate. The study also divide interest rates into VND deposit rate and

USD deposit rate because the movements of these two variables are different and both of them play a

crucial role in the monetary policy decision-making (Luc 2008)

- ER: ER is defined as the exchange rate of VND per 1 USD and extracted from IFS database.

Page 6: Final Exchange rate pass-through to inflation in Vietnam_on Journal.pdf

6

- OIL: Oil price (USD/barrel) is extracted from the IFS, where the UK Brent oil price index (2005 =100)

is found.

As the purpose of this study is to estimate the impact of exchange rate and other macroeconomic

shocks on domestic prices and also other possible interactions among them, we generate the

structural shocks using a Cholesky decomposition of the matrix Ω, a variance-covariance matrix

of the reduced-form VAR residuals. Building such a matrix, however, requires logically

arrangement of the order of the variables in Cholesky decomposition. This order will identify

structural shocks and thus have certain impacts to research results. The different orders would

create different results.

We follow closely the arguments suggested by Bernanke and Mihov (1998), in which the non-

policy variables (OPGAP and CPI) are arranged first and then followed by the variables related

to policy (money supply, interest rates and exchange rate). This seems to be reasonable and

consistent with actual behavior of Vietnam economy because firms do not change their output as

well as prices immediately in response to the changes in monetary policies within the same

period due to adjustment costs, while the SBV could set its policies quickly in response to the

shocks to output and prices. In other words, the output and prices stand above the policies in the

Cholesky decomposition. Moreover, the CPI is assumed to response to output immediately

because the output is often more rigid than the price and the changes in output lead to the

changes in prices.

After the changes in output and prices, monetary aggregates are assumed to change

contemporaneously. When it comes to the order between these two variables, the VNDM will

respond first to the changes in output and prices because it is the main currency in Vietnam and

then FCD represented by foreign currency deposits will vary.

Due to the changes in output, prices and money demand, the SBV could determine the deposit

interest rate on VND and USD by different monetary instruments. Furthermore, the interest rate

on VND is assumed to arrange before that on USD because the role of VND to the economy is

more essential than that of USD.

According to the asset approach, the changes in interest rate causes the changes in exchange rate

in short-term, therefore the ER variable is put after the interest rates and responses immediately

with the changes in output, prices, money supply and interest rates. If the exchange rate varies,

the SBV will have interfering policies to stabilize the exchange rate in the next period because

Vietnam follows the pegged exchange rate regime.

From the above arguments, the relationship between the reduced-form VAR results (εt) and

structural disturbances (et) for the Vietnam economy can be proposed in Table 3.2.

Page 7: Final Exchange rate pass-through to inflation in Vietnam_on Journal.pdf

7

Table 3.2: Cholesky Decomposition for Vietnam economy

11

21 22

31 32 33

41 42 43 44

51 52 53 54 55

61 62 63 64 65 66

71 72 73 74 75 76 77

opgap opgap

cpi cpi

vnd vnd

usd usd

rvnd rvnd

rusd rusd

er er

b e

b b e

b b b e

b b b b e

b b b b b e

b b b b b b e

b b b b b b b e

Source: Authors’ own tabulation

3.2. Tests for model validity and interpretation of the estimation results

Before analyzing the estimation results through impulse response and variance decomposition,

we need to implement some necessary tests, including the unit root tests, lag length and stability

of the VAR model.

The ADF unit root test is used to determine the stationarity of time series. The stationarity is a

very important condition as the model could lead to spurious estimation if the below hypotheses

are not checked.

The null hypothesis Ho: the data series is not stationary.

As showed in Table 3.3, most initial data series, except that of OPGAP3, are non-stationary and

integrated of Order 1 I(1). As a result, they are replaced by the first difference of logarithm of the

variable, which reveals I(0) or stationary.

