Inflation Dynamics in Transition Economy of Lao PDR.

31
The Australian National University Asia Pacific School of Economics and Government International and Development Economics Quantitative International Economics Inflation Dynamics in Transition Economy of Lao PDR. Submitted by: Fongchinda Sengsourivong U4082083 Submitted to: Tom Kompas and Ligang Song Date submitted: 8 June 2005

description

This paper investigates the relationship between inflation and macroeconomic variables using a sample of Laos in the period 1992-2004. Based on one-step error correction model of Engle-Granger approach, foreign price and exchange rate show the strong correlation with inflation process in Laos. This can be explained by a high degree of trade dependency in Lao economy. However, monetary base has the low impact on the price pressure partly because of monetary policy actions. Dollarization has influenced price instability in the short-run influences rather than in the long-run. In addition, there is evidence of inflation inertia in the process of inflation in Laos. Foreign interest rates positively relates to long-run inflation pressure in Laos through the transmission of traded good prices and dollarization channels. This paper also takes the endogeneity issue of explanatory variables into account by using the Hausman test approach. The endogeneity test shows that the model of single price equation may experience the endogeneity problem because monetary base shows evidence of interrelationship with price process.

Transcript of Inflation Dynamics in Transition Economy of Lao PDR.

Page 1: Inflation Dynamics in Transition Economy of Lao PDR.

The Australian National University Asia Pacific School of Economics and

Government

International and Development Economics

Quantitative International Economics

Inflation Dynamics in Transition Economy of Lao PDR.

Submitted by: Fongchinda Sengsourivong

U4082083

Submitted to: Tom Kompas and Ligang Song

Date submitted: 8 June 2005

Page 2: Inflation Dynamics in Transition Economy of Lao PDR.

Table of Contents

TABLE OF CONTENTS................................................................................................................................. 2 ACKNOWLEDGEMENTS............................................................................................................................. 3 ABSTRACT ...................................................................................................................................................... 4

INTRODUCTION............................................................................................................................................ 5

LITERATURE REVIEW................................................................................................................................ 6

MODEL SPECIFICATION ............................................................................................................................ 9

DATA AND ESTIMATION METHODOLOGY ........................................................................................ 10 DATA DESCRIPTION...................................................................................................................................... 10 ESTIMATION METHODOLOGY....................................................................................................................... 10

Unit roots test.......................................................................................................................................... 10 Johansen co-integration test ................................................................................................................... 11 Engle and Granger: Error Correction Models ....................................................................................... 12 Hausman endogeneity test ...................................................................................................................... 13

EMPIRICAL RESULTS................................................................................................................................ 13 UNIT ROOTS TEST RESULTS........................................................................................................................... 13 JOHANSEN CO-INTEGRATION TEST RESULTS................................................................................................. 14 THE ERROR CORRECTION MODEL RESULTS ................................................................................................... 14 LONG-RUN PRICE EQUILIBRIUM .................................................................................................................... 16 HAUSMAN ENDOGENEITY TEST RESULTS..................................................................................................... 17

DISCUSSION AND ANALYSIS OF RESULTS ......................................................................................... 18

CONCLUSION............................................................................................................................................... 21

REFERENCES ............................................................................................................................................... 23

APPENDIX A ................................................................................................................................................. 25 APPENDIX B.................................................................................................................................................. 30

2

Page 3: Inflation Dynamics in Transition Economy of Lao PDR.

Acknowledgements

I wish to express my gratitude to my research essay supervisor Dr. Tom Kompas for his insightful comments and support. I also would like to thank to our Academic and Research Skills Advisor, Anne Patching, for her English editing and structure of the essay.

3

Page 4: Inflation Dynamics in Transition Economy of Lao PDR.

Abstract

This paper investigates the relationship between inflation and macroeconomic variables

using a sample of Laos in the period 1992-2004. Based on one-step error correction model

of Engle-Granger approach, foreign price and exchange rate show the strong correlation

with inflation process in Laos. This can be explained by a high degree of trade dependency

in Lao economy. However, monetary base has the low impact on the price pressure partly

because of monetary policy actions. Dollarization has influenced price instability in the

short-run influences rather than in the long-run. In addition, there is evidence of inflation

inertia in the process of inflation in Laos. Foreign interest rates positively relates to long-

run inflation pressure in Laos through the transmission of traded good prices and

dollarization channels. This paper also takes the endogeneity issue of explanatory variables

into account by using the Hausman test approach. The endogeneity test shows that the

model of single price equation may experience the endogeneity problem because monetary

base shows evidence of interrelationship with price process.

4

Page 5: Inflation Dynamics in Transition Economy of Lao PDR.

Introduction Inflation is a global problem because almost all countries have suffered from one time or

another. Generally, inflation creates lower purchasing power and generates lower incomes,

especially in developing countries. Based on the recent history of high inflation in transition

economies, determinants of price instability in those emerging countries have been

extensively studied in the academic literature. Four main sources have focused on

examining the inflation process. First, fiscal dominance has associated to money growth

(Walsh 2003:145). Loungani and Swagel (2001); Fisher et al. (2002); and Lissovolik

(2003), found that fiscal imbalance related to money growth is highly possible to push the

price up. Second, balance of payment views related to a high level of openness - goods and

finance markets – assert that exchange rate pass-through has a high potential affecting

inflation instability. Montiel (1989); Bailliu et al. (2003); Lissovolik (2003);

Sinxayvoravong (2003); Aron et al. (2004) and Domac (2004) suggest that exchange rate

plays a crucial role in determining the volatility of inflation. Furthermore, expectation in

currency devaluation appears to be the most important factor fueling inflation pressure

(Himarios 1987; Goldfajn and Werlang 2000; Bilan and Siliverstovs 2005). Third, inertial

factors influenced from existence of wage contracts or rigidity adjustment of inflationary

expectations may be significant inflation components (Chopra 1985). Finally, discussed by

Bahmani-Oskooee and Domac (2002) and Sinxayvoravong (2003), dollarization leads to

velocity in inflation pressure given the government deficit financed by inflation tax.

Over the last ten years of economic development, Laos has experienced high and volatile

inflation (Appendix B: Figure 1). In mid 1995, the inflation rate went up to 30 per cent

related mainly to expansionary macroeconomic policies both fiscal and monetary policies

and depreciation in the exchange rate. The price velocity again with higher degree of

inflation during the era of Asian crisis in 1997 combined with expansionary fiscal policy in

1997/98 and 1998/99, peaked at 167 per cent in 1999. In dealing with price volatility,

tightening policies were introduced leading to inflation reduction.

