Granger

Money supply and agricultural prices 193

MONEY SUPPLY AND AGRICULTURAL PRICESA CAUSALITY ANALYSIS FOR PAKISTAN ECONOMY (QUARTERLY DATA ANALYSIS) Qazi Muhammad Adnan Hye*

ABSTRACT The present study was conducted at Applied Economics Research Centre, University of Karachi, Pakistan during 2008. This study empirically investigates the cointegration and causal relationship between agricultural prices and money supply during a period 1971 to 2007. For empirical analysis JJ cointegration for long run and Toda Yamamoto Modified Granger Causality Test for causal association were used. The results show that a long run relationship existed between both variables and long run elasticity of agricultural prices with respect to money supply is 0.79. The causality analysis indicates unidirectional causality from money supply to agricultural prices. Thus money supply is not neutral in determining agriculture prices in an agricultural based economy of Pakistan. KEYWORDS:

Money supply; agricultural prices; economy; Pakistan.

INTRODUCTION Agriculture sector plays a vital role in the process of economic growth of the country. It contributed 21.5 percent to GDP during 2008. For desirable economic growth sustainability in agricultural growth is necessary (6). Agricultural prices are important for maintaining agricultural growth, farmer’s living standard and investment decisions. Thus factors that influence the agricultural prices is a fundamental issue. Conventional agricultural economies examine that agricultural prices are determined by the interaction of supply and demand forces. The latest studies on agricultural economies examined that macroeconomics, particularly monetary factors, affect the agricultural prices. Tweeten (12) finds that monetary shocks have a little effect on agricultural prices. David et.al. (2) empirically indicates unidirectional causality from money supply to agricultural prices in Brazilian data. Frankel (5) argues that monetary policy has significant effects on *M. Phil Student, Applied Economics Research Centre, University of Karachi, Karachi, Pakistan.

J. Agric. Res., 2009, 47(2)

194 Q. M. A. Hye

agricultural prices as there is flexible compared to other goods prices. Devadoss et.al (3) support the hypothesis that agricultural prices faster respond than manufacturing product prices to change money supply in U.S.A. Saghaian, et.al (10) empirically demonstrates that in long run money neutrality does not hold in determination of agricultural prices. Peng et.al (8) observed monetary variables impact on food prices in China. Asfaha and Jooste (1) reject the money neutrality hypothesis and also explain that in case monetary shock occurs, agricultural sector will have to bear the burden of adjustment because of increase in farmers’ financial vulnerability. Most of the empirical research regarding monetary shocks impact on agricultural prices was conducted on well developed market economies. Comparing with these markets, Pakistan’s agricultural commodity markets are not well developed. But due to financial reforms in Pakistan, it is anticipated that monetary policy plays more vigorous role, affecting agricultural prices in Pakistan. Hence, it is important to verify the monetary impacts on Pakistan agricultural prices through quantitative methods. The present study was conducted to explore causal relationship between money supply and agricultural prices in Pakistan by employing JJ cointegration for long run relationship and causal relationship determined through Toda and Yamamoto (11) modified granger causality test. METHODOLOGY The study covers quarterly data from 1971 to 2007 phase. Broad money supply (M2) (measured in million of rupees) was taken from international financial statistics and agricultural price index (quarterly) developed by author. Both series were transformed in natural logarithm for econometric analysis.

J. Agric. Res., 2009, 47(2)

Money supply and agricultural prices 195

This empirical work uses Phillips and Perron (9) unit root test to determine time series properties. Phillips and Perron (PP) test proposes an alternative (non-parametric) method of controlling for serial correlation while testing unit root of time series data. PP method estimates non-augmented Dickey Fuller equation (1). The test detects the presence of a unit root in a series, say Xt by estimating as

∆X

t

= α + ρ X

t −1

+ ε t − − − − − − − (1 )

PP test estimates the modified t-value associated with estimated coefficient of ρ so that serial correlation does not influence the asymptotic distribution of test statistic. PP test is based on following statistic:-

γ ~ tρ = t ρ  0  f0

  

1

2

−

T ( f 0 − γ 0 )( se( ρ~ )) 1

2 f0 2 s

− − − − − − − (2)

where ρ~ is the estimate and t ρ the t-ratio of ρ , se( ρ~ ) is coefficient standard error and s is the standard error of test regression. In addition, y0 is a consistent estimate of the error variance (in eq.1) which was calculated as.

