ESTIMATION OF THE MONEY DEMAND FUNCTION IN A HETEROGENEOUS PANEL FOR SELECTED ASIAN COUNTRIES

Transcription

1 Indian Journal of Economics & Business, Vol. 12, No. 1, (2013) : ESTIMATION OF THE MONEY DEMAND FUNCTION IN A HETEROGENEOUS PANEL FOR SELECTED ASIAN COUNTRIES MOHAMMED NUR HUSSAIN * AND ALBERT WIJEWEERA ** Abstract Despite the large number and variety of studies examining the money demand function, only a few have a utilized panel data approach in estimating the open money demand function. This paper uses the Pedroni panel cointegration approach along with a fixed-effect estimation method to estimate the open money demand function for 12 selected countries over the period of Cointegration test results indicate that money demand as measured by M1 and M2 shares a long run equilibrium relationship with key explanatory variables. As anticipated, we find that real GDP exerts significantly positive impacts on the demand for real money balances. Among other variables, the interest rate and the exchange rate are found to influence a negative impact on money demand of M1 in the 12 selected countries. We further find that, compared with M2, the cointegration relationship is much stronger for the M1 money demand function. Keywords: Money demand function; Panel cointegration; Panel estimation methods Field: Economics 1. INTRODUCTION Estimation of the money demand function is a critical task for both researchers and policy makers when considering the effects of monetary policy on the economy. Monetary policy is often used to pursue macroeconomic goals of ensuring price stability, economic growth and low unemployment rates. In attempting to achieve these three goals central banks adopt various measures of monetary policy. Monetary policy in general achieves its ultimate target setting an intermediate target. For most countries, operational instrument is the level of short term interest rates (Miles and Scott, 2005). Interest rate is the most important variable in a * Lecturer, Faculty of Business and Design, Swinburne University of Technology Sarawak Campus, Jalan Simpang Tiga Kuching Sarawak, Malaysia ** Senior Lecturer, Southern Cross Business School, Southern Cross University, Tweed Heads, NSW, Australia

2 24 Mohammed Nur Hussain and Albert Wijeweera money demand function. A clear understanding of the money demand function is essential to comment on the efficacy of the monetary policy of a certain economy. In general, there are two types of money demand functions in the literature - conventional money demand function and open economy money demand function. Over the last two decades, the conventional money demand function has been found more useful in estimating the domestic money demand function because it is related to only two key variables, namely income and interest rate. Although the conventional money demand function has been widely used, its relevance to the current global economic environment is quite limited given the substantially increased interrelations among countries. Today most countries are open to foreign trade as well as foreign finance. Given the dramatic policy changes over the last two decades due to liberalization and globalization, it is imperative to estimate the open money demand function which controls for global impact factor by including exchange rate variable in addition to income and interest rates that are found in closed economy conventional type money demand models. Most of the existing literature is based on the estimation of the domestic money demand function for developed countries. (See for example Dreger and Roffia 2007, Pradhan and Subramanian 2003, Calza and Sousa 2003, Bahmani-Oskooee and Chomisisengphet 2002, Hamori and Tokihisa 2001). Very few studies deal with an open economy money demand function and this is particularly the case for developing countries in Asia. Most of the Asian countries liberalized their financial markets after 1990 under the guidance of IMF and its structural adjustment program. Financial liberalization has encompassed interest rate deregulation, introduction of competition among banks, removal of restrictions on capital outflow, lower transaction costs, widespread use of automatic teller machines, and increased use of credit cards. Given these widespread changes it is worthwhile to estimate an open economy money demand function for the selected Asian countries. Impacts of the structural adjustment program and financial liberalization policy have been quite impressive for some of the countries that are subject to this study. For instance, countries such as Singapore, Hong Kong and South Korea have experienced a rapid growth in their per capita income during period. Even those countries such as Bangladesh, India, Nepal and Pakistan that were late in liberalizing the financial sector have registered around two fold increase in per capita income. This trend was temporarily disrupted by the 1997 financial crisis, but quickly regained the lost output once the conditions returned to normal. In general, consequences of the financial crisis were felt more in ASEAN countries than in other Asian countries. In this study we use two measures of money supply, M1, or narrow money and M2, or broad money. M1 comprises of currency in circulation and transferable deposits. Currency in circulation refers to notes and coins issued by the Central Bank. Transferable deposits refer to current account deposits in national currency of nonbank depository corporation, nonbank financial institutions, state and local governments, public nonfinancial corporations, and private sectors with the Central