The study on the factors affecting the fluctuation of inflation in Laos during 1992 to 2001

was carried out by Sinxayvoravong (2003). He found that inflation was affected by foreign

price, exchange rate, monetary base, dollarization, and income. Engle-Granger approach

5

Page 6: Inflation Dynamics in Transition Economy of Lao PDR.

was used for a single equation model to estimate one-step error correction model. However,

I find that his methodology missed the test of endogeneity, which might lead to invalidity

of the model estimation. In addition, I also find that he missed an important factor in

determining inflation process in Laos that is foreign interest rate. Therefore, this research

will include the foreign interest rate into the inflation model and will conduct endogeneity

test, in order to make his model more valid and accurate.

The process of modeling the price equation in Laos from 1992 to 2004 for this research

includes; firstly conducting the price model based on Sinxayvoravong framework (2003),

secondly, using econometric approaches as tools for estimating and testing the model: (i)

Augmented Dickey-Fuller (ADF) test of unit roots; (ii) Johansen co-integration test; (iii)

Engle-Granger Method – one-step error correction model – to estimate a single price

equation; and (iv) The Hausman edogeneity test.

The rest of this paper is organized as follows. Section 2 discusses findings of previous

research regarding determinants of inflation in emerging economies including Laos.

Section 3 briefly outlines the theoretical framework and inflation models. Section 4

presents data and estimation methodology. Section 5 reports the obtained results. Section 6

discusses and analyzes the results. Section 7 concludes.

Literature Review Some researchers focused on the identification of the factors to determine the inflation

pressure. Despite the variation of using methods and study on countries, the common views

on inflation factors in transition economies are exchange rate, expectation of exchange rate,

fiscal factor and dollarization influence. Montiel (1989); Bailliu et al. (2003); Lissovolik

(2003); Sinxayvoravong (2003); Arong et al. (2004); and Domac (2004) found that

exchange rate plays a crucial role in determining the volatility of inflation. These

researchers argued that substantial effects of exchange rate on inflation process were the

results of high degree of dependency. However, Reyes (2004) found a low relationship

between exchange rate and price owing to the intervention of monetary authority under

inflation targeting economies.

Expectation on exchange rate influencing inflation process is another factor that has been

explored. Himarios (1987); Goldfajn and Werlang (2000); and Bilan and Siliverstovs

6

Page 7: Inflation Dynamics in Transition Economy of Lao PDR.

(2005) had a similar view that devaluation expectations appear to be the most important

factor fuelling price innovation. In respect of this finding, there are uncertain methods for

identifying the proxy for expectation. As a result, these researchers applied different

proxies such as Himarios; and Bilan and Siliverstovs used wage rate whereas Goldfajn and

Werlang used survey data for future exchange rate to calculate exchange rate expectation.

A further inflation factor is fiscal dominance. On the basis of study across countries in a

similar period (1960s-1990s) but different methods, Loungani and Swagel (2001) and

Fisher et al. (2002) argued in the same view that fiscal imbalance related to money growth

is possible to push the price up. However, Loungani and Swagel found that this

phenomenon will exist merely in countries with floating exchange rate regimes. In contrast,

study in Ukraine under a period of relative macroeconomic stability (1996-2002) of

Lissovolik (2003) and Bilan and Siliverstivs (2005) found money growth has a negligible

impact on inflation even they used different approach. Similarly, Sinxayvoravong (2003)

by using case study of Laos found that the monetary base is a weak relationship in both the

short and the long term. Conversely, under the period of high inflation in Ukraine (1993-

1995), Lissovolik found that the budget deficit with almost full financing from the

monetary authority led to broad money high co-integrating price instability.

Effects of dollarization on inflation process are also investigated due to dominancy of

foreign currency in emerging market economies. Bahmani-Oskooee and Domac (2002)

employing VARs models for Turkey from 1990 to 2001 and Sinxayvoravong (2003) using

the single equation for Laos between 1992 and 2001 affirmed that dollarization leads to

velocity in inflation pressure given the government deficit financed by inflation tax.

Nonetheless, Sinxayvoravong (2003) showed additional evidence that the relationship

between dollarization and inflation pressure appear strongly only in the short run, not in the

long term.

Model comparison is another aspect that has been focused in order to conduct appropriate

macroeconomic policies, apart from identification the inflation factors. However, the

findings are various depending on the countries and situations of studying. Bailliu et al.

(2003) testing three existing models – Mark-up model, Monetary equation and Phillips

curves – found the mark-up model is the best model to forecast inflation in the context of

Mexico, while money model did not appear to be useful in its current form and the Phillips

7

Page 8: Inflation Dynamics in Transition Economy of Lao PDR.

curves could not explain when it comes to forecast inflation. Lissovolik (2003) using VARs

estimation asserted similarly to Bailliu et al. that the mark-up model was fitted to the

framework of Ukraine under the stable phenomenon, though the money model was more

applicable under high inflation circumstance. Domac (2004) tested the model under

different exchange rate regimes in Turkey and suggested that the mark-up model was more

appropriate under fixed exchange rate regimes, whereas in respect to floating exchange rate

regimes the Phillips curves is better.

Based on empirical evidence, the general agreement on inflation factors for emerging

countries are possible to be (i) monetary factor related to the budget deficit; (ii) exchange

rate; (iii) exchange rate expectation; and (iv) dollarization influence. However, the strong

relationship between inflation and its factors depends on country episodes.

There is only one study applying the econometric approach with a single equation model to

determine the inflation dynamics in Laos from 1992 to 2001 (Sinxayvoravong 2003). His

analysis focused on identifying the effects of dollarization on inflation. He also concerned

other factors such as foreign price, exchange rate, money reserve and income. Even he

provided intuitive explanations of the macroeconomic variables affecting the inflation

process in Laos, he did not confirm the model by testing endogeneity. This is a crucial part

of testing the consistency of model estimators. Therefore, this paper will use Hausman test

to examine the existing of endogeneity in the inflation model. In terms of completeness of

the inflation model, Sinxayvoravong model did not consider foreign interest rate. On the

basis of transition economies as Laos that is dominated by some certain degrees of

dollarization and domestic interest rate is fixed, foreign interest rate becomes a choice for

resource allocations. In addition, the accepting foreign exchange of banking system makes

foreign interest rate more crucial role as a rate of return of storing foreign assets or an

opportunity cost of holding domestic assets. Therefore, this paper will take foreign interest

rate incorporating in the Sinxayvoravong (2003) framework.

There are two hypotheses for this research. First foreign interest rate is an important factor

in explaining the inflation process in Laos. Second the need to test endogeneity of

explanatory variables in order to achieve the validity of model.

8

Page 9: Inflation Dynamics in Transition Economy of Lao PDR.

Model Specification Practically, inflation process is dynamics affected from various shocks. Many empirical

studies used structural models of inflation that consider all channels of feedback in

explaining inflation. However, it is a-theoretic due to less prior information (Gujarati

1995:749). Hence, it is less suited for policy analysis. Furthermore, on the basis of

underdeveloped financial system in Laos, few choices can be made for resource allocation.

