Monetary uncertainty and stock prices The case of Malaysia ABSTRACT

Monetary uncertainty and stock prices: The case of Malaysia ABSTRACT While studies to determine the relationship between money supply and stock prices have long been pursued, little effort has been place in identifying the impact of monetary uncertainty on stock prices. The element of uncertainty is acknowledged to increase perceived riskiness of financial assets and this in turn will have adverse effect on stock prices. As such, monetary uncertainty is hypothesized to have a negative relationship with stock prices. This study is designed to determine the relationship in an emerging market (i.e. Malaysia) by adopting the commonly used econometric tools. The results show that contrary to the hypothesized findings, monetary uncertainty is found to have no influence on stock prices. However, the former is found to have a significant long-run dynamics with the uncertainty in stock prices. To the extent that market is informational efficient, these findings demonstrate evidence that are consistent with Efficient Market Hypothesis (EMH). Author: Tel: Fax: Email: Noor Azuddin Yakob Department of Finance Faculty of Business Management Universiti Kebangsaan Malaysia 43600 Bangi Selangor Darul Ehsan MALAYSIA 603-8929 3681 603-8929 3306 azuddin@pkrisc.cc.ukm.my 1 Monetary uncertainty and stock prices: The case of Malaysia Introduction Studies to examine the relationship between money supply and stock prices are abundant and extensive particularly with reference to advanced markets1 . Existing theoretical frameworks 2 advocate for a direct relationship between the two variables with the causation running uni-directionally from the former to the latter. Nonetheless, empirical findings fail to reach a conclusive agreement3 . Needless to say, the issue continues to receive attention from both finance scholars and researchers4 . Conceptually, stock prices5 are determined by the present value of the discounted expected cash inflows to be received by investors, in this case from dividend payments. As such, the ability of a company to pay dividends will influence the price of its stock6 . An argument that links money supply with stock prices can be traced within the framework of monetary economics 7 . Thorbecke (1995) suggests that “positive monetary shocks increase industry stock returns” which indicates that “expansionary monetary policy exerts real effects by increasing future cashflows or by decreasing the discount factors at which those cashflows are capitalized”. Therefore, monetary policy has been acknowledged to have “real and quantitatively important effects on the economy”. 1 Documented evidence suggests that the issue has been investigated as early as the sixties by Brunner (1961), Friedman (1961), Friedman and Schwartz (1963) and Sprinkel (1964). Studies continue to be expanded in the seventies through the nineties, covering various markets worldwide (see Ghazali and Yakob (1997) for an overview of the related studies). 2 Rozeff (1974) provides a comprehensive description of the two theoretical foundations namely Monetary Portfolio (MP) and Efficient Market (EM). 3 While most findings show consistent results with the proposed theories, bi-directional causality relationship is also detected in certain markets (see Rogalski and Vinso (1977) and Hashemzadeh and Taylor (1988)). 4 The issue benefits from the advancement of the econometric tools with modern researchers adopting various advanced techniques such as the causality and cointegration tests (see, for example, Mak and Cheung (1991)). 5 The most common model of stock valuation is P0 = Σ [ (D0 (1 + g t ) / (1 + rt + ρ)], where: P0 = current price of common stock; D0 = current level of dividend; g t = dividend growth rate; rt = riskless rate of interest: ρ = risk premium; and, t = time, running from 1 to ∝. 6 According to the signalling hypothesis, firms pay high dividend payout in view of positive anticipation of future economic condition and vice versa. 7 The monetarists propose that money supply affects the economy via its transmission effect on interest rate which is a crucial determinant of economic stimulus or impediment. 