3 Estimation of the Money Demand Function in a Heterogeneous Panel Table 1 Macroeconomic Performance in the Selected Countries (Pre-structural period, and Post-structural period Country Average Per Average Average Average Average Average Average Average Average Capita GDP Real GDP Change in Nominal Nominal M1/ M2/ Change in Discount ($US) Growth per cent of M1 M2 GDP % GDP % per cent of Rate GDP Deflator Growth Growth Nominal Exchange Rate Bangladesh India Nepal Pakistan Sri Lanka Indonesia Malaysia Philippines Singapore Thailand Hong Kong Korea Source: Authors calculation based on various IMF statistics

4 26 Mohammed Nur Hussain and Albert Wijeweera Bank and commercial banks. On the other hand, M2 comprises M1 and time and savings deposits in national currency and foreign currency deposits of nonbank depository corporation, nonbank financial institutions, state and local governments, public nonfinancial corporations, and private sector with commercial banks. (International Financial Statistics Country Notes, 2007, pp. 25). Following financial liberalization, nominal growth for both M1 and M2 for the ratio of M2/GDP, which is the main economic indicator of financial liberalization increased in all countries from pre liberalization to post-liberalization. Most countries have also experienced depreciations of their currencies against the US Dollar. Singapore and Hong Kong are exceptions. As far as an interest rate is concerned, the average rate in the poststructural period is lower compared with that of the pre-structural adjustment period. However, Pakistan, Sri Lanka, Indonesia, and the Philippines experienced an increase in the average discount rate from the pre-liberalization to postliberalization period. This study attempts to contribute to the existing literature by examining whether there exists a long-run equilibrium relationship between money demand and its major determinants; real income, nominal interest rate, and nominal exchange rate. While most of the previous studies have focused on the estimation of money demand separately in each country, (see for example Ahmed and Islam 2007, Hossain 2007, Hussain and et al 2006, Islam 2000, Bahmani-Oskooee 2002, Rajapaksa 2004, Hafer and Kutan 2003, Khalid 1999, Chowdhury 1997, Dekle and Pradhan 1997 and so on) we empoly the panel cointegration tests developed by Pedroni (2004) for estimating the relationship. This paper is divided into 6 sections. Following the introduction, the theoretical money demand function is described in Section 2. Section 3 reviews relevant literature relating to the empirical analysis of the money demand function. Econometric methodology is presented in Section 4. Section 5 provides empirical results, and Section 6 concludes with policy recommendations. 2. THEORETICAL MODEL The amount of money that individuals and business hold varies with three macroeconomic variables: the nominal interest rate (i), real output (Y), and the price level (P). A negative relationship exists between money demand and nominal interest rate because the nominal interest rate is a macroeconomic factor that determines the cost of holding money. On the other hand, as real income increases, the demand for money goes up because individuals and businesses need more money to satisfy an increase in transactions. If the real money balance is given by, M P the money market equilibrium can be expressed as follows: M P L i, Y (1)