γ0 =

(T − k ) s 2 − − − − − − − (3) T

where k is the number of regressors and T tabulated value. Remaining term, f 0 , is an estimator of residual spectrum at frequency zero. The series is stationary if ρ is negative and significant. JJ cointegration test If hypothesis of non-stationary is established for the underlying variables, it is desirable and important that time series data are examined for cointegration. Engle and Granger (4) approach for cointegration is simple and popular for its certain agreeable attributes. This study used maximum likelihood procedure of Johansen (7) because this is based on well-established likelihood ratio principle. The advantage of Johansen’s procedure is that several cointegration relationships can be estimated and it fully captures the underlying time series properties of the data. Johansen’s method tests the J. Agric. Res., 2009, 47(2)

196 Q. M. A. Hye

restrictions imposed by cointegration on unrestricted VAR involving the series. Consider a VAR of order p.

y t = A1 y t −1 + ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ + A ρ y t − ρ + Bx t + ε t − − − − − − − ( 4 ) where yt is a k-vector of non-stationary I(1) variables ,xt is a d-vector of deterministic variables, and Єt is a vector of innovations. We can write the VAR as ∆ y

t

= Π y

t−1

+

p −1

∑

i=1

where

Π = Γ

i

= −

Γi∆ y

∑ ∑

i

t

− 1 + Bx

t

+ ε

t

− − − − − − − (5 )

A i − I , i = 1 , ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅, ρ

j = r + 1

A

i

− − − − − − − (6 )

Granger’s representation theorem asserts that if coefficient matrix П has reduced rank r < k, then there exist k × r matrices α and β each with rank r such that П = α and yt stationary is the number of cointegrating relations (the cointegrating rank) and each column of β is the cointegrating vector. The elements of α are known as the adjustment parameters in vector error correction model. Johansen’s method is to estimate the П-matrix in an unrestricted form, and then test whether we can reject the restrictions implied by reduced rank of П. Johansen’s method uses two test statistics for the number of cointegrating vectors: the trace test and maximum eigenvalue ( λ max ) test. The ( λtrace ) statistic tests H0, that the number of distinct cointegrating vectors is less than or equal to r against a general alternative. The second statistic tests H0 that the number of cointegrating vectors is r against the alternative of r+1 cointegrating vectors. Toda and Yamamoto modified Granger causality test To establish a causal relationship between monetary expansion, and agricultural prices, we also employed a modified version of Granger causality test, which is robust for cointegration features of the process. This procedure was suggested by Toda and Yamamoto (11) to overcome problem of invalid asymptotic critical values when causality tests are performed in presence of non-stationary series. This procedure essentially suggests the determination of d-max, i.e. maximal order of integration of J. Agric. Res., 2009, 47(2)

Money supply and agricultural prices 197

series in the model, and to intentionally over fit the causality test underlying model with additional d-max lags- so that VAR order is now ρ = k + d , where k is the optimal lag order. Toda and Yamamoto (11) augmented Granger causality test was obtained in present study by estimating a two equation system using the seemingly unrelated regressions (SUR) techniques. Therefore, model can be specified as follow:k +d

k +d

Ln( AP ) = ∑ α 1i Ln( AP ) t −i + ∑ β 1i Ln ( MS ) t −i + µ1t − − − − − (7) i =1 k +d

i =1 k +d

i =1

i =1

Ln ( MS ) = ∑ α 21i Ln( MS ) t −i + ∑ β 2i Ln( AP ) t −i + µ 2t − − − − − (8) where Ln(MS) and Ln(AP) are respectively the natural logarithms of money supply and natural logarithms of agricultural prices. ‘ k ’ is the optimal lag

order, d is the maximal order of integration of series in system and µ1 and µ 2 are error terms that are assumed to be white noise. Conventional Wald tests were then applied to first k coefficient matrices using standard χ 2 -statistics. The main hypothesis set can be drawn as in equation (7), money supply “Granger-causes” agricultural prices if it is not true that β1i = 0∀i ≤ k ; in equation (8), agricultural prices “Granger-causes” money supply if it is no true that β 2i = 0∀i ≤ k . RESULTS AND DISCUSSION The results about the order of integration of series, derived from Phillips and Perron (PP) unit root test (Table-1.) indicate that natural logarithms of agricultural prices and natural logarithms of money supply are not stationary in their levels. But stationary after first difference of both variables, the null hypothesis of no unit root is rejected at 0.01 significance level. Table 2.

Phillips-Perron unit root test.

Variable I(0) Ln(AP) -2.68 Ln(MS) -2.92 *Significant at P=0.01 Optimal lag based on AIC.