Table 3.3: Results from Unit Root Tests

Time series Levels First Differences

CPI 0.38

(0.98)

-3.88***

(0.0042)

FCD 0.2

(0.97)

-6.15***

(0.00)

VNDM -0.32

(0.9)

-5.5***

(0.00)

RVND -2.8*

(0.05)

-5.1***

(0.00)

RUSD -1.03

(0.73)

-7.9***

(0.00)

ER 2.4

(1.00)

-5.1***

(0.00)

OPGAP -8.8***

(0.00)

Note: *,* and *** denote significance at 10%, 5% and 1% respectively; P-values are in parentheses.

Source: Authors’ estimation

3 OPGAP is a stationary series because it is extracted from HP method.

Page 8: Final Exchange rate pass-through to inflation in Vietnam_on Journal.pdf

8

The lag length of variables plays also a crucial role in forming the VAR model. According to the

criteria below, suggested lag lengths are of 0, 2 and 3. The lag length of is chosen because of

some following reasons. Firstly, the lag length of 0 is not reasonable because the values in the

previous periods often have certain influences to subsequent periods and future expectations are

generally based on present and past values. Secondly, the smaller the lag length is, the better the

quality of results is because an increase in the lag length will make the degrees of freedom

reduce, thus affecting the quality of the estimates. Last but not least, the following diagnostic

tests demonstrate that the lag length of 2 is suitable (Table 3.4).

Table 3.4: Lag length criteria

Lag LogL LR FPE AIC SC HQ

0 212,1675 NA 5,14e-13 -8,432659 -7,881551* -8,225273*

1 272,1337 96,96663 3,33e-13 -8,899306 -6,419321 -7,966071

2 326,1330 71,23307* 3,17e-13* -9,112041 -4,703178 -7,452956

3 384,7112 59,82462 3,29e-13 -9,519627* -3,181887 -7,134693

Source: Authors’ estimation.

Moreover, using the inverse roots of the characteristic AR polynominal, the stability test shows

that the estimated VAR model is stable and no roots lie outside the unit circle (Appendix 1).

Thus, the standard errors in the impulse responses in the model are valid.

The diagnostic tests generally provide also satisfactory outcomes. The results show that the

errors of most components are normally distributed, except VNDM. This is because the null

hypothesis of normally distributed errors cannot be rejected at the significant level. The serial

correlation also arises at lag 1 and the variance of three errors has changes at the significant level

(Appendix 2). However, the main objective of the VAR model is to analyze the dynamic

relationships among variables rather than the estimation of the parameters in any particular

equation, so the presence of this phenomenon may not be the main concern in this study. In

addition, the stability of the VAR model proven above suggests that the standard errors of the

impulse responses, which are the keys in the VAR analysis, are significantly valid.

Now we can examine the estimation results. Two statistical properties of the VAR models used

to assess the pass-through of exchange rate shock to consumer prices are impulse response and

variance decomposition. Firstly, the impulse responses to the related variables are estimated over

a period of 40 months horizon. These responses are standardized to determine the response of the

other exogenous variables with a 1% - increase of the exchange rate or 1% - devaluation of

VND. In addition, other shocks are standardized to 1% - increase. Then, the variance

decomposition is applied to separate the variation in an endogenous variable into the component

shocks to the VAR. Thus, the variance decomposition provides information about the relative

importance of each random innovation in affecting the variables in the VAR, especially CPI.

Figure 3.2 illustrates the impact of one-time shock to exchange rate to inflation. Because the ER

is placed behind the CPI in the Cholesky triangular matrix, the impact of this shock to inflation

only happens in the next period. In the first period - the 1st month – prices do not change. As it is

Page 9: Final Exchange rate pass-through to inflation in Vietnam_on Journal.pdf

9

expected, prices increase after the exchange rate increases and the biggest impact falls into the

second month after the shock, followed a decrease in the subsequent months. After about 13

months, inflation back to its initial level before the shock. Moreover, the upper and lower dotted

lines represent two standard error bands are rather narrow, meaning the reliability of the result.

Figure 3.2: Impulse response of CPI to a shock to the exchange rate

Source: Authors’ estimation.