5 Estimation of the Money Demand Function in a Heterogeneous Panel The above function does not provide a complete picture of money market equilibrium relationship due to the developments in global financial market over the last two decades. As we know, most of the economies are increasingly becoming interdependent on foreign trade and foreign capital. Therefore it can easily be imagined that money demand during the post-liberalization period should not only depend on the real income and interest rate, but also determined by some international factors such as; foreign money, the foreign interest rate and the exchange rate. We follow Mundell (1963) to construct the open money demand function. The Exchange rate variable is included to capture the effect of expected changes in the foreign exchange rate. Following Harb (2004), open money market demand is specified in a log-linear form as in equation (2). d m a b y c i d ER (2) it i i it i it it it it Here m d, y it, is the natural logarithm of real M1 and M2, real GDP. i it and ER it are the nominal interest rate and nominal exchange rate respectively. Macroeconomic theory suggests that the estimated coefficient of b to be positive and c to be negative. The coefficient of the exchange rate variable can be either positive or negative. The reason for this theoretical ambiguity is that a depreciation of the domestic currency raises the value of the foreign assets in terms of domestic currency. If this increase is perceived as an increase in wealth, then the demand for domestic money increases. Therefore the sign of d is expected to be positive. On the other hand, if an increase in the exchange rate induces an expectation of further depreciation of the domestic currency, the public may have less of a demand for domestic currency and more of a demand for foreign currency. So, the sign of d is expected to be negative (see for example Bahmani-Oskooee and Rehman (2005). 3. EMPIRICAL ANALYSIS OF MONEY DEMAND FUNCTIONS There have been many empirical studies to estimate the money demand function. Some (see, inter alia, Ahmed and Islam 2007, Hussain et al. 2006, Rajapaksa 2004, Pradhan and Subramanian 2003, Hasan 2001, Ramajo 2001, Islam 2000, Siddiki 2000, Khalid 1999 Weliwita and Ekanayake 1988, Dekle and Pradhan 1997) studies utilize individual country data along with time series econometric techniques, while others (see, inter alia, Mark and Sul 2003, Jun 2004, Harb 2004, Alfredo and Leonel 2006, Narayan et al. 2008, Rao and Kumar 2008) use data from more than one country to make use of panel estimation methods. Recent studies seem to favor a panel approach over a time series method of estimation. To illustrate, Mark and Sul (2003) estimated the money demand function using the Pedroni s panel cointegration technique for 19 OECD countries and found evidence of the long run relationship between money demand and key macroeconomic variables. Harb (2004) using similar techniques, estimated money demand function for six Gulf Co-operation Council (GCC) countries namely Bahrain, Kuwait, Oman, Qatar, Saudi Arabia and the United Arab Emirates (UAE) and confirms that money demand is found to be cointegrated with relevant variables.

6 28 Mohammed Nur Hussain and Albert Wijeweera He also found that the income elasticity and the interest elasticity were consistent with the earlier literature. Jun (2004) utilizing a Fully modified ordinary least squares (FMOLS) approach with the data panel of G6 member countries, namely; Canada, France, Japan, U.K, and the U.S employing quarterly data for the period 1982:1 to 1998:4 estimated the money demand function to find that all income elasticities are positive with the exception of Italy. Among recent researchers who used the panel approach, Alfredo and Leonel (2006) investigated the money demand function for 27 countries using the static and dynamic fixed effects model for the period They used money demand specification suggested by Cagan (1956) in the estimation and found anticipated signs for income and interest rate elasticity. However, the magnitude of the income elasticity was rather low, which was attributed to the considerable differences between the countries and the issues associated with the method of conversion of GDP from local currency to the US dollar. Narayan et al. (2008) examined the open money demand function for five South Asian Countries, namely Bangladesh, India, Nepal, Pakistan and Sri Lanka using a panel cointegration test. They found that the money demand and relevant macroeconomic variables are cointegrated for all five countries. Income elasticities were positive and significant as expected, but domestic interest rate elasticities turned out to be negative for only two countries namely Bangladesh and India. They utilized three different methods of estimating long-run elasticities; dynamic SUR, panel DOLS (ordinary) and panel DOLS (common time effects). All three methods have in general produced results consistent with the existing literature. Along the same line, Rao and Kumar (2008) estimated the money demand function for 14 selected Asian countries from Interestingly, all income elasticities and interest elasticities were significant and signs were as anticipated. They further tested whether the financial reforms have had any effect on the money demand function, but did not find strong evidence in favor of the said hypothesis. We differ from Rao and Kumar (2008) because their money demand function is based on a conventional money demand and we instead employ the open money demand function in the current study. 4. DATA SOURCES AND METHODOLOGY Data Data covering the period 1974 to 2010 for twelve selected Asian countries has been included in this study. Macroeconomic variables such as, gross domestic product (GDP), GDP deflator, narrow money (M1), broad money (M2), nominal interest rate and nominal exchange rate are used in the estimation. Due to the unavailability of a unique interest rate variable, a proxy rate is used as the nominal interest rate. For instance, the discount rate is used for Bangladesh, Pakistan, Nepal, Thailand, Philippines and Korea, while the bank rate is used for India and Nepal. Our preferred choice is the deposit rate for Indonesia, the money market rate for Malaysia, the