Consequently, there is a limitation of inflation transmissions. Thus, it would be suitable to

conduct the model in single equation.

For inflation dynamic model in Laos, the research will use the inflation model of small

open-economy that popularly has been used in many studies of modeling inflation process

in transition economies (Himarios 1987; Lissovolik 2003; Sinxayvoravong 2003; Reyes

2004). This model assesses overall price in the economy by considering the effects of non-

traded and traded goods price. If the law of one price1 holds and foreign interest rate is a

price for resource allocation2, the single price model from Sinxayvoravong3 (2003) will

present as: f

tf

tttttt ipfyeMp 654321 loglogloglogloglog ββββββ +++++= (1) where, is the overall price in natural logarithm form; tplog

is monetary base in natural logarithm form; tMlog

is exchange rate of local currency (kip)/ US dollar, natural logarithm form; telog

is gross domestic products (GDP) in natural logarithm form; tylog

is dollarization index in natural logarithm form; tflog

is foreign price index in natural logarithm form; and ftplog

is foreign interest rate in natural logarithm form. fti

QTM provides the link between price and money moving in the same direction. Domestic

price development also responds positively with a change in exchange rate and foreign 1 Empirical exercise is performed PPP test by using FS test, the long-horizon approach. The results of PPP test satisfy the assumption. 2 Domestic deposit interest rate almost all the time is less than inflation rate. So, it cannot be an incentive choice for money demand in Laos. On the other hand, foreign interest rate (if ) seems to be an alternative choice because banking system has been allowed to accept foreign currency from society in terms of deposit and borrowing. 3 See his paper for more detail on the derivation of the model.

9

Page 10: Inflation Dynamics in Transition Economy of Lao PDR.

price level. This mechanism works through the ideas of purchasing power parity of open

economy. In contrast, Phillips curves present a trade-off between output and inflation

particularly in the short run. An increase in output reduces inflation. Capital mobility and

asset substitution show the link between foreign interest rate and traded good prices. When

foreign interest rate rises, foreign asset is more attractive. It leads to capital outflow and

then local currency depreciates. In addition, in case of dollarization environment, society

will shift to hold more foreign assets. High demand for foreign exchange push domestic

price higher through exchange rate depreciation.

Data and Estimation Methodology

Data Description The data used in this analysis is taken from the Bank of Lao PDR and International

Monetary Funds (IMF). The estimated sample uses monthly data in the period from

January 1992 to December 2004. Money supply that indicates the movement of fiscal

budget deficit is defined as monetary base. Exchange rate devaluation is defined as parallel

exchange rate in unit of Kip per US dollar because the authority uses this rate to adjust the

official exchange rate. Dollarization factor is measured as a proportion of foreign deposit in

the baking system to broad money (Sahay and Vegh 1995 and Balino et al. 1999). Foreign

prices are weighted by 60 percent for Thai prices and 40 percent for US prices

(Sinxayvoravong 2003). Foreign interest rate is defined as an average value between Thai

deposit interest rate and US deposit interest rate. Dummy variables are included in the price

model corresponding to particular policy and scenario such as Financial Crisis (July-

December 1997) and dramatic tightening policy (April-September 1999). Monthly GDP is

interpolated from annual data averaged by 12 in each year4 owing to the indifference

between actual GDP and its trend. Data descriptions are shown in Table 1 (Appendix A).

Estimation Methodology

Unit roots test Examining the integrating order of series, Augmented Dickey-Fuller (ADF) test is

conducted with the auto-regressions, AR (p=13). The test is chosen on the basis of

following considerations: (i) Even though the sequences of price index and parallel

exchange rate jumped rapidly during the period of 1997-1999 in Figure 2 and 3 (Appendix

4 Estimated by author.

10

Page 11: Inflation Dynamics in Transition Economy of Lao PDR.

B) and then came back to the same trend as usual by the early 2000, there are no priori

reason to expect a structural change; (ii) According to Said and Dickey (1984), if data is

finite data it can infer that autoregressive integrated moving average, ARIMA (p,1,q) is

equivalent to ARIMA(p,1,0). Hence, AR (p) is applied for ADF test; and (iii) Lag lengths

are determined by Schwarz Bayesian Criterion (SBC) test with the maximum of 13 lag

lengths because monthly data should consider possibility lag lengths no less than 12 months

(Enders 1995:236).

Johansen co-integration test If unit roots test suggest non-stationary series with indifference in integrating orders,

Johansen test approach is conducted to identify the long run equilibrium of variables in the

structural forms. This test can avoid the problems occurring from a single regression

(Enders 1995:385) such as: (i) There is no systematic procedure in the single regression

method because of separating estimation of the multiple co-integrating vectors; and (ii)

Two step of Engle-Granger procedure create problem because the estimated residual may

not appropriate to be a proxy for co-integration test due to the OLS properties.

The Johansen test procedures also consider deterministic specification and number of lag

lengths. Setting different deterministic will make the asymptotic distribution of likelihood

ratio (LR) test various. Unit root statistic test will provide this reference. Sufficient lag

lengths are determined on the basis of LR statistic tests5 recommended by Sims in 1980

with traditional VARs approach6 (Enders 1995:396-397). So, the estimated form of the

model to test co-integrating relations in the first differencing is

tpt

p

iitit xxAx εππ ++Δ+=Δ −

=−∑

1

10 (6)

The basic idea of this test is that if coefficient matrix (π ) is of rank n, the vector process is

stationary. However, if rank (π ) is equal to 1, there is a single co-integrating vector and the

5 LR test statistic: )loglog)(( urcTLR Σ−Σ−= ; Where, T is number of observations; c is number of parameters in the unrestricted system: c = pn+1, n is number of equations in the system (n=6), p is number of lags; iΣlog is natural logarithm of the determinant

of , i = r, u. r is number of lag restrictions and u is number of lag un-restrictions. The null hypothesis of

restricted VARs model is under . (df = rniΣ

2,05.0 dfχ 2).

6 Traditional VARs approach use stationary variables in the VAR model. So, VARs system is estimated by level form as suggested by Sims in 1980.

11

Page 12: Inflation Dynamics in Transition Economy of Lao PDR.

expression ptx −π is the error-correction factor. For other cases in which 1< rank (π ) <n,

there are multiple co-integrating vectors. The number of distinct co-integrating vectors can

be obtained by checking the significance of the characteristic roots (λ ) of π . Two test

statistics can be conducted to test for the numbers of characteristic roots that are maxλ and

traceλ statistic tests. It can calculate maxλ and traceλ 7 statistic from the characteristic roots of

π .

Engle and Granger: Error Correction Models Error correction model (ECM) is used in order to determinants of price and explaining their

dynamics of the economic model equation (5) if observed variables are non-stationary and

they are co-integrated (Engle and Granger 1987). If the obtained results from unit root tests

and co-integration test of Johansen approach are provide as Engle and Granger

representation theorem, then the short run dynamics of inflation can be described by ECM.