2 On that note, a direct relationship between monetary policy and stock prices is expected. However, the impact of monetary uncertainty on stock prices is still unclear. According to Friedman (1984), monetary growth variability increases the degree of perceived uncertainty8 . Given the dependence of (financial) asset prices on investor expectations, Boyle (1990) argues that changes in uncertainty regarding money stock will affect these prices 9 . He claims that changes in monetary uncertainty alter the equity risk premium to reflect the additional expected return investors require for bearing the risk of holding stocks 10. Thus, monetary uncertainty is implied to have an adverse relationship with stock prices. In view of that proposition, this paper intends to determine the relationship between monetary uncertainty and stock prices, with a special reference to the Malaysian stock market. The objectives of this paper are twofold – first, to test for the existence of a relationship between the uncertainty associated with the variability of the past values of money growth and the stock prices. Second, to detect for the presence of long-run dynamics between uncertainty in monetary aggregate and stock prices. The next section of the paper will discuss the background of the study followed by the presentation of data, methodology and findings. A brief discussion of the findings and its conclusion will end the paper. Background of the Study Within the context of Malaysia, one of the comprehensive studies to determine the relationship between money supply and stock prices is performed by Ghazali and Yakob (1997). They employ the Vector Autoregression (VAR) methodology in identifying the relationship between the two variables by incorporating four variables, namely stock prices, money, income and interest rate, into the VAR system. The result of the Granger causality test shows the presence of a significant uni-directional relationship running from money supply to stock prices. The impulse response function (IRF) indicates that 8 9 Indeed, uncertainty is an important consideration in finance since it is normally being associated to risk. Through his Proposition 5 (on page 1050), Boyle states that “a rise in monetary uncertainty, as measured by coefficient of variation of the money growth rate, increases the equity premium”. 3 stock prices response positively following monetary expansion and the impact peaks after seven month from the initial shock. With the belief that stock prices reflect real economic performance, the authors conclude that their finding is consistent with the long-run effect of money on the real sector. Similar result is also reported by Habibullah (1998). Using the cointegration and error correction model (ECM), he investigates the relationship between money supply and stock prices in the KLSE to determine the level of market informational efficiency. He discovers that the two variables – stock prices and money, are nonstationary in their level form but are cointegrated in the long-run with the presence of error correction representation. Most importantly, he finds from the error -correction model that money supply, represented by M3, Granger cause stock prices but not otherwise. He concludes that his finding is inconsistent with the efficient market hypothesis (EMH) since market participants will be able to predict stock prices in the market using information on broad money supply, M3, as a trading rule to earn excess returns. The empirical results that suggest the importance of monetary policy in influencing stock prices in Malaysia is not surprising since the Government has long been known to pursue monetary policy in attaining economic goals, one of which is price stability in the country. The central bank, Bank Negara Malaysia (BNM), is entrusted with the responsibility of formulating and implementing the country’s monetary policy. One of the earlier strategies has been targeting the monetary aggregates which saw the emphasis given on M1 and eventually M3 to ensure sufficient liquidity in the system to meet the demand of the economy 11 . The success of the monetary targeting strategy is evident in its ability to spur economic growth until mid-1990’s. The Malaysia economy continues to record unprecedented growth rates from 1988 until then. 10 Given the stock valuation model presented in note 5, increase in risk premium will result in decline in stock price. 11 Chapter 4 of the “Central Bank and the Financial System in Malaysia: A Decade of Change 1989-1999” outlines the conduct of monetary policy pursued by Bank Negara Malaysia. 