7 Estimation of the Money Demand Function in a Heterogeneous Panel treasury bill rate for Singapore, and the lending rate for Hong Kong, respectively. Since interest rates move together, this should not alter the results considerably. To obtain Real GDP and real money demand, nominal GDP and M1 and M2 are deflated by the GDP deflator with the base year of Data were mainly extracted from IMF Statistical Database except for M1 and M2 data for Hong Kong, which were obtained from Hong Kong Monetary Authority website. Methodology Panel data framework is used for the analysis in this study. The general regression model for panel data can be expressed as follows: Y it = it + it X it + it The dependent variable is Y to the i th country in year t and X is a vector of control variables. it are the usual disturbance terms, which assumed to have zero mean and constant variance. Intercept term can vary for different countries and for different time periods. Estimation technique is usually based on the assumptions that we impose on the model. In this paper we assume that slope coefficients are constant, but the intercepts vary to incorporate heterogeneity among the countries. To determine the applicability of this assumption, the F-test has been utilized. Here the null hypothesis is that country specific effects do not exist and OLS is the best linear unbiased estimator. The restricted residual sum of squares is obtained from the OLS on the pooled model and the unrestricted residual sum of square is obtained from the least squares dummy variable regression. The F ratio can be calculated as follows: F n 1, nt n k 2 2 R unrestricted Rrestricted / n Runrestricted / nt n k The F-Test suggests that individual country effects in fact exist and model should be estimated with different intercepts accordingly. Having resolved the problem of intercept, we have to decide whether different intercepts should be represented as fixed effects or random effects. According to Baltagi (2003) the fixed effects model is more suitable if the model is taking a specific set of countries in the panel and the cross section component is small. Our panel consists of 12 Asian countries and the time spans for more than two decades. Recent developments in the panel data method suggest that testing for a unit root or the order of the integration of the variables is necessary before we proceed to estimation of the model. If there is a unit root, then we say that the particular series is considered to be non-stationary. Estimation based on non-stationary variables may lead to spurious results which may inflate R 2 and show statistical significance even though there is no meaningful relationship between the variables. In this paper, a host of panel unit root tests have been utilized to test for the

8 30 Mohammed Nur Hussain and Albert Wijeweera stationarity. These tests are namely Levin, Lin and Chu (LLC) test, the Breitungt, the Im, Pesaran and Shin (IPS) test, Fisher typing ADF and PP and Hadri test. If the series are found to be integrated in the same order, a cointegration test needs to be conducted to see whether they share a long run equilibrium relationship. Pedroni Panel Cointegration technique is widely used to test for the presence of long-run relationship among the variables in panels. If X it and Y it are integrated of the same order and the residual u it is stationary then we say X it and Y it are cointegrated. Pedroni test consists of seven different cointegration statistics to capture within and between effects in the panel. These seven tests are divided into two groups. The first group includes four tests namely, the panel statistics, the panel, the panel t-statistic (non-parametic) and the panel t statistics (parametric) based on the pooling along within dimension. The second group includes three tests are namely the group statistic (parametric), the group statistic (non-parametric), the group statistic (parametric) based on pooling the between dimension. The null hypothesis of each test is that there is no cointegration between the variables whereas the alternative hypothesis states that cointegration relationships between the variables exist. 5. RESULTS 5.1. Panel Unit Root Tests All five variables were tested for stationarity by the panel unit root test. The null hypothesis of the unit root is rejected for two variables, namely M2 and NINR in the case of LLC panel unit root test, whereas this hypothesis is rejected only for the interest rate variable (NINR) in the case of Breitung-t test. However, Hardi test rejects the null hypothesis of stationary for all five variables. The results of panel unit root tests are shown in Table Panel Cointegration Tests Following the panel unit root test, possible long-run equilibrium relationships are investigated by Pedroni panel cointegration test. Results reported in table 3 indicate that the null hypothesis of no cointegration is rejected at different levels of significance for the case of M1 model. As far as the M2 model is concerned, the null hypothesis of no cointegration is rejected by only two out of seven Pedroni type panel cointegartion tests. This cointegartion test results reveal that the money demand model for M1 is more cointegrated than the M2 money demand model. The results imply that the cointegration relationship is stronger for M1 compared to M Estimations As panel cointegration results confirm the existence of a long-run equilibrium relationship, we proceed to estimate the money demand function as specified in equation (1) by the fixed effects model. The results are reported in Table 4. The

