Internatinoal Journals Call for paper: http://www.iiste.org/Journals

- Transcript Header:
- Money, income, and prices in bangladesh a cointegration and causality analysis

- Transcript Body:

- 1. Journal of Economics and Sustainable Development www.iiste.orgISSN 2222-1700 (Paper) ISSN 2222-2855 (Online)Vol.3, No.7, 2012 Money, Income, and Prices in Bangladesh: A Cointegration and Causality Analysis Md. Nisar Ahmed Shams Associate Professor, Department of Economics Jahangirnagar University, Savar, Dhaka -1342, Bangladesh E-mail: nishu11us@yahoo.comAbstractThis paper re-examines the causal relationship between money, income and prices in Bangladesh during theperiod 1972/73 to 2009/10. Cointegration analysis indicates a long-run relationship among the variables. Basedon the Error Correction Model (ECM), a bidirectional causality between money and income has been observed.Therefore, monetary policy should be formulated by taking into account the feedback effects of output onmoney. Money supply can be considered as an effective control variable as causality is found to run from moneyto prices supporting the Monetarists.Key Words: Cointegration, Error Correction Model, Bivariate Causality, Trivariate Causality, Bangladesh.1. IntroductionThe relationship between money, income, and prices has been a matter of debate among different economistsparticularly between the Monetarists and Keynesians. The Monetarists consider money supply to be theimportant factor leading to changes in income and prices. Thus, the direction of causation runs from money toincome and prices without any feedback. On the contrary, the Keynesians asserts that changes in income leads tochanges in the stock of money through the demand for money. Therefore, the direction of causation runs fromincome to money without any feedback.The causal relationship between money and the other two variables, i.e., income and prices has been an issueargued by economists particularly after the seminal paper by Sims (1972). Using post-war quarterly data forU.S. in a bivariate framework, he found evidence of unidirectional causality from money to income as claimedby the monetarists. However, this result was not obtained by subsequent studies. Replicating Sims’ test in theCanadian economy, Barth & Bannett (1974) showed bidirectional causality between money and income.Williams et al. (1976) employing a similar approach found evidence of unidirectional causality from income tomoney in case of U.K., opposite to Sims’ findings. However, Dyreyes et al. (1980) showed evidence ofbidirectional and unidirectional causality between money and income in U.S. and Canada respectively.Concerning the relationship between money and prices, the studies undertaken by Bengali et al. (1999), Husain& Rashid (2006) found evidence of unidirectional causality running from money to prices in Pakistan. Lee & Li(1983) showed unidirectional causality from money to price in Singapore. Causality is also found to run frommoney supply to price movements in Malaysia as observed by Ghazali et al. (2006). The results obtained bythese studies confirm the claim made by the Monetarists. On the contrary, Jarrah (1996) found money to beGranger caused by prices in Saudi Arabia. Mishra et al. (2010) also found unidirectional causality from pricelevel to money supply in India.Regarding Bangladesh, Jones & Sattar (1988) examined the causal relationship between money-income andmoney-inflation. They found evidence of unidirectional causality from money to output. Similar result was alsoobtained regarding the relationship between money and prices. Therefore, monetary expansion could have asignificant impact on output growth while there might be inflation in the economy. Shams et al. (2010) foundunidirectional causality from money to income in Bangladesh. Ahmed (2003) while investigating multivariatecausality among money, interest rates, prices and output, identified bidirectional causality between money andprices in Bangladesh.The objective of this paper is to re-examine the causal relationship between money-income and money- prices inBangladesh. The study concentrates on Cointegration and Error Correction Model (ECM) to look into thebivariate causal relationship by taking care of the stochastic properties of the variables. The major drawback ofbivariate analysis is the exclusion of relevant variable(s). Such omission may lead to erroneous conclusions. 82