4 However, the large capital influx into the country since early 1990’s has caused considerable instability in the relationship between monetary aggregates and nominal GDP. Output growth is seen to cause monetary growth and not vice versa, forcing the central bank to shift its strategy from monetary aggregate targeting to interest rate targeting. Nevertheless, BNM still monitors the monetary aggregates very closely despite suggestions that they become unreliable indicators of economic activity. During this period, monetary velocities – the ratios of nominal GDP to various monetary aggregates, are reported to frequently depart from the historical patterns 12 . This suggests the instability of monetary aggregates over those periods which is believed to have an affect on the level of economic activity. Considering this development, it is deemed appropriate to pursue empirical evidence to establish the relationship between monetary uncertainty and stock price behaviors. On that note, this paper is designed to test two hypotheses, namely (i) monetary uncertainty is negatively related to stock prices 13 , and (ii) stock price uncertainty is directly related to monetary uncertainty14 . The major contribution of this paper is in its attempt to elucidate the issue of monetary uncertainty and its effect on the Malaysian stock market. The findings will be of use for market participants and regulators alike. Data, Methodology and Findings This study employs monthly data running from 1989:01 to 2001:03. Using M1 15 to represent monetary aggregate and Composite Index16 as a proxy of stock prices, the month-to-month (i.e. January-to-January) rate of change is computed to generate new series that represent changes in M1 and KLCI. The monetary variability is measured by calculating the standard deviation 17 of changes in the M1 series over one year period 18 . 12 13 See Bank Negara Malaysia (1999), page 141. A contention that echoes the proposition made by Boyle (1990). 14 This assumption is drawn upon the contribution of monetary targeting strategy on the country’s economic well-being in the past. 15 Friedman (1984) recommends the use of M1 as it is the closest approximation of monetary aggregate. 16 The Composite Index (KLCI) is the key indicator of the performance of the Kuala Lumpur Stock Exchange (KLSE) since it comprises of 100 dominant counters in the market. 17 Standard deviation is a common measurement of risk since it reflects the variability from the mean. 18 In an earlier study, Ghazali and Yakob (1997) discover that stock prices respond positively following monetary shock and the impact last for almost a year. 5 To measure the past history of monetary instability, a series of one-year moving average of the standard deviation is constructed. Similar process is also performed on the KLCI series to accomplish the second objective of the study. Overall, three series are generated - namely moving average of standard deviation for monetary growth, stock price growth and moving average of standard deviation for stock price growth, with 113 observations each. Before choosing the appropriate econometric approach to analyze the data, the stationarity of the time series is first determined. A time series is said to be stationary in level form, I(0), if it does not contain a unit root. A time series that contains a unit root and requires first-differencing in order to obtain stationarity is said to be first -order integrated, I(1). In general, a time series is said to be integrated in order d if it achieves stationarity after being differenced d times. The presence of unit roots and stationarity of time series can be tested using the Augmented Dickey-Fuller (ADF) test 19 . The ADF test is conducted from estimating the following ordinary least square (OLS) equation: ∆Xt = ρ 0 + ρ 1 Xt-1 + Σi=1m ϕi ∆Xt-i + ε t where, ∆ is the first-differenced operator, ε t is a white noise error term, and m is chosen such that the residuals are serially uncorrelated. The null hypothesis of nonstationary is rejected if ρ1 < 0 and statistically significant. The results from the ADF tests are presented in Table 1. They indicate that the three series are nonstationary at level form but integrated of first-order, I(1). Since the series are integrated of the same level, it suggests the presence of cointegration relationships that validates the use of cointegration analysis to ascertain the long-run dynamic between the series. 19 The econometric procedures used in this study are based on the specification of Eview software. 