In order to calculate the pass-through coefficients, it is better to translate the shock into one

percent shock in exchange rate. In addition, the change of the exchange rate because of its shock

in the following periods should be also considered. Leigh and Rosi (2002) measure the pass-

through coefficients as follows.

,

,

t t i

t t i

t

PPT

E

Where Pt,t+i is the change in indices in period i in response to the initial shock in exchange rate,

Et is the accumulated impact change of exchange rates to their own shocks. Therefore, the pass-

through coefficients are resulted as indicated in Table 3.5.

Table 3.5: The exchange rate pass-through coefficients

Source: Authors’ calculation.

-.004

-.003

-.002

-.001

.000

.001

.002

.003

5 10 15 20 25 30 35 40

Period CPI Period CPI

1 0 9 -0,01

2 0,069567 10 -0,09

3 -0,32682 11 -0,12

4 -0,52774 12 -0,19

5 -0,46276 15 0

6 -0,18333 20 0

7 -0,03 40 0

8 -0,05

Page 10: Final Exchange rate pass-through to inflation in Vietnam_on Journal.pdf

10

-.003

-.002

-.001

.000

.001

.002

.003

.004

.005

.006

5 10 15 20 25 30 35 40

-.005

-.004

-.003

-.002

-.001

.000

.001

.002

.003

.004

5 10 15 20 25 30 35 40

-.002

-.001

.000

.001

.002

.003

.004

.005

5 10 15 20 25 30 35 40

-.004

-.003

-.002

-.001

.000

.001

.002

.003

.004

.005

5 10 15 20 25 30 35 40

Table 3.5 shows that 1% - increase in exchange rate causes CPI to rise 0.07% in the second

month after the initial shock. In other words, the ERPT coefficient in Vietnam is 0.07. Moreover,

this effect is completely eliminated in the third month and after 13 months the CPI converges to

its pre-shock level in the long-run.

The responses of CPI to other macro shocks can be seen in Figure 3.3. As it is expected, when

aggregate demand (OPGAP) increases, prices will rise immediately because aggregate demand is

ranked before CPI in VAR model (Figure 3.3a). This increase occurs continuously since the first

month to fourth month, and reaches a peak in the second month since the initial shock. The result

is opposite to that by Minh (2009) that concludes that the changes in OPGAP do not affect

inflation. This contradiction may be due to the differences in research period or the method of

setting up VAR model in term of the order of variables. However, aggregate demand has certain

influences into inflation in the case of Vietnam. For example, in the early of 2008, a fast increase

in aggregate demand was considered one of the factors causing inflation rise considerably.

Figure 3.3: Impulse responses of CPI to other shocks

(a) OPGAP (b) VNDM

(c) RVND (d) RUSD

Source: Authors’ estimation.

Page 11: Final Exchange rate pass-through to inflation in Vietnam_on Journal.pdf

11

Figure 3.3b illustrates the response of CPI to the increase in money supply. Prices increase

significantly after a rise in money supply and reach a peak in the third month. After 13 months,

the impacts of this shock on CPI are completely removed. Thus, the impacts of the changes in

money supply on inflation are rather high and persistent. This has certain policy implication. The

increase in VND interest rate has a considerably negative impact on inflation (Figure 3.3c) and

this is also consistent with the economic theories. The response of prices to the change in USD

interest rate is somewhat different, positive in second month before decreasing remarkably in the

next month. This might be caused by the increased firms’ costs in USD borrowings.

As argued by Taylor (2000), a high level of the ERPT implies a high transmission from

exchange rate change to prices, but if the change in exchange rate plays a small role in variance

of prices, the exchange rate will not be important in determining the movement of prices. It is

necessary, therefore, to analyze the variance decomposition of targeted variable, which is CPI in

this case.