- 2. Journal of Economics and Sustainable Development www.iiste.orgISSN 2222-1700 (Paper) ISSN 2222-2855 (Online)Vol.3, No.7, 2012Therefore, this study also attempts to investigate the causal relationship through trivariate causality. The rest ofthe paper is arranged as follows. Section 2 provides the data sources. The methodology and empirical results arepresented in section 3. The final section contains the conclusions and policy recommendations.2. DataThis study is based on annual data covering the period from 1972/73 to 2009/10. Gross Domestic Product (GDP)has been considered to represent income (I). Data on Gross Domestic Product (GDP) and Broad Money (M)which includes time deposits along with narrow money have been obtained from various publications ofEconomic Trends, published by the Bangladesh Bank. GDP and M are expressed in terms of Taka (DomesticCurrency of Bangladesh) in Millions. To measure prices (P), the Consumer Price Index (CPI) is used which hasbeen collected from different issues of Statistical Yearbook of Bangladesh. Econometric estimations have beendone by using STATA 9.2.3. Methodology and Empirical ResultsThe causal relationship between two variables is tested through the standard Granger (1969) causality frameworkby estimating the following equations: m n (1 − L)Yt = α 0 + ∑ α i (1 − L)Yt −i + ∑ β j (1 − L) X t − j + ε t (1) i =1 j =1 m n (1 − L) X t = δ 0 + ∑ δ i (1 − L) X t −i + ∑ γ j (1 − L)Yt − j + ωt (2) i =1 j =1where L is the lag operator, ε and ω are mutually uncorrelated white noise series and t denotes time period.Causality may be determined by estimating equations (1) and (2) by testing the null hypothesis that β j = γ j = 0 for all j’s against the alternative hypothesis that β j ≠ 0 and γ j ≠ 0 for at least some j’s. If thecoefficients β j ’s are statistically significant but γ j ’s are not, then Y is said to have been caused by X. Thereverse causality holds if γ j ’s are statistically significant while β j ’s are not. If both β j and γ j are significantthen causality runs both way.In addition to bivariate causality, this paper also attempts to examine the causal relationship through trivariatecausality i.e., the causal relationship between money and income conditional on the presence of prices. Similarly,it tests the causal relationship between money and prices conditional on the presence of income. In order to testthe joint influence of two variables on the third variable, the joint trivariate causality model is specified as: m n p (1 − L)Yt = χ 0 + ∑ χ i (1 − L)Yt −i + ∑ ϕ j (1 − L) X t − j + ∑ λ k (1 − L) Z t − k +u t (3) i =1 j =1 k =1 m n p (1 − L) X t = λ0 + ∑ λi (1 − L) X t −i + ∑ψ j (1 − L)Yt − j + ∑ φ k (1 − L) Z t − k +vt (4) i =1 j =1 k =1In the trivariate specification, (i) X and Z Granger-cause Y if ϕ j = λ k = 0 is not true i.e., ϕ j = λ k = 0 isrejected, (ii) if ψ j = φ k = 0 is rejected, Y and Z Granger-cause X. A feedback system exists if (i) – (ii) holdsimultaneously. Finally, X, Y, and Z are causally independent if all the coefficients of X and Z in equation (3), Yand Z in equation (4) are not statistically different from zero.The causality tests involving money, income, and prices are carried out in the following three steps. Step Iconsists of identifying the order of integration of the variables under consideration. Cointegration is determinedthrough the maximum likelihood procedure established by Johansen (1991) in step II. In step III, we perform thecausality tests.3.1 Testing for the Order of IntegrationThe econometric methodology first examines the stationarity properties of univariate time series. This isnecessary to avoid the potential problem of estimating spurious relationships. The Augmented Dickey-Fuller 83