6 Table 1. ADF tests for the presence of unit root. Variable Statistic for level Statistic for first-differences MONMASD STKGROW STKMASD -1.9243 -1.9704 -1.9681 -3.7119* -5.5009* -3.2719** Note: MONMASD = moving average of standard deviation for money growth. STKGROW = stock price growth. STKMASD = moving average of standard deviation for stock price growth. * = significant at one percent level. ** = significant at five percent level. Conventionally, when two time series are nonstationary in level but some linear combination of them is stationary, then the series are said to be cointegrated. A cointegrated series suggest that they do not move far away from one another. In this study, the Johansen’s (1991) cointegration test is adopted to determine whether the linear combination of the series possesses a long-run equilibrium relationship. The test applies the maximum likelihood procedure to determine the presence of cointegrating vectors in nonstationary time series. It considers a vector autoregression (VAR) equation of order p such as: yt = A1 yt-1 + …… + Ap yt-p + Bxt + ε t where y is a k-vector of non-stationary, I(1) variables, x is a d vector of deterministic t t variables and ε t is a vector of innovations. The system is in reduced form with each variable in yt regressed on only lagged values of both itself and all other variables in the system. The equation can be reformulated into a vector error correction model (VECM) form: ∆yt = Πyt-1 + Σ i=1 p-1 Γi ∆yt-i + Bxt + ε t where, Π = Σ i=1 p Ai – I, and Γi = -Σ j=i+1 p Aj 7 This way of specifying the system contains information on both short- and long-run adjustments to changes in yt , via the estimates of Γi and Π respectively. If the coefficient matrix Π has reduced rank r < k, then there exist k x r matrices α and β each with rank r such that Π = αβ’ and β’yt is stationary. r is the number of cointegrating relations and each column of β is the cointegrating vector. The elements of α are known as the adjustment parameters in the vector error correction model. In general, the Johansen’s method estimates the Π matrix in an unrestricted form and then test whether the restrictions implied by the reduced rank of Π can be rejected. In this study, the series are assumed to have linear trends but the cointegrating equations have only intercepts Πyt-1 + Bxt = α(β’yt + ρ 0 )α⊥γ0. The results from cointegration analysis are presented in Table 2. Table 2. Res ults from Johansen’s cointegration analysis. Panel A. Series: MONMASD STKGROW Likelihood Ratio 7.606680 3.206030 5 Percent 1 Percent Hypothesized Critical Value Critical Value No. of CE(s) 15.41 3.76 20.04 6.65 None At most 1 Eigenvalue 0.039928 0.029249 Note: L.R. rejects any cointegration at 5% significance level. Panel B. Series: MONMASD STKMASD 5 Percent 1 Percent Hypothesized Critical Value Critical Value No. of CE(s) 15.41 3.76 20.04 6.65 None* At most 1** Eigenvalue Likelihood Ratio 0.081992 0.064923 16.48893 7.249614 Note: L.R. test indicates 2 cointegrating equations at 5% significance level. *(**) denotes rejection of the hypothesis at 5% (1%) significance level. 8 Based on the results presented in Panel A of Table 2, it is found that the two series – moving average of standard deviation for monetary growth and stock price growth, are not cointegrated in the long-run despite both being integrated in the first-order. Nonetheless, Panel B shows that the series of moving average of standard deviation for both monetary and stock price growth do indeed possess a long-run equilibrium relationship between the themselves20 . Figure 1 shows the movement of the two series over the period understudy. Given the vector error correction mechanism that is embedded in the Johansen’s procedure, the deviation from long-run equilibrium is corrected through a series of partial short-run adjustments. The VECM specification restricts the long-run behavior of the variables in the system to converge to their long-run relationship while allowing a wide range of short -run dynamics. Table 3 presents the results of VECM. The coefficients for the error correction terms (ECM) which represent the speed of adjustment are found to be significantly different from zero (in this case, the coefficients are negative albeit small ones). This indicates that the two series simultaneously correcting for the disequilibrium resulting from momentary deviation from their long-run equilibrium path (at different rates), thus explaining the cointegration relationship between the series. The significant negative coefficients suggest that the current period innovations respond to the previous period’s deviation from equilibrium. However, the results from the short-run dynamics fail to suggest causal relationship between the series. None of the lagged coefficients for the series are found to be significant in influencing one another. This suggests the absence of causation between the series. This finding, however, contradicts the result from Granger (1969) causality test21 which detects the presence of bi-directional relationship. The result is shown in Table 4. 20 The finding is consistent with the result of correlation analysis which shows that moving average of standard deviation for both stock price and monetary growth are highly positively correlated (0.91) and significant at one percent level. 