9 Estimation of the Money Demand Function in a Heterogeneous Panel Table 2 Panel Unit Root Tests (Maximum Lag 1) Series LLC Breitung-t IPS-W ADF PP Hadri Ln (RGDP) (0.30) (0.95) (0.99) (0.74) (0.81) (0.00)* DLn (RGDP) (0.00)* (0.00)* (0.00)* (0.00)* (0.00)* (0.00)* Ln (M1) (0.56) (0.18) (0.99) (0.95) (0.92) (0.00)* DLn (M1) (0.00)* (0.00)* (0.00)* (0.00)* (0.00)* (0.04)** Ln (M2) (0.00)* (0.92) (0.56) (0.38) (0.52) (0.00)* DLn (M2) (0.00)* (0.00)* (0.00)* (0.00)* (0.00)* (0.00)* NINR (0.10)*** (0.00)* (0.15) (0.24) (0.37) (0.00)* D(NINR) (0.00)* (0.00)* (0.00)* (0.00)* (0.00)* (0.49) NEXR (0.76) (0.63) (0.99) (0.99) (0.99) (0.00)* D(NEXR) (0.00)* (0.00)* (0.00)* (0.00)* (0.00)* (0.89) Notes: The Panel unit root tests are: Levin, Lin and Chu (LLC), Breitung (2000), Im, Pesaran and Shin (2003) (IPS), ADF Fisher Chi-Square (ADF), PP Fisher Chi-Square (PP), and Hadri (2000). Ln (RGDP), Ln (M1), Ln (M2), NINR and NEXR denotes log of Real GDP, log of Real M1, Log of Real M2, Nominal Discount Rate, Nominal Exchange Rate, D first difference respectively. Figures in parentheses at probability value *, ** and *** denotes the rejection of the null at 1%, 5% and 10% respectively. Table 3 Cointegration Tests (Deterministic Intercept and Trend) Tests Real LM1 Real LM2 Panel v-stat (0.00)* (0.00)* Panel rho-stat (0.29) (0.79) Panel pp-stat (0.00)* (0.36) Panel ADF-Statistics (0.03)** (0.10)*** Group rho-stat (0.49) (0.97) Group PP-stat (0.00)* (0.74) Group ADF-Statistics (0.03)** (0.59) Notes: one maximum lag is included. Figures in parentheses at probability value,*, ** and *** denotes the rejection of the null at 1%, 5% and 10% significant level respectively

10 32 Mohammed Nur Hussain and Albert Wijeweera results help us to uncover some interesting features of the money demand function of these selected Asian countries. First, real income seems to exert a significantly positive impact on the real money demand. Results suggest that demand for M1 is income inelastic and the demand for M2 is income elastic. Findings are consistent with some of the existing studies. For example, Harb (2004) estimated the money demand using the FMOLS (Fully modified least squares) for GCC countries. He found that this elasticity 0.59 for pooled panel, and 0.78 for group-mean. However, the magnitude of the elasticity is considerably different from Alfredo and Leonel (2006) who examined the money demand function for 27 different countries and found that income elasticity ranged from 0.18 to On the contrary, Rao and Kumar s (2008) find that the income elasticity is close to one which is consistent with the estimation result of M2. Why are our estimates considerably different from similar recent studies? One may attribute this to the heterogeneity of the selected countries. This argument gains further strength from the country specific effects shown in Table 5. Table 4 Estimation Results Fixed Effect Variable LM1 LM2 C (0.88)a (0.76)a (0.00)* (0.89) LRGDP (0.08)a (0.07)a (0.00)* (0.00)* NINR (0.003)a (0.003)a (0.00)* (0.00)* NERX -2.79E E-05 (9.75E-06)a (8.51E-06)a (0.00)* (0.06)*** R Adjusted R F-test for cross-sectional effect (0.00)* (0.00)* F-test for time-effect (0.00)* (0.00)* Notes: a- indicates the standard error and,* denotes the rejection of the null at 1%, significant level at probability value. Second, the coefficient of the interest rate is found to be negative and significant. This suggests that a one percentage point increase in interest rate decreases money