- 3. Journal of Economics and Sustainable Development www.iiste.orgISSN 2222-1700 (Paper) ISSN 2222-2855 (Online)Vol.3, No.7, 2012(ADF) (Dickey & Fuller, 1981) test is used for this purpose. The ADF test is derived from the regressionequation: n (1 − L) X t = π 0 + π 1 X t −1 + ∑ π 2 (1 − L) X t − i + et (5) i =1where L is the lag operator and n is the number of lags on the dependent variable. The null hypothesis is that X isgenerated by a unit root process i.e. π 1 = 0. The ADF test statistic is obtained by dividing the estimate of π 1 byits standard error. If the calculated ADF test statistic is less than the critical value (in absolute terms), the nullhypothesis of a unit root can not be rejected and the series is said to be non-stationary. The order of integrationof X is determined by conducting the ADF test on its first difference. The series will be integrated of order 1 if itsfirst difference does not possess a unit root. The ADF test is carried out by replacing X t with I t , M t and Pt inequation (5) respectively. The results of the unit-root tests are reported in Table 1. The results indicate that in allcases money (M), income (I), and prices (P) are nonstationary at their levels. Therefore to achieve stationarity,the variables must be first-differenced. The ADF statistics are significant only for the first-differenced series.Thus, M, I and P all appear to be I(1).3.2 Testing for CointegrationCointegration test helps to identify the long-run relationship among nonstationary time series. Two or morevariables are said to be cointegrated if they are integrated of the same order. Having determined that thevariables are stationary at first differences, the Johansen cointegration test (1991) is used to examine whether thevariables in question have common trend. The, Johansen procedure assumes that Wt has a vector autoregressive(VAR) representation such that: Wt = δ + Π1Wt −1 + Π 2Wt − 2 + ... + Π kWt − k + ε t (6)where δ is the intercept and εt are the disturbance terms. Equation (7) also yields: ∆Wt = δ + Γ1∆Wt −1 + Γ2 ∆Wt − 2 + ... + Γk ∆Wt − k + ε t (7)where ∆ being the first difference operator, W is the vector of variables, δ is a drift parameter, and Γ1..........Γkare the coefficient matrices. The number of cointegrating vectors is equal to the rank of Γk , denoted by r.Johansen (1991) suggests two test statistics to determine the cointegration rank. The first of these is the tracestatistic: n λ trace ( r ) = −T ˆ ∑ ln (1 − λ ) i (8) i = r +1The second test, known as the maximum eigen value test, is computed as: λ max( r , r +1) ˆ = −T ln(1 − λ ) r +1 (9)where λ i ’s are the ordered (estimated) eigen values of the matrix Γ and T is the available observations. Thenull hypothesis of no cointegration i.e., r = 0 is tested against the alternative of r + 1 cointegrating vectors intrace and maximum eigen values tests.The results of the Johansen’s maximum likelihood method for determining the number of cointegrating vectorsare summarized in Table 2. The number of lags (i.e., two) is chosen by Akaike information criterion (AIC).Concerning the relationship between money-income and money-prices, the null hypothesis of no cointegration isrejected using either statistics. However, the null hypothesis of at most one cointegrating vector cannot berejected in favour of r = 2 . Besides, the null hypothesis of no cointegration is also rejected whereas the nullhypothesis of at most two cointegrating vectors cannot be rejected in favour of r = 3 regarding the relationshipamong money, income, and prices. Thus, the empirical support for one and two cointegrating vectors impliesthat the variables money- income, money-prices and money-income- prices are cointegrated and have a long-runrelationship. 84

- 4. Journal of Economics and Sustainable Development www.iiste.orgISSN 2222-1700 (Paper) ISSN 2222-2855 (Online)Vol.3, No.7, 20123.3 The Causality TestsAs the variables turn out to be cointegrated, the lagged values of the residuals ( η t −1 , η t′−1 , µ t −1 , µ t′−1 ) obtainedfrom the cointegrating regressions are used as error correction terms to amend the standard Granger test. Thebivariate and trivariate tests are specified as generalized extensions of the standard case (Granger, 1969) asfollows: m n (1 − L)Yt = α 0 + ρ1η t −1 + ∑ α i (1 − L)Yt −i + ∑ β j (1 − L) X t − j + ε t (10) i =1 j =1 m n (1 − L) X t = δ 0 + ρ 2η t′−1 + ∑ δ i (1 − L) X t −i + ∑ γ j (1 − L)Yt − j + ωt (11) i =1 j =1 m n p (1 − L)Yt = χ 0 + σ 1 µ t −1 + ∑ χ i (1 − L)Yt −i + ∑ ϕ j (1 − L) X t − j + ∑ λ k (1 − L) Z t − k +u t (12) i =1 j =1 k =1 m n p (1 − L) X t = λ0 + σ 2 µ t′−1 + ∑ λi (1 − L) X t −i + ∑ψ j (1 − L)Yt − j + ∑ φ k (1 − L) Z t − k +v t′ (13) i =1 j =1 k =1For the bivariate as well as the trivariate analysis, the F- value is calculated as: 2 2 ( RUR − R R ) / l F= 2 (14) (1 − RUR ) /( n − q ) 2 2where RUR and RR are obtained from the unrestricted and restricted causality regressions respectively, n is thetotal number of observations, l is the number of lagged terms of the variables which are chosen by Akaike’sInformation Criterion (AIC), and q is the number of parameters estimated in the unrestricted regression.The findings of the bivariate analysis are presented in Table 3(a). These results show bidirectional causalitybetween money and income. Thus, due to the mixed direction of causation found between money and income, itis difficult either to accept or reject the Keynesians or the Monetarists view in Bangladesh. Regarding, money-price relationship, the results suggest a unidirectional causality from money to price supporting the Monetarists.An increase in the money supply increases the price level which does not in turn cause the money supply toincrease. This implies that monetary expansion increase inflation in Bangladesh. Finally, the causal relationshipis also examined on the basis of trivariate causality. In view of the presence of a long-run relationship amongmoney, income, and prices, Table 3(b) shows the causal relationship between money and income conditional onthe presence of prices. Similarly, it shows the causal relationship between money and prices conditional on thepresence of income. The results are similar to those found in the bivariate case, i.e., bidirectional causalitybetween money and income and a unidirectional causality from money to prices conditional on the presence ofprices and income respectively.4. Conclusions and Policy RecommendationsThe objective of this paper has been to examine the causal relationship between money-income and money-prices in Bangladesh over the period 1972/73 to 2009/10. Cointegration analysis suggests a long-run relationshipbetween money and the other variables i.e., income and prices. Based on the Error Correction Model (ECM), thebivariate analysis indicates a bidirectional causality between money and income and a unidirectional causalityfrom money to prices. This implies that an increase in money supply raises the general price level. Therefore, itis money that takes lead in increasing inflation in Bangladesh. The absence of bidirectional causality betweenmoney and prices indicate that money supply can be considered as exogenous which can be taken as an effectivecontrol variable. However, these results may be inaccurate if relevant variable(s) are omitted. The trivariateanalysis also confirms the results obtained from the bivariate case.In view of the causal effect of money over income and prices, a number of policy implications can be inferred.An increase in money supply leads to an increase in income which ultimately increases the demand for money tofinance higher level of economic activity. However, this also increases the price level resulting in inflation. Ahigher level of income without inflation may be achieved if the growth rate of money supply is fixed roughly at a 85