21 A direct test of Granger causality between two variables, X and Y, is performed by estimating the following equations: ∆Xt = α0 + Σ i=1 kαi ∆Xt-i + Σ j=1 n βj ∆Yt-j + µ1t ∆ Yt = α0 + Σ i=1 kαi ∆Xt-i + Σ j=1 n βj ∆Yt-j + µ2t (1) (2) 9 Table 3. VECM estimates of the adjustment coefficients ∆MONMASD ∆STKMASD Dependent Variable ECMt-1 ∆MONMASD t-1 ∆MONMASD t-2 ∆STKMASDt-1 ∆STKMASDt-2 constant R2 SEE -0.0052 (-3.1635)* 1.4490 (19.1639)* -0.6293 (-9.0872)* 0.0291 (1.4157) 0.0103 (0.4562) -1.74E-06 (-0.02767) -0.0012 (-1.7381)*** 0.3630 (1.1077) -0.4064 (-1.3536) 1.4282 (16.0383)* -0.4802 (-4.8943)* 0.0001 (0.4473) 0.9698 0.0006 0.9530 0.0008 Note: * significant at one percent level. ** significant at five percent level *** significant ten percent level Table 4. Results of the Granger causality test Panel A: Causality test. F-stat. H0 : STKMASD does not Granger cause MONMASD H0 : MONMASD does not Granger cause STKMASD 17.4635* 4.7429** The definition of causality relies on the predictability of the time series whereby if σ2 (XX.Y) < σ2(X X), then Y is said to cause X. The term σ2 (XX.Y) is the prediction error variance of X derived from the information set that includes past values of X and Y. The term σ2 (XX) is the variance of the prediction error of X based on information contained only in the past values of X. Similarly, if σ2 (Y Y.X) < σ2 (Y Y), then X is said to cause Y. Bi-directional causality is said to occur when the above outcomes occur simultaneously. However, if σ2 (X X) < σ2 (XX.Y) and σ2 (YY) < σ2 (YY.X), then the two series are independent of each other. 10 Panel B: Estimation of causality test. Dependent: Independent: ∆STKMASDt-1 ∆STKMASDt-2 ∆MONMASDt-1 ∆MONMASDt-2 constant R2 SEE Note: * significant at one percent level. ** significant at five percent level *** significant ten percent level ∆STKMASD ∆MONMASD 1.4702* -0.5517* 0.5801*** -0.5485*** 0.0153*** 0.0467** -0.0197 1.5410* -0.6890* 0.0012 0.9516 0.2831 0.9669 0.0674 The result from Panel A suggests the presence of bi-directional relationships between the two series due to the rejection of both null hypotheses. This means that the uncertainty in the stock price is influenced by the uncertainty in the monetary policy and vice versa. Judging by the result presented in Panel B, the lagged values (up until t-2) of moving average of standard deviation for monetary growth are found to influence the present value of the moving average of standard deviation for stock prices. On the other hand, only the t-1 value of the moving average of standard deviation for stock prices is found to influence the present value of moving average of standard deviation for monetary growth. This may indicate that the uncertainty in monetary growth has a more prevalent effect on the uncertainty in stock prices than vice versa 22 . 22 The interpretation of this finding has to be taken with caution since Granger (1988) points out that when the series are cointegrated, the direct test of causality will result in invalid causal inferences due to the exclusion of the error correction terms. 11 Discussion and conclusion The paper aims at accomplishing two objectives – first, to test for the existence of a relationship b etween the uncertainty associated with the variability of (past values of) money growth and the (contemporaneous) stock prices. Through the analysis that has been conducted, it is found that the variability of the past values of money growth has no significant long-run relationship with stock prices, as evident by the lack of cointegration between the moving average of standard deviation for monetary growth and stock prices 23. This finding rejects the proposition made by Boyle (1990) who argues that changes in uncertainty regarding money stock will affect stock prices thus implying a negative relationship. But such discovery on the Malaysian stock market is nonetheless consistent with the concept of efficient market since past information does not seem to influence the contemporary stock prices. As such, it suggests that market has already considered past information of market uncertainty in determining stock prices. Secondly, the study intends to detect for the presence of long-run dynamics between uncertain ty in monetary aggregate and stock prices. Using Johansen (1991) cointegration analysis, the long-run relationship between the uncertainty of the two variables is detected. This finding conforms to the suggestion that monetary policy has real and quantitat ively important effects on the economy since the uncertainty in the latter is reflected in the uncertainty in stock prices which is a proxy of economic prosperity. However, the extent of causal relationship between the two variables is still vague despite evidence from Granger causality test that suggests bi-directional relationships. Being 23 The result from correlation analysis shows that the two series are lowly (0.08) and not significantly correlated while the result from Granger causality test suggests that the two series are independent. 