Table 3.4: Variance Decomposition of CPI

Period S.E. OPGAP DLOG(CPI) DLOG(VNDM) DLOG(FCD) DLOG(RVND) DLOG(RUSD) DLOG(ER)

1 4863.302 1.050354 98.94965 0.000000 0.000000 0.000000 0.000000 0.000000

5 6286.387 9.436129 66.44666 7.730055 0.931146 5.364308 4.020434 6.071270

10 6342.600 8.973445 63.96962 8.841355 1.546502 6.551830 3.999989 6.117255

15 6344.096 8.977591 63.83607 8.845178 1.573841 6.622441 4.012543 6.132335

20 6344.209 8.977506 63.83208 8.844949 1.573890 6.624366 4.014191 6.133022

25 6344.218 8.977466 63.83168 8.844945 1.573957 6.624674 4.014257 6.133017

30 6344.218 8.977467 63.83166 8.844942 1.573960 6.624695 4.014260 6.133019

35 6344.218 8.977467 63.83166 8.844942 1.573960 6.624696 4.014260 6.133019

40 6344.218 8.977467 63.83166 8.844942 1.573960 6.624696 4.014260 6.133019

Source: Authors’ estimation.

Table 3.4 shows that both domestic OPGAP and money supply are the main factors affecting the

inflation rate in Vietnam. At the same time, the analysis above finds that the impact of money

supply to inflation is more persistent and higher than that of OPGAP. Although, the exchange

rate is responsible for 6% of changes in inflation, but the magnitude and timing are rather small.

Thus the impact of the changes in exchange rate on inflation is not significant. Moreover, both

VND and USD interest rates are effective to control inflation; however, the VND interest rate

plays a more important role compared to that of USD.

On average, the ERPT coefficient in Vietnam is 0.07 after 2 months. In order to have a better

idea of the level of the ERPTIP in Vietnam, it is worth to compare the ERPTIF coefficients

between Vietnam and some similar developing economies. According to Mwase (2006), this

figure in Tanzania was 0.023. Ito and Sato (2006) also calculate the ERPTIF in some East Asia

economies such as Korea (0.08), Thailand (0.07), Malaysia (0.03), Philippines (0.03), and

Indonesia (0.31). It may be concluded that the ERPT coefficient in Vietnam is at a moderate-

sized level. This moderate impact might be explained by: (1) the competitiveness is increasingly

improved, which limits firms’ ability to adjust the price when an exchange rate change occurs;

Page 12: Final Exchange rate pass-through to inflation in Vietnam_on Journal.pdf

12

(2) Vietnam adopts a pegged exchange rate regime and still controls the capital flows, thus the

impact of a shock to exchange rate to prices is restricted.

4. CONCLUSIONS

The aim of this paper is to apply the recursive VAR model to estimate the exchange rate pass-

through into consumer prices in Vietnam in recent years. It is found that the ERPT coefficient in

Vietnam is 0.07 after two months. The impact of a shock to exchange rate on consumer prices is

completely removed in the third month. Such an influence is rather small, insignificant and only

in a very short time. At the same time, the results of the variance decomposition and impulse

response analysis shows that money supply plays a crucial role in controlling inflation in

Vietnam. In addition to the issue of managing money supply, VND deposit interest rate is

another influential channel affecting inflation. Last but not least, the study also finds that an

increase in aggregate demand will cause prices increase. This result is quite different from that of

some previous studies on the relationship between inflation and aggregate demand.

These findings can have some important policy implications. Firstly, as the ERPTIF is relatively

moderate and not persistent, the SBV can adopt a more flexible exchange rate regime. This in

turn will make monetary policy more freedom to achieve other macroeconomic targets.

Secondly, high inflation in Vietnam is mainly due to the expansion of money supply. Controlling

inflation, therefore, first and foremost depends on how the SBV can manage the money

supply. Thirdly, if the money supply is effectively managed, an exchange rate arrangement

following market determinants will not cause inflation. In addition, VND interest rate is one

powerful tool to achieve the inflation target. But, as demand shocks affect inflation only in a

period of 4 months, the SBV need to make sure whether the increase in aggregate demand is

persistent before deciding to use such an instrument.