The model in general form presents as:

ttjt

n

iitji

n

iitit ppp εχγγχβββ +++Δ+Δ+=Δ −+−

=−

=− ∑∑ 1111

0110 logloglog (7)

where tχ is set of explanatory variables and 1γ is a speed of adjustment to log run

equilibrium. Equation (7) will be estimated by OLS estimation and the long run

relationship between price and independent variables can be derived from this short run

dynamic equation.

Assumptions:

(1) error term is white-noise process

Σ=′= )(;0)( ttt EE εεε for all t, where is mxm positive

definite matrix; for all t ≠ t ’ and

},.......2,1,,{ 2 mjiij ==Σ σ

0)( ' =ttE εε 0)/( =txE tε

7 Trace statistic for the null hypothesis of r co-integrating relations, while maximum eigenvalue statistic tests

the null hypothesis of r co-integrating relations against the alternative r+1 co-integrating relations. They are

computed as: ; ∑+=

−−=n

riitrace Tr

1

)ˆ1ln()( λλ )ˆ1ln()1,( 1max +−−=+ rTrr λλ

Where, is the estimated value of characteristic roots obtained from the estimated iλ̂ π matrix. T is the

number of usable observations. And r is the number of co-integrating equations.

12

Page 13: Inflation Dynamics in Transition Economy of Lao PDR.

(2) it−χ ’s are not perfectly correlation or no multi-correlation.

Hausman endogeneity test OLS estimates will be biased and inconsistent if endogeneity presents. The Hausman

specification test of endogeneity is conducted for testing the independence between the

stochastic regressor and the disturbances. Granger causality may not appropriate because

the weak condition to obtain the condition of exogeneity8 (Enders 1995:315). In this paper

will apply the Hausman test proposed by Davidson and Mackinnon (1989), which carries

out the test by running an auxiliary regression. Based on the relationship in equation (5),

three exogenous variables are suspected, namely , and , will be tested

whether they present an endogeneity or not. Nested models and instrument variables

criteria

Mlog elog flog

9 are considered. The test follows the step as: (i) identifying instrument variables:

lag terms in explanatory variables; (ii) regress three explanatory variables on other

explanatory variables ( and ) and the instrumental variables, then save their

residuals; (iii) Include those residuals in the equation (5); and (iv) Jointly test the

coefficients of those residuals. If they are jointly significant, reject the null hypothesis of

exogeneity.

fplog fi

Empirical Results

Unit roots test results Series are expected to have a time-variant mean due to trend movement in each series. In

addition, investigating autocorrelation from correlogram, all variables have slowly decayed

and Q- statistics provide significant of autocorrelation in each series. ADF test is used and

suggest the results that output series has trend stationary, I(0), whereas the other variables

have mean-reverting properties in the first differencing; I (1) process with intercept

(Appendix A: Table 2). Q-statistic from Box and Jenkins also shows correlation

coefficients decreasing of these non-stationary variables in the first differencing.

8 A necessary condition for the exogeneity of explanatory variables is that current and past values of dependent variable do not affect these explanatory variables. 9 Obtaining a suitable set of instrumental variables that are both sufficiently uncorrelated with the stochastic disturbance terms and sufficiently correlated with the relevant explanatory variables, T-test and F-test are used to test the significance of the instruments. Moreover, the Hausman test apply with nested models, hence each equation is one of the other.

13

Page 14: Inflation Dynamics in Transition Economy of Lao PDR.

Johansen co-integration test Results Even Johansen test can detect differing orders of integration; it is good not to mix variables

with multiple integrating orders. As a result of ADF test, GDP is not included. Then, to

ensure the white-noise process of residuals, LR statistic test is conducted. As shown in the

Table 3 (Appendix A), under null hypothesis of restricted VARs model, the lower value of

LR test statistics than critical value ( ), which is more acceptable the restricted

model. Hence, up to 4 lag lengths, it produces acceptable model compared to the other

choices.

2,05.0 dfχ

The Johansen co-integration test includes intercept (Ao), but no deterministic trend that is

determined by unit roots test. The LR test indicates setting 4 lags (p = 4). The test consists

of 6 non-stationary variables I (1) – domestic price index, monetary base, foreign interest

rate, foreign price index, and parallel exchange rate. So, equation (6) is estimated. Then, the

result from traceλ statistic suggests that the vector process has less than or equal to four co-

integration vectors, whereas that of from the maxλ statistic provides merely 3 co-integrating

equations at 95 per cent critical value (Appendix A: Table 4). As such case, the results of

traceλ and maxλ are conflict. Enders (1995) suggests that maxλ test is better than traceλ test

due to the sharper alternative hypothesis. Therefore, it can specific the number of co-

integrating vectors as 3 equations.

The error correction model results As a result of non-stationary I (1) process in each series and co-integrating relations, the

error correction model is estimated in order to capture the long run relationship of price

instability. Even ADF test and Johansen test provide different lag lengths results, but

Johansen test done is to verify the co-integration of multivariate non-stationary. On account

of VARs method, it considers effects of all series in the whole systems, whereas ADF test

carries out the test in particular series for identifying the integrating order of univariate.

Thus, to avoid misspecification, this paper will apply lag lengths outcomes from Johansen

test. As a result, the error correction model is estimated in the first differencing form with

up to four lags. The short-run dynamics presents in the specific form as:

14

Page 15: Inflation Dynamics in Transition Economy of Lao PDR.

)8(log

loglogloglog)log

logloglog(loglog

1615

141312111615

1413

4

112

4

110

tf

tt

tf

tttf

itif

iti

itif

itii

itii

itit

if

epMpip

epMpp

εγγ

γγγγββ

βββββ

+++

++++++

+Δ+Δ+Δ+=Δ

−−

−−−−+−+−

+−+−=

+−=

− ∑∑

Long-run relation can be derived from the short-run dynamics as:

fttt

fttt ifepMp

1

6

1

5

1

4

1

3

1

2

1

0 logloglogloglogγγ

γγ

γγ

γγ

γγ

γβ

−+

−+

−+

−+

−+

−= (9)

OLS estimate is applied for this one-step error correction model, equation (8). In this stage,

dummy variables for Asian financial crisis during July to December 1997 (D1) and

tightening monetary policy from April to September 1999 (D2) are incorporated in the

equation (8) in order to capture the effects of dramatically change. Furthermore, on account

of consistency, it is sensible to include these dummies placed the issue of omitted relevant

variables. The result is presented in Table 5 (Appendix A).

The short run dynamic model has a sensible statistic test. All coefficients are significant

and reasonable explaining the model by approximately 77 percent. Durbin-Watson statistic

shows the overall model serially uncorrelated. However, not all signs have an intuitive and

plausible. Clearly, dummy variables for Asian Crisis in 1997 and tightening monetary

policy in 1999 have apposite signs from expectations.