- 5. Journal of Economics and Sustainable Development www.iiste.orgISSN 2222-1700 (Paper) ISSN 2222-2855 (Online)Vol.3, No.7, 2012rate equal to the growth rate of the economy. Thus, if a policy objective is to achieve a high rate of economicgrowth as well as restrain inflation, money supply should be considered as the most suitable target. Moreover, inview of the bidirectional causal relationship between money and income, monetary policy should be devised byconsidering the feedback effects of output on money (Mehrara & Musai, 2011).ReferencesAhmed, M. (2000). Money-Income and Money-Price Causality in Selected SAARC Countries: SomeEconometric Exercises. The Indian Economic Journal, 50 (1), 55-62.Al-Jarrah, M. (1996). Money, Income, and Prices in Saudi Arabia: A Cointegration and Causality Analysis.Pakistan Economic and Social Review, 34 (1), 41-53.Barth, J. & Bannett, J. (1974). The Role of Money in the Canadian Economy: An Empirical Test. CanadianJournal of Economics, 7, 306 -311.Bengali, K., Khan, A. & Sadaqat, M. (1999). Money, Income, Prices and Causality: The Pakistani Experience.Journal of Developing Areas, 33 (4), 503-514.Dickey, D. A. & Fuller, W. A. (1981). Likelihood Ratio Statistics for Autoregressive Time Series with a UnitRoot. Econometrica, 49, 1057-1072.Dyreyes, F., Starleaf, D. & Wang, G. (1980). Test of Direction of Causation between Money and Income in SixCountries. Southern Economic Journal, 47(2), 477-487.Ghazali, F. M. (2008). Linkage between Money and Prices: A Causality Analysis for Malaysia. InternationalBusiness Research, 1 (4), 82-87.Granger, C. W. J. (1969). Investigating Causal Relations by Econometric Models and Cross-spectral Methods.Econometrica, 37, 424-436.Husain, F. & Rashid, A. (2006). Economic Liberalization and the Causal Relations among Money, Income, andPrices: The Case of Pakistan. [Online] Available: http://mpra.ub.uni-muenchen.de/3195/ (June 1, 2012)Johansen, S. (1991). Estimation and Hypothesis Testing for Cointegration Vectors in Gaussian VectorAutoregressive Models. Econometrica, 59, 1551-1580.Jones, J. D. & Sattar, Z. (1988). Money, Inflation, Output and Causality: The Bangladesh Case, 1974-1985. TheBangladesh Development Studies, 16 (1), 73-83.Lee, S. & Li, W. (1983). Money, Income, and Prices and their Lead-Lag Relationship in Singapore. SingaporeEconomic Review, 28, 73-87.Mehrara, M. & Musai, M. (2011). The Dynamic Causal Relationships among Money, Output and Prices in Iran.Asian Journal of Business and Management Sciences, 1 (5), 31 -37.Mishra, P. K., Mishra, U. S. & Mishra, S. K. (2010). Money, Price and Output: A Causality Test for India.International Research Journal of Finance and Economics, 53, 26-36.Shams, M. N. A., Hossain, M. M. & Hassan M. M. (2010). Money, Income, and Causality: The BangladeshExperience. Asian Economic Review, 52 (2), 231-236.Sims, C. (1972). Money, Income and Causality. American Economic Review, 62, 540-552.Williams, W., Goodhart, C., & Gowland, D. (1976). Money, Income, and Causality: The U. K. Experience. TheAmerican Economic Review, 66, 417- 423. 86