This paper, however, still has some limitations. The exchange rate shock affects the domestic

prices along the distribution chain from the import prices to production prices and then consumer

prices. With the availability of production prices and import prices we would have better VAR

model as well as better results. Moreover, given the limited span of data, covering only 5 years

from 2005 to 2009, it is not possible to capture the long run relationship between the exchange

rate and prices. Another regards the method to decompose residuals. In this paper, the Cholesky

decomposition is applied, but one can establish the decomposition by taking into account the

relationships between variables in the model with more economic senses. Hopefully, in the near

future, we can improve our research with much more interesting findings.

REFERENCES

1. Bank for Investment and Development Vietnam, BIDV.

2. Belaisch, A., (2003), “Exchange Rate Pass-Through in Brazil,” IMF Working

Paper,WP/03/141, International Monetary Fund.

3. Berbanke, B.S abd Mihov,I., (1998), “The liquidity effect and long-run neutrality”, NBER

Working paper 6608, National Bureau of Economics Research, Cambridge

4. Campa, J. M. and L. S. Goldberg, 2005, “Exchange Rate Pass Through into Import Prices,”

Review of Economics and Statistics, 87(4), pp. 679-690.

5. Campa, M. Jose and Goldberg, Pinelopi Koujianou (2002), “Exchange Rate Pass-Through

into Import Prices: A Macro or Micro Phenomenon”, NBER Working Paper 8934.

6. Dornbusch, Rudi (1987), “Exchange Rates and Prices”, NBER Working Paper, No.1769.

Page 13: Final Exchange rate pass-through to inflation in Vietnam_on Journal.pdf

13

7. Faruqee, H. (2006), “Exchange Rate Pass-Through in the Euro Area”, IMF Staff Papers,

53(1), pp.63-88.

8. Feinberg, R. (1986), “The Interaction of Foreign Exchange and Market Power Effects on

German Domestic Prices”, Journal of Industrial Economics 35 (1), pp. 61-70.

9. General Statistics Office – GSO.

10. Hahn, E. (2003), “Pass-Through of External Shocks to Euro Area Inflation”, Working

Paper, 243, European Central Bank.

11. Hang, Nguyen Thi Thu et al (2010), “Lựa chọn chính sách tỷ giá”, Báo cáo kinh tế thường

niên, VEPR.

12. Heidi Cigan, Anton Jevčák, Perceval Pradelle and Pavlina Žáková (2008), “Exchange Rate

Pass-Through to inflation in Slovakia”, ECFIN country focus, Vol 5, Issue 8.

13. International Financial Statistics - IFS.

14. Ito, Takatoshi and Sato, Kiyotaka (2006), “Exchange Rate Changes and Inflation in Post-

Crisis Asian Economies: VAR Analysis of the Exchange Rate Pass-Through”, NBER

working paper 12395.

15. Leigh, D. and M. Rossi (2002), “Exchange Rate Pass-Through in Turkey”, IMF Working

Paper, WP/02/204, International Monetary Fund.

16. Lian, An (2006), “Exchange Rate Pass-Through: Evidence Base on Vector Autoregression

with Sign Restrictions”, MPRA paper, No. 527.

17. Luc, Dieu Khanh (2008), “Monetary Transmission Mechanism in Vietnam after the Asia

Financial crisis: A structural VAR model”, Dissertation, Australia National University

18. McCarthy, J. (2000), “Pass-Through of Exchange Rates and Import Prices to Domestic

Inflation in Some Industrialized Economies”, Staff Reports, 111, Federal Reserve Bank of

New York.

19. Minh, Vo Van (2009), “Exchange rate pass-through and its implications for inflation in

Vietnam”, VDF Working Paper.

20. Nicoleta, C. (2007), “Estimating the exchange rate pass through into inflation in a vector

autoregressive framework”.

21. Nkunde Mwase. (2006), “An Empirical Investigation of the Exchange Rate Pass-Through

to Inflation in Tanzania”, IMF Working Paper, WP/05/150.

22. Olivei, G. P. (2002), “Exchange Rates and the Prices of Manufacturing Products Imported

into the United States,” New England Economic Review, Federal Reserve Bank of Boston,

First Quarter, pp. 3-18.