Therefore, the model will be examined for its adequacy. It also looks for remedies such as

omitted an important variable or have used the wrong function form. To determine whether

model inadequacy results from one or more of these problems, various methods to test

residual performance will be used. The results of the diagnostic test (Appendix A: Table 6)

suggest that the error term fulfill the classical assumptions. In specific, Cramer-von Mises

(W2) and Anderson-Darling (A2) test cannot reject the null of normal distributed error at 1

per cent significant level (Appendix B: Figure 4). In addition, LM test for serial correlation

and auto-regression conditional heteroskedasticity (ARCH) also depicts the satisfied

residuals that cannot reject the null of no serial correlation and no ARCH only up to order

1. In addition, the diagnostic test for model specification of RESET cannot reject the null of

homoskedasticity, independence of regressors and the correct of model specification up to

two fitted values. CUSUM test for the stability of error square also shows moderately stable

(Appendix B: Figure 5).

15

Page 16: Inflation Dynamics in Transition Economy of Lao PDR.

On the basis of the diagnostic tests, the short run dynamic model of inflation provides the

validity of outcomes. The adjustment coefficient of error correction ( ) for long run

equilibrium shows the intuitive sign with adjustment speed almost 6 months that is similar

to Sinxayvoravong results (2003). This result can reflect through the inertia in inflation that

100 percent change in price index in three months ago still influences the current change by

around 32 per cent, regarding to the effects of other explanatory variables.

1log −tp

As an expectation, exchange rate and foreign price are strongly significant having

contemporaneous effects on the price adjustment particularly foreign price effect. Precisely,

the elasticity of current price differencing with respect to that of exchange rate and foreign

price are 0.3 per cent and 2 per cent, respectively. This implication clearly reflects the

strong dependence of Lao economy on abroad.

Surprisingly, dummy for Asian crisis shows the negative impact on price. It indicates that

even there is high dependency; negatively external impact does positively influence Lao

economy. On the other hand, tightening monetary policy seems ineffective due to the

positive relation between price and the policy. These consequences differ from

Sinxayvoravong outcome (2003).

Holding constant the effects of other factors, current effect of monetary factor is

unexpectedly small. The current price elasticity with respect to current money base is

merely about 0.06 per cent. Dollarization influences and foreign interest rate present the

different outcomes. While, as an expectation, dollarization index shows positively effect on

current domestic price, foreign interest rate affects negatively.

Long-run price equilibrium The speed of adjustment coefficient, -0.23, shows the consistency in long-run relationship

(Enders 1995:371). So, it can derive the long-run equilibrium from the short-run relation in

Table 5 (Appendix A) as: f

ttttf

tt ifeMpp 02.0log36.0log785.0log173.0log892.0906.7log +−+++−= (10)

This long-run relationship is rational economic explanation. All signs are intuitive and

plausible. Specifically, factors from external influences such as foreign price and foreign

interest rate explore significant in terms of economic intuition and magnitudes particularly

16

Page 17: Inflation Dynamics in Transition Economy of Lao PDR.

foreign price. For instance, with considering the other effects, an increase in foreign

inflation rate by 100 per cent raises domestic inflation by almost 90 per cent. Beside that an

increase in foreign interest rate by 100 percentage point raises domestic inflation by around

2 per cent. Similarly, depreciation in local currency by 100 per cent boosts inflation up by

about 79 per cent regarding to other effects. These statistic relationships do not only

meaningful in economic intuition; they also reflect a high integration of Lao economy with

abroad.

Even monetary factor has a consistent sign in relation to inflation instability; it has small

impact on domestic price by only 17 per cent. This effect is possibly dominated by foreign

price and exchange rate factors, which show the strong dependency of Lao economy. In

contrast to the short run outcome, dollarization index in long run equilibrium has reverse

relation to inflation rate. The increase in the level of dollarization leads to the reduction in

inflation as similar to Sinxayvoravong results (2003).

Hausman Endogeneity Test Results The suspected variables as , , and regress on its exogenous and

instrument variables regarding to significant test for coefficients. The relationships present

as:

Mlog elog flog

tttf

tt eppe ναααα ++++= −− 131210 loglogloglog (11)

(12) ttf

ttf

tt MifpM ϑφφφφφ +++++= −−−− 141312110 loglogloglog

(13) tttf

ttf

tt fMiepf ωδδδδδδ ++++++= −−−− 1514312110 logloglogloglog

Artificial model (14) for Hausman test recommended by Davidson and Mackinnon (1989)

is constructed on the basis of the relationship in equation (5)10 and estimated residuals,

ttt andωϑν ˆˆ,ˆ obtained from (11), (12) and (13) respectively.

ttttf

ttttf

tt ifMepp εωϕϑϕνϕϕϕϕϕϕϕ +++++++++= ˆˆˆlogloglogloglog 876543210

(14)

The model (14) is estimated by OLS approach that provides the statistic value for testing

being endogenous of explanatory variables. The Hausman test results are illustrated in table

7 (Appendix A). Residuals of three suspected explanatory variables are jointly significant

rejecting the null of exogeneity in favor of endogeneity. However, two residuals of

10 GDP does not include because the different order of intergration.

17

Page 18: Inflation Dynamics in Transition Economy of Lao PDR.

exchange rate and dollarizarition index together cannot reject the null of exogeneity.

Therefore, it can conclude that the model (5) may face endogeneity problem particularly

monetary factor (reserve money) cannot meet the condition of exogeneity. This problem

will lead to the classical assumptions violated with inconsistency.

Discussion and Analysis of Results The overall results indicate that the single inflation model is plausible to interpret inflation

phenomenon in Laos. Both the short-run and long-run relationships are meaningful to cite

the intuitive explanations for price instability. On the basis of the results and Laos

situations, five important points are worthwhile for this discussion. First, the substantial

effects of foreign price and exchange rate on inflation process in Laos express the

explanatory power. Clearly, this outcome closely links to the high degree of international

integration of small-open economies as Laos.

Second, the significantly reserve money entirely reflects the underlying the phenomenon of

Lao economic development. It shows the evidence of fiscal dominance matters particularly

in 1995 and during 1998 to 1999 that monetary base requires to be adjusted endogenously

to fulfill the expansionary fiscal policy. This economic phenomenon is known as Non-

Recardian regime (Walsh 2003). However, even monetary base depicts positively

significant but relatively low effects compared to foreign price and exchange rate. One

reason can be the dominant effects of foreign price and exchange rate. Another reason is

that the monetary policy actions have performed a relatively-fixed domestic interest rate

hence money needs to adjust endogenously to ensure the economy on the track of

development target (controlling inflation).