- 6. Journal of Economics and Sustainable Development www.iiste.orgISSN 2222-1700 (Paper) ISSN 2222-2855 (Online)Vol.3, No.7, 2012Table 1Unit Root Test with ADF for the period 1972/73 to 2009/10 ADF ADF Variables C First Difference C, T First Difference I 2.184 (2) 2.681 (2)* 2.676 (2) 4.406 (2)*** M 1.392 (2) -3.621 (1)** 2.584 (2) 4.831 (2)*** P 3.008 (1) 2.754 (1)* 1.665 (1) 3.658 (1)**Notes: i) Figures within parentheses indicate lag terms chosen by the Akaike information criterion (AIC); ii)***, ** and * denote rejection of the null hypothesis of unit root at the 1%, 5% and 10% levels respectively; iii)C = constant term included in the unit root test, C,T = constant and trend term included in unit root test. Table 2 Johansen’s Maximum Likelihood Procedure Variables Hypotheses Test Statistics Null Alternative Trace λ - Max r=0 r=1 15.97** 15.97** (M,I) r≤1 r=2 2.65 3.16 r=0 r=1 19.61** 19.26*** (M, P) r≤1 r=2 0.36 0.36 r=0 r =1 29.68** 25.52*** (M, I, P) r≤1 r=2 6.61 6.49 r≤2 r=3 0.12 0.12 Notes: i) r denotes the number of cointegrating vectors; ii) *** and ** denote rejection of the null hypothesis at the 1% and 5% levels respectively. Table 3 (a) Bivariate Analysis of Causal Relationship between Money (M) - Income (I) and Money (M) - Prices (P) for the period 1972/73 to 2009/10 Causation M→ I I→ M M→P P→M F – Values 7.54*** 7.38*** 8.83*** 1.57 Notes: (i) Critical F-Values: 1% = 5.34, 5% = 3.29, 10% = 2.48, df = (2, 32); (ii) *** indicates a significant causal relationship at the 1% level; (iii) M→ I: M causes I; (iv) I→ M: I causes M; (v) M→P: M causes P; (vi) P→M: P causes M. 87

- 7. Journal of Economics and Sustainable Development www.iiste.orgISSN 2222-1700 (Paper) ISSN 2222-2855 (Online)Vol.3, No.7, 2012Table 3 (b)Trivariate Analysis of Causal Relationship among Money (M), Income (I), and Prices (P) for the period 1972/73to 2009/10Causation M(P) → I I(P) → M M(I) → P P(I) → MF – Values 4.57** 3.16* 5.12** 2.26Notes: (i) Critical F-Values: 1% = 5.39, 5% = 3.32, 10% = 2.49 , df = (2, 30); (ii) ** and * indicate a significantcausal relationship at the 5%, and 10% levels respectively; (iii) M(P) → I: M and P jointly cause I after including Pin the unrestricted regression; (iv) I(P) → M: I and P jointly cause M after including P in the unrestricted regression;(v) M(I)→P: M and I jointly cause P after including I in the unrestricted regression; (vi) P(I)→M: P and I jointlycause M after including I in the unrestricted regression. 88

- 8. This academic article was published by The International Institute for Science,Technology and Education (IISTE). The IISTE is a pioneer in the Open AccessPublishing service based in the U.S. and Europe. The aim of the institute isAccelerating Global Knowledge Sharing.More information about the publisher can be found in the IISTE’s homepage:http://www.iiste.orgThe IISTE is currently hosting more than 30 peer-reviewed academic journals andcollaborating with academic institutions around the world. Prospective authors ofIISTE journals can find the submission instruction on the following page:http://www.iiste.org/Journals/The IISTE editorial team promises to the review and publish all the qualifiedsubmissions in a fast manner. All the journals articles are available online to thereaders all over the world without financial, legal, or technical barriers other thanthose inseparable from gaining access to the internet itself. Printed version of thejournals is also available upon request of readers and authors.IISTE Knowledge Sharing PartnersEBSCO, Index Copernicus, Ulrichs Periodicals Directory, JournalTOCS, PKP OpenArchives Harvester, Bielefeld Academic Search Engine, ElektronischeZeitschriftenbibliothek EZB, Open J-Gate, OCLC WorldCat, Universe DigtialLibrary , NewJour, Google Scholar
View More

# Money, income, and prices in bangladesh a cointegration and causality analysis

Internatinoal Journals Call for paper: http://www.iiste.org/Journals