I(1) -17.01* -16.31*

J. Agric. Res., 2009, 47(2)

198 Q. M. A. Hye

After both series were found to be integrated of order one, cointegration hypothesis between variables is examined by Johansen cointegration test (Table 3). Using the trace statistic test, null hypothesis of no cointegrating vector (R=0) can be rejected at 5 percent level of significance, while null hypothesis (R ≤ 1 ) cannot be rejected, the results indicating one cointegrating vector. Therefore, results support the hypothesis of cointegration between the agricultural prices and money supply. At the bottom of Table-3, estimated cointegrating vector shows positive long run elasticity (equal to 0.79) of agricultural prices with respect to money supply. Table 3.

Johansen cointegration test.

Null hypothesis Trace statistic 5 percent critical value R=0 64.56 20.26 8.13 9.17 R≤1 Cointegration equation { normalized to Ln(AP)} Ln(AP) = 2.17 + 0.79 Ln(MS) Table 4. Test for Granger-causality applying Toda and Yamamoto modified wald test. Null hypothesis

χ2

P-Value

MS does not Granger Cause AP AP does not Granger Cause MS

10.64 8.53

0.05 0.13

The underlying model for the two-equation system is a SUR model; the lag order (k) is 1 based on AIC.

The results of causality wald test, obtained from SUR estimations are illustrated in Table- 4. Null hypothesis that money supply does not cause agricultural prices can be rejected at 10 percent level of significance. On the other hand, hypothesis that agricultural prices do not cause money supply, cannot be rejected at 10 percent level of significance. Thus, there is unidirectional causality from money supply to agricultural prices in case of Pakistan’s economy. CONCLUSION The study concludes; first, there is one cointegrated vector between money supply and agricultural prices. Secondly, estimated cointegrated vector indicates 0.79 long run elasticity of agricultural prices with respect to money supply. Thirdly, causal analysis demonstrates that there is unidirectional causality from money supply to agricultural prices. Present findings guide to J. Agric. Res., 2009, 47(2)

Money supply and agricultural prices 199

policy implication that in long run money supply positively causes agricultural prices. This implies that loose monetary policy can be used to boost the agricultural prices which leads to an increase in farmer’s income or to use tight monetary policy in order to control agricultural prices for easing consumers. This study also recommends that closed managed coordination is necessary between monetary, agricultural policy makers and price control authority to achieve desired goals to facilitate consumers or farmers. REFERENCES 1. 2. 3. 4. 5. 6. 7. 8.

9. 10. 11. 12.

Asfaha, T.A. and A. Jooste. 2007. The effect of monetary changes on the relative agricultural prices. Agrekon. 46(4): 460-474. David, A. B, R. C. Barnett. and R. L. Thompson 1983. The money supply and nominal agricultural prices. Amer. J. Agric. Econ. 65(2):303307. Devadoss, S. and W.H. Meyers.1987. Relative prices and money. further results for the United States. Amer. J. Agric. Econ. 69: 838-842. Engle, R. F. and C.W. J. Granger 1987. Co-integration and error correction: Representation, estimation, and testing. Econometrica. 55(2):251-276. Frankel, J.A. 1986: Expectations and commodity prices dynamic: The Overshooting Model. Amer. J. Agric. Econ. 67: 344-348. Hye, Q.M.A. 2009, Agriculture on the Road to Industrialization and Sustainable Economic Growth: An Empirical Investigation For Pakistan Economy. Paper Presented in 5th ISOSS Conference, Lahore. Johansen, S, 1991. Estimation and Hypothesis Testing of Cointegration Vectors in Gussian Vectors Autoregressive Models. Econometrica. 59: 1551-1580. Peng, X, M.A Marchant and M.R Reed. 2004. Identifying Monetary Impacts on Food Prices in China: A VEC Model Approach. Paper Presented in the American Economics Association Annual Meeting. Denver, Colorado. August 1-4, 2004. Phillips, P.C. and P. Perron. 1988, Testing for a Unit Root in a Time Series Regression. Biometrica. 75: 335-346. Saghaian, S. H., M. R. Reed and M. A. Marchant. 2002. Monetary Impacts and Overshooting of Agricultural Prices in an Open Economy. Amer. J. Agric. Econ. 84: 90-103. Toda, H. Y. and T. Yamamoto 1995 Statistical Inference in vector autoregressions with possibly integrated processes. J Econometrics. 66: 225-250. Tweeten, L.G. 1980. Macroeconomics in crisis: Agriculture in an underachieving economy. Amer. J. Agric. Econ. 62: 853-865. J. Agric. Res., 2009, 47(2)