Third, even foreign interest rate changes the sign of correlation from negative in the short-

run to positive in the long-run relationship; it results of estimation method of the error

correction model that can be varied. This method has considered the first differencing as

dependent variable in order to address the issue of non-stationary variables. However,

positive relationship in the long-run is more considerable. The positive correlation between

foreign interest rate and domestic price index reflects the evidence of asset substitution

particularly in terms of foreign exchange. Domestic price will be pulled up through two

major channels, namely dollarization influences and traded good prices.

18

Page 19: Inflation Dynamics in Transition Economy of Lao PDR.

The first channel, on the basis of less financial development, there is less choice for asset

allocations. Thus, under foreign exchange officially held by public and deposited at the

banking system, higher foreign interest rate is attracted by public to increase their demand

for foreign currency11. Asset substitution will shift domestic assets to foreign assets. The

high demand for foreign exchange will raise a higher degree of dollarization both in

banking system and non-banking systems. It leads to lower credibility of local currency.

The depreciation in exchange rate will push the imported price in unit of local currency

more costly. Then, the general price level will go up. The second channel, on account of

Fisher identity, an increase in foreign interest rate will infer that foreign price will go up as

constant real rate of return. Clearly, traded good prices will positively respond and lead to

the general price in domestic rising.

It shows that even low magnitude, foreign interest rate has the crucial relationship on

inflation process by working through traded good prices and dollarization12. Therefore,

Sinxayvoravong paper that missed this variable may provide incomplete explanation of

inflation process in Laos.

Fourth, Inflation inertia appears significantly in the short-run price equation. Even price

liberalization implemented after market reform, but many price products still continue

under control or carefully managed. It might be the effects of price controlling on strategy

products mainly relating to the traded goods. For instance, fuel products with price ceiling

are the major import goods. Some domestic goods are also under control such as service fee

of electricity, water supply and some agricultural products. In banking system, interest rate

is set relatively fixed. This controlling leads to slowly adjustment of current price index.

The final result is Hausman test that shows an evidence of single equation model is

experiencing inconsistency because monetary base shows the significant interrelationship

with price index. This finding of the Hausman test for endogeneity also reflects the policy

actions that affect reserve money responding endogenously to ensure those policies being

11 Doolarization is associated with an increase in banking spreads because an increase in deposit dollarization is related to an increase in offshore deposits (Honohan and Shi 2001). 12 Bearing in mind, the long-run relation of dollarization has negative impacts on price index. Sinxayvoravong (2003) explained that high dollarization level implies low possibility of the government to generate inflation through inflation tax of financing budget deficit. It makes inflation lower. He also pointed that it may be just the simple relationship of these two series.

19

Page 20: Inflation Dynamics in Transition Economy of Lao PDR.

able to aim their targets. Therefore, it would violate the results of Sinxayvoravong with less

validity (2003).

Nevertheless, in the short-run dynamics model, it is exclusively composed the difference in

lag variables and its own lags. So, the implications from the short-run model still remain

validity and adequacy. It implies that the long-run equation derived from the consistent

estimators will provide valid outcomes (Enders 1995).

In terms of methodology, the substantial changes in particular period normally lead to

heteroskedasticity problem and instability of residual. These problems exist when it comes

to the divergence in data series. Figure 5 (Appendix B) shows the cumulative sum

(CUSUM) of the recursive residuals going outside the area between the two critical lines

particularly in the first period of the sample. On the other hand, after the critical period,

financial crisis and stabilizations, this problem has no longer substantial effect and CUSUM

of residual again move back to lie inside the lines. RESET test also confirms no

heteroskedasticity of residual (Appendix A: Table 6).

These valid implications are presented underlying the data structure, theoretical framework

and econometric methodology. Clearly, data require the critical considerations with reliable

collection system and the substantial changes. Basic ADF13 test for non-stationary in series

sequences is sufficiently valuable to identify the existing of unit roots. However, output14

with deterministic trend stationary possibly results from the basic approach of monthly

interpolation (average by 12).

The model framework and the estimation methodology lead to applicable outcomes for

reality in Lao economy. The model is set on the basis of the small open economy in the

period of transitional market that is possible to have a high degree of dependency.

Therefore, all explanatory variables are significantly correlated with price level. However,

it is important to note that one-step estimation method of error correction model with the

first differencing as a dependent variable may give the results variation. The short-run

dynamic price model may provide various signs that differ from the economic intuition 13 Econometric exercises to test unit roots by Perron’s test for structural change and ADF-detrending provide similar results to basic ADF test. 14 If unit roots test provides incorrect result, this model is not possible to face the problem of omitting relevant variables. According to the econometric exercise, output does not show significant correlation with price index.

20

Page 21: Inflation Dynamics in Transition Economy of Lao PDR.

such as foreign interest rate and dummy variables for Asian financial crisis and tightening

monetary policy.

These variations tend to appear with the longer lag lengths. So far the longer lag lengths

incorporate in the model mainly because of correcting auto-correlations. Furthermore, the

sufficient lag lengths will ensure the adequacy of the short-run model that will bring about

the consistency of long-run relationships.

Conclusion This paper is aimed at exploring dynamic relationship between inflation and five other

macroeconomic variables: money growth, exchange rate, foreign price, dollarization and

foreign interest rate. This analysis bases on one-step error correction model of Engle-

Granger that similar to the previous paper in Laos. However, this paper makes two new

principal contributions for quantitative analysis in Laos. First, awareness of the problem

from the single equation model, endogeneity test by Hausman test method is implemented.

The implication is that money growth shows an evidence of deteriorating exogeneity

conditions, in which the monetary base has substantial interrelationship with inflation.

Second, foreign interest rate newly introduced into the price single equation; in the long-run

it expresses the positively significant effects on inflation through the channels of traded

good prices and dollarization.

The other substantial consequences present as (i) there is an evidence of ample influence of

foreign price and exchange rate on inflation dynamics in Laos. This outcome associates

with a high degree of Lao economic dependency; (ii) even the effects of money growth on

price dynamics are weak owing to the monetary policy actions, it is important to explain the

role of fiscal policy; (iii) dollarization plays positive role in determining the shot-run

inflation process in Laos, but not in the long-run; and (iv) the evidence of substantial

inflation inertia can be attributed to high degree of price control in the country.

Even there is an issue of endogeneity of explanatory variables; the short-run dynamic price

model remains validity and consistency due to its lags and the differencing in lag variables

as the components of explanatory variables. The long-run price equilibrium derived from

the short-run model should produce consistent results. However, seeking for instrument

21

Page 22: Inflation Dynamics in Transition Economy of Lao PDR.

variables or changing methods of estimation from considering one-side effect to

interrelationships of system equations should be considered in the next step of study.