12 cointegrated and represented by error correcting mechanism, the causal relationship between the two variables cannot be determined by the direct test of causality as cautioned by Granger (1988). Yet, the result of causality test drawn from VECM fails to place support on the bi-directional relationship, indicating instead that the two variables are independent. This finding, nonetheless, would still be compatible with the premise of EMH since stock prices respond accordingly to monetary uncertainty. Perhaps, future research can be geared towards clearing the air pertaining to the causal relationship between uncertainty in both monetary policy and stock prices. This paper provides evidence that (i) monetary uncertainty has no significant relationship with the contemporaneous stock prices, and (ii) the uncertainty in monetary policy is cointegrated in the long-run with the uncertainty in stock prices, at least within the context of Malaysia. Given the importance of the issue of uncertainty in financial markets, this paper virtually contributes to beef up the literature on the subject particularly with reference to emerging markets like Malaysia. It is hope that this effort will instigate further investigations on the topic in view of its great importance to market participants and regulators. 13 References Bank Negara Malaysia. (1999), “The Central Bank and the Financial System in Malaysia: A Decade of Change 1989-1999,” Kuala Lumpur. Boyle, G.W. (1990), “Money Demand and the Stock Market in a General Equilibrium Model with Variable Velocity,” Journal of Political Economy 98, pp. 1039-1053. Brunner, K. (1961), “Some Major Problems in Monetary Theory,” American Economic Review Proceedings, May. 47-56. Eviews 3: User’s Guide 2nd Edition 1994-1998. Quantitative Micro Software, Irvine, CA. Friedman, M. (1961), “The Lag in Effect of Monetary Policy,” Journal of Political Economy, pp.447-66. Friedman, M. & Schwartz, A. (1963), “Money and Business Cycle,” Review of Economics and Statistics Supplement, pp.32-64. Friedman, M. (1983), Monetary Variability: United States and Japan,” Journal of Money, Credit and Banking 15 (3). pp. 339-343. Friedman, M. (1984), “Lessons from the 1979-82 Monetary Policy Experiment,” AEA Papers and Proceedings, May 1984. pp. 397-400. Friedman, M. (1988), “Money and the Stock Market,” Journal of Political Economy 96, pp. 221-45. Ghazali, N. A. & Yakob, N. A. (1997), “Money Supply and Stock Prices: The C of ase Malaysia,” Proceeding Seminar Antar Bangsa Managing Growth and Changes, Bengkulu, Indonesia, pp. 605-617. Granger, C. W. J. (1969), “Investigating Causal Relations by Econometric Models and Cross Spectral Methods,” Econometrica 37, pp. 424-38. Granger, C. W. J. (1988), “Some Recent Developments in the Concept of Causality,” Journal of Econometrics 36 , pp. 199-211. Habibullah, M.S. (1998), “The Relationship between Broad Money and Stock Prices in Malaysia: An Error Correction Approach,” Jurnal Ek onomi Malaysia 32, pp. 5173. Hashemzadeh, N. & Taylor, P. (1988), “Stock Prices, Money Supply, and Interest Rates: The Question of Causality,” Applied Economics 20, pp. 1603-1611. 14 Johansen, S. (1991), “Estimation and Hypothesis Testing of Cointegration Vectors in Gaussian Vector Autoregressive Models,” Econometrica 59, pp. 1551-1580. Mak, B. S. C. & Cheung, D. W. W. (1991), Causality Tests of the United States Weekly Money Supply and Asian-Pacific Stock Markets. Asia Pacific Journal of Management 9 (2), pp. 252-260. Rogalski, R.J. & Vinso, J.D. (1977), “Stock Returns, Money Supply and the Direction of Causality,” The Journal of Finance 32, pp. 1017-1030. Ross, S. A., Westerfield, R. W. & Jordan, B. D. (1995), “Fundamentals of Corporate Finance, Third Edition. (Chicago, Illinois: Richard Irwin, Inc). Rozeff, M.S. (1974), “Money and Stock Prices: Market Efficiency and the Lagged Effect of Monetary Policy,” Journal of Financial Economics 1, pp. 245-302. Sprinkel, B.W. (1964), “Money and Stock Prices”. (H omewood, Illinois: Richard Irwin, Inc). Thorbecke, W. (1995), “On Stock Market Returns and Monetary Policy,” Working Paper No. 139, April 1995. The Jerome Levy Economics Institute of Bard College. 15