23. Otani, A., S. Shiratsuka and T. Shirota (2005), “Revisiting the Decline in the Exchange

Rate Pass-Through: Further Evidence from Japan’s Import Prices”, IMES Discussion, No.

2005-E-6, Institute for Monetary and Economic Studies, Bank of Japan.

24. Paul Krugman (1986), “Pricing to Market when the Exchange Rate Changes”, NBER

Working Papers 1926, National Bureau of Economic Research, Inc.

25. Taylor, J.B. (2000) “Low Inflation, Pass-through, and the Pricing Power of Firms”,

European Economic Review, 44, pp. 1389-1408.

Page 14: Final Exchange rate pass-through to inflation in Vietnam_on Journal.pdf

14

APPENDIX

Appendix 1: Stability of VAR model

Roots of Characteristic Polynomial

Endogenous variables: OPGAP DLOG(CPI) DLOG(VND) DLOG(FCD) DLOG(RVND)

DLOG(RUSD) DLOG(ER)

Exogenous variables: C DLOG(OIL)

Lag specification: 1 2

Date: 07/30/10 Time: 13:02

Root Modulus

0.367512 - 0.668206i 0.762604

0.367512 + 0.668206i 0.762604

-0.425861 - 0.598481i 0.734532

-0.425861 + 0.598481i 0.734532

0.696142 - 0.167243i 0.715950

0.696142 + 0.167243i 0.715950

-0.543306 - 0.111454i 0.554620

-0.543306 + 0.111454i 0.554620

-0.101345 - 0.491341i 0.501684

-0.101345 + 0.491341i 0.501684

-0.347862 0.347862

0.073805 - 0.204161i 0.217092

0.073805 + 0.204161i 0.217092

0.210858 0.210858

No root lies outside the unit circle.

VAR satisfies the stability condition.

Page 15: Final Exchange rate pass-through to inflation in Vietnam_on Journal.pdf

15

Appendix 2: Diagnostic tests

1. Variance of errors no change

Null hypothesis Ho: variance of errors are the same

Dependent Chi-sq(30) Prob.

res1*res1 23.61313 0.7892

res2*res2 32.75736 0.3332

res3*res3 28.29743 0.5547

res4*res4 39.24267 0.1204

res5*res5 38.23255 0.1439

res6*res6 44.20604 0.0457*

res7*res7 23.86202 0.7782

res2*res1 30.51614 0.4395

res3*res1 27.20582 0.6124

res3*res2 39.97541 0.1053

res4*res1 31.32896 0.3994

res4*res2 31.89555 0.3724

res4*res3 37.50074 0.1630

res5*res1 34.43757 0.2637

res5*res2 33.77062 0.2901

res5*res3 30.76108 0.4272

res5*res4 38.41713 0.1393

res6*res1 40.01251 0.1046

res6*res2 40.28299 0.0995

res6*res3 34.38598 0.2657

res6*res4 43.79419 0.0498*

res6*res5 46.71394 0.0265*

res7*res1 33.25259 0.3117

res7*res2 38.59700 0.1350

res7*res3 40.22355 0.1006

res7*res4 38.05574 0.1483

res7*res5 36.88892 0.1804

res7*res6 41.88165 0.0733

Page 16: Final Exchange rate pass-through to inflation in Vietnam_on Journal.pdf

16

2. Normality

Null hypothesis Ho: normal distribution

Component Jarque-Bera df Prob.

1 3.038227 2 0.2189

2 3.271247 2 0.1948

3 7.299513 2 0.0260*

4 5.981509 2 0.0502

5 1.888228 2 0.3890

6 5.029677 2 0.0809

7 3.240797 2 0.1978

3. VAR residual serial correlation LM tests

Null hypothesis Ho: no serial correlation at lag order h

Lags LM-Stat Prob

1 74.83582 0.0102

*

2 60.28117 0.1296

3 42.76038 0.7228

Probs from chi-square with 49 df.

Note: * insignificance at level of 5%