This paper also considers the limitations in the research. First, according to unknown the

form of the data-generating process whether AR (P) or MA (q) or ARIMA (p,I,q), ADF test

may mislead. Second, if some series sequences have structural change, ADF test will be

biased and tends to accept the null of unit roots. Third, it may true that our series may have

trend stationary process. So it could lead to misspecification of inflation dynamic model.

Fourth, the short-run model may face the problem of third factor argument because the

evidence of dropping either one or another variable will make the remaining variables have

low power of test (Appendix A: Table 8). Fifth, the long-run equilibrium may misinterpret

in terms of significant results even it is the outcome of consistent derivation from the short-

run dynamics. It requires specifying the asymptotic statistic references. Sixth, limitation of

variables may provide inappropriate instrument variables that will lead to incorrect results

of the Hausman test. Finally, forward-looking factor is not considered, it may lead to model

inconsistency.

Taken together the limitations, structural system equations is a competitive choice for the

next study, in which it can deal with problem such endogeneity of explanatory variables

and less chance to omit relevant variables.

22

Page 23: Inflation Dynamics in Transition Economy of Lao PDR.

References Aron, J., Muellbauer, J. and Smit, B., 2004. ‘Modeling the Inflation Process in South

Africa’, Centre for the Study of Agrican Economies, Oxford University.

Bahmani-Oskooee and Domac, I. 2002. ‘On the Lonk between Dollarization and Inflation: Evidence from Turkey’, http://www. temb.gov.tr/research/discuss /dpaper59.pdf,(23 March 2005).

Bailliu, J., Garces, D., Kruger, M. and Messmacher, M., 2003. ‘Explaining and Forecasting Inflation in Emerging Markets: The case of Mexico’, Bank of Canada, Working Paper 2003-17.

Bilan, O. and Siliverstovs, B. 2005., ‘Inflation Dynamics in the Transition Economy of Ukraine’, Institute for Economic Research and Policy Consulting, Working Paper, No. 28.

Chopra, A. 1985., ‘The Speed of Adjustment of the Inflation Rate in Developing Countries: A study of Inertia’, IMF Staff Papers, 32(4):693-733.

Davdison, R. and Mackinnon, J. 1989., ‘Testing for Consistency using Artificial Regressions’, Econometric Theory 5: 363-384.

Domac, I., 2004., ‘Eplaining and Forecasting Inflation in Turkey’, World Bank Research

Paper, No.3287.

Enders, W. 1995. Applied Econometric Time Series, Wiley, New York. Engle, Robert E. and Granger, C. 1987., ‘Cointegration and Error-Correction:

Representation, Estimation, and Testing’, Econometrica. 55:251-76. Fisher, S., Sahay, R. and Vegh, C. 2002., ‘Modern Hyper- and High Inflation’, Journal of

Economic Literature, Working Paper 8930.

Goldfajn, I. and Werlang, S. 2000., ‘The Pass-through from Depreciation to Inflation: a Panel study’, Working Paper Series. No.5.

Gujarato, D. 1995. Basic Econometrics, McGraw-Hill, Singapore. Himarios, D. 1987., ‘Devaluation, Devaluation Expectations and Price Dynamics’,

Economica, 54(215):299-313.

Honohan, P. and Shi, A., 2001. ‘Deposit Dollarization and the Financial Sector in Emerging Economies’, World Bank Policy Research Working Paper, WPS 2748.

Lissovolik, B. 2003., ‘Determinants of Inflation in a Transition Economy: The case of

Ukraine’, IMF Working Paper, WP/03/126.

23

Page 24: Inflation Dynamics in Transition Economy of Lao PDR.

Loungani, P. and Swagel, P. 2001., ‘Sources of Inflation in Developing Countries’, IMF Working Paper, WP/01/198.

Montiel, P.J. 1989., ‘Empirical Analysis of High-Inflation Episodes in Argentina, Brazil and Israel’, IMF Staff Papers, 36 (3):527-549.

Reyes, J. 2004., ‘Exchange Rate Pass-through Effect and Inflation Targeting in Emerging Economies: what is the relationship?’, University of Arkansas.

Said, S. and Dickey, D. 1984., ‘Testing for Unit Roots in Autoregressive-Moving Average Models with Unknown Order’, Biometrica, 71:599-607.

Sinxayvoravong, B. 2003., ‘Doolarization and Monetary Implications: Case of Laos PDR’, PhD dissertation with National Centre for Development Studies. Australian National University.

Wallace, F. and Shelley, G. 2005., ‘An Alternative Test of Purchasing Power Parity’, Prairie View A&M University and East Tennessee State University.

Walsh, C. E., 2003. Monetary Theory and Policy, 2nd edn, MIT Press, Cambridge.

24

Page 25: Inflation Dynamics in Transition Economy of Lao PDR.

Appendix A

Table 1 Descriptive statistics for main series

Variables

Mean

Maximum

Minimum

Std. Dev.

No. of Obs

plog 3.614 5.036 2.163 1.078 156 fplog 4.527 4.693 4.300 0.118 156

Mlog 5.473 7.350 3.254 1.268 156 elog 7.962 9.302 6.583 1.165 156

fi 5.296 9.795 1.025 2.647 156 flog -0.644 -0.178 -1.313 0.337 156 ylog 1.993 2.339 1.617 0.228 156

Source: Bank of the Lao PDR. (BOL), IMF: International Financial Statistics (IFSs) and author estimates. Table 2 Unit root tests: Augmented Dickey-Fuller (ADF) test

Level First differences

Variables ADF-test statistic Lags Variables ADF-test statistic Lags

Log(P)

Log(MB)

Log(f)

Log(pf)

if

log(MER) Log(y) #

-0.6579 -0.4475 -1.3102 -2.4259 -0.545245 -0.5777 -4.841***

2 2 1

1

1

1

1

Dlog(P) Dlog(MB) Dlog(f) Dlog(pf) Dif

Dlog(MER)

-4.5275*** -7.851*** -12.9994*** -8.8747*** -9.604693*** -7.3506***

1

1

0

0

0

0

Note: test critical value with intercept at 1per cent, 5 per cent and 10 per cent provides significant level as -3.4731; -2.8802 and -2.5768. # depicts that output is tested with deterministic trend and critical value at 1 per cent, 5 per cent and 10 per cent is -4.02, -3.44 and -3.14 respectively. The superscript ***; ** and * denotes rejecting null hypothesis (unit roots) at 1 per cent; 5 per cent and 10 per cent. Source: Author’s calculation

25

Page 26: Inflation Dynamics in Transition Economy of Lao PDR.

Table 3 Likelihood ratio test: lag lengths test

Null

hypothesis

P=8 against

p=12

P=4 against

p=8

P=2 against

p=4

P=1 against

p=2

P=1 against

p= 4

P=3 against p=4

LR 61.009*** 114.819*** 128.478 28.354*** 154.016 81.551

Df 144 144 72 36 108 36 2

,05.0 dfχ 124.34 124.34 92.8 51 124.34 51

Note: p is number of lags. The superscript *** denotes that cannot reject the null of restricted VARs model at

5% critical value.

Source: Author’s calculation

Table 4 Johansen test: the maxλ and traceλ tests

Null hypothesis Alternative hypothesis 95% Critical Value

traceλ tests

0=r

1≤r

2≤r

3≤r

4≤r

maxλ tests

0=r

1=r

2=r

3=r

0>r

1>r

2>r

3>r

4>r

1=r

2=r

3=r

4=r

traceλ value

176.848

116.263

69.657

35.465

15.258***

maxλ Value

60.584

46.607

34.191

20.207***

103.847

76.973

54.079

35.193

20.262

40.857

34.806

28.588

22.299

Note: r is number of co-integrating relations (the co-integrating rank).

The superscript *** denotes that cannot reject the null hypothesis at 5 per cent critical value.

Source: Author’s calculation

26

Page 27: Inflation Dynamics in Transition Economy of Lao PDR.

Table 5 Short run dynamic estimation, dependent variable is tplogΔ

Variable

Coefficient

Std. Error

t-Statistic

Constant -1.812*** 0.324 -5.591 f

tplogΔ 1.926*** 0.561 3.434 f

tp 2log −Δ -1.069* 0.594 -1.800

tMlogΔ 0.059** 0.029 2.033

telogΔ 0.294*** 0.039 7.466

2log −Δ te -0.218*** 0.050 -4.335

3log −Δ te -0.132*** 0.043 -3.099 * 4log −Δ te -0.180*** 0.040 -4.454

fti 3−Δ -0.013** 0.005 -2.581

2log −Δ tp 0.322*** 0.066 4.849

1log −Δ tf 0.065* 0.036 1.786 f

tp 1log − 0.204*** 0.076 2.693

1log −tM 0.039*** 0.012 3.191

1log −te 0.179*** 0.022 8.329 f

ti 1− 0.005** 0.002 2.516

1log −tf -0.083*** 0.022 -3.799

1log −tp -0.229*** 0.027 -8.556 D1 -0.033*** 0.009 -3.614 D2 0.037*** 0.009 3.901 R-squared 0.774788 F (13, 137) 25.22861 Adj. R-squared 0.744078 Prob(F-statistic) 0.000 RSS 0.038599 DW statistic 2.059396

Note: the superscripts ***, ** and * denote rejection at 1 per cent, 5 per cent and 10 per cent critical values.

*it includes based on the empirical work to make the error correction model well-behave. Source: Author’s calculation

27

Page 28: Inflation Dynamics in Transition Economy of Lao PDR.

Table 6 Diagnostic tests for the short run dynamic price model, equation (8)

Tests

Methods

Statistic test

Probability

Normality test Cramer-von Mises (W2) 0.157 0.019 Anderson-Darling (A2) 0.948 0.017 Serial Correlation LM test up to order 1 F(1, 131) 0.183 0.669 Obs*R-squared 0.211 0.646 ARCH Test up to order 1 F (1,148) 1.089 0.298 Obs*R-squared 1.096 0.295 Ramsey RESET Test: up to 2 fitted terms F (2,130) 1.972 0.143 Log likelihood ratio 4.513 0.105

Source: Author’s calculation

Table 7 Hausman test for edogeneity

Null hypothesis

Statistic tests

P-value

0ˆˆˆ === ttt ωϑν F (3,146) = 3.922*** 0.0099 0ˆˆ == tt ων F (2, 146) = 1.030 0.359 0ˆˆ == tt ϑν F (2,146) = 5.655*** 0.004

0ˆ =tϑ F (1,146) = 7.998*** 0.005 Note: the superscripts *** denotes rejection at 1 per cent critical values.

Source: Author’s calculation

28

Page 29: Inflation Dynamics in Transition Economy of Lao PDR.

Table 8 The short-run dynamic model with dropping ftplogΔ

Variable Coefficient Std. Error t-Statistic Constant -1.982*** 0.346 -5.726

ftp 2log −Δ -1.731*** 0.621 -2.789

tMlogΔ 0.049 0.032 1.511 telogΔ 0.299*** 0.042 7.050

2log −Δ te -0.241*** 0.054 -4.497

3log −Δ te -0.128*** 0.045 -2.826 * 4log −Δ te -0.215*** 0.043 -5.007

fti 3−Δ -0.0002* 0.002 -0.125

2log −Δ tp 0.347*** 0.072 4.813

1log −Δ tf 0.064 0.039 1.647 f

tp 1log − 0.229*** 0.079 2.867

1log −tM 0.039*** 0.013 3.085

1log −te 0.193*** 0.023 8.480 f

ti 1− 0.005*** 0.002 2.795

1log −tf -0.095*** 0.024 -4.015

1log −tp -0.242*** 0.028 -8.553 D1 -0.028*** 0.009 -2.871 D2 0.036*** 0.011 3.320 R-squared 0.743 F (1, 129) 21.966 Adj. R-squared 0.709 P_value 0.000 RSS 0.044 DW statistic 1.828

Note: the superscripts ***, ** and * denote rejection at 1 per cent, 5 per cent and 10 per cent critical values. *it includes based on the empirical work to make the error correction model well-behave. The difference in monetary base, dollarization and foreign interest rate has lower power of test in the short-run model with dropping comparing to the results from table 5. f

tplogΔ Source: Author’s calculation

29

Page 30: Inflation Dynamics in Transition Economy of Lao PDR.

Appendix B

Figure 1 inflation rate in Laos from 1991to 2004

Yearly Inflation rate

020406080

100120140160180

M1

1990

M3

1991

M5

1992

M7

1993

M9

1994

M11

199

5

M1

1997

M3

1998

M5

1999

M7

2000

M9

2001

M11

200

2

M1

2004

Source: Bank of the Lao PDR and IMF

Figure 2 General Price Innovation, 1992-2004

2.0

2.4

2.8

3.2

3.6

4.0

4.4

4.8

5.2

1992 1994 1996 1998 2000 2002 2004

LNP

Source: Author’s calculation

30

Page 31: Inflation Dynamics in Transition Economy of Lao PDR.

Figure 3 Exchange Rate, 1992-2004

6.5

7.0

7.5

8.0

8.5

9.0

9.5

1992 1994 1996 1998 2000 2002 2004

LNE

Source: Author’s calculation

Figure 4 Normal distribution of residuals

0

4

8

12

16

20

24

28

32

-.04 .00 .04

RESID01

Kernel Density (Normal, h = 0.0045)

Normality of Residual

Source: Author’s calculation

Figure 5: Cumulative Sum of Squares of Recursive Residuals

-0.2

0.0

0.2

0.4

0.6

0.8

1.0

1.2

1999 2000 2001 2002 2003 2004

CUSUM of Squares 5% Significance

Source: Author’s calculation

31