11 Estimation of the Money Demand Function in a Heterogeneous Panel demand by 0.12 per cent which is similar to that of other studies (see for example Harb 2004, Rao and Kumar 2008). Third, our results suggest that the exchange rate is a significant variable in a money demand function. This supports the use of an open money demand model instead of a conventional money demand model. According to the results, the coefficient estimate of the exchange rate of money demand (M1) is negative in the case of M1 but the size of the coefficient is negligible. This result contradicts Harb (2004), in which the value was 0.37 for pooled panel and 0.04 for group mean. Table 5 Country Specific Effects Country M1 M2 Intercept Intercept Bangladesh India Nepal Pakistan Sri Lanka Indonesia Malaysia Philippines Singapore Thailand Hong Kong Korea CONCLUSION In order to analyze money demand function, we employ the panel cointegration method along with the fixed effects model over the period of for selected 12 Asian countries. Included in the model are real money demand, real GDP, nominal interest rate, and nominal exchange rate. The results indicate that there exists a significantly positive money demand and real income relationship for the countries investigated. Although the direction of the relationship is consistent with the literature, the size of the income elasticity coefficient for M1 considerably differs from prior studies such as Rao and Kumar (2008). Two other variables; nominal interest rate and nominal exchange rate exert significantly negative impacts on real money demand M1. The direction of the relationship and the magnitude of interest rate elasticity and exchange rate elasticity are consistent with the prior literature. These relationships are stronger for the money defined in narrow term rather than for the money defined in broad term. The study supports the use of the open money demand function in the analysis and implies that authorities should take both internal and external factors into consideration when setting monetary goals.

12 34 Mohammed Nur Hussain and Albert Wijeweera References Ahmed, S. and M. E. Islam (2007), A Cointegration Analysis of the Demand for Money in Bangladesh. Bangladesh Bank, Working paper WP Alfredo, G. H. and C. Leonel (2006), Empirical Evidence for a Money Demand Function: A Panel Data Analysis of 27 Countries Applied Econometrics and International Development 6(1): Bahmani-Oskooee, M. and S. Chomisisengphet (2002). Stability of M2 Money Demand Function in Industrial Countries. Applied Economics 34: Bahmani-Oskooee, M. and H. Rehman (2005), Stability of the Money Demand Function in Asian Developing Countries. Applied Economics 37: Calza, A. and J. Sousa (2003), Why has Broad Money Demand been More Stable in the Euro Area than in other Economies? A Literature Review. European Central Bank Working Paper No Dekle, R. and M. Pradhan (1997), Financial Liberalization and Money Demand in ASEAN Countries: Implications for Monetary Policy. IMF Working Paper WP/97/36. Dreger, C. R. and B. H. E. Roffia (2007), Long-run Money Demand in the New EU Member States with Exchange Rate Effects. Eastern European Economics 45: Green, W. (2003), Econometric Analysis, Prentice Hall, Upper Saddle River. Hamori, S. and A. Tokihisa (2001), Seasonal Cointegration and the Money Demand Function: Some Evidence from Japan. Applied Economics Letters 8: Harb, N. (2004), Money Demand Function: A Heterogeneous Panel Application. Applied Economics Letters 11: Hasan, M. S. (2001), Monetary and Fiscal Impacts on Economic Activities in Bangladesh: Further Evidence. The Bangladesh Development Studies XXVII(4): Hussain, Z., H. Awan, et al. (2006), Demand for Money in Pakistan. International Research Journal of Finance and Economics 5(September): Islam, A. M. (2000), Money Demand Function for Bangladesh. The Bangladesh Development Studies XXVI(4): Jun, S. (2004), Dynamic Panel Cointegration Approaches to the Estimation of Money Demand Functions. Global Economic Review 33(3): Kao, C. (1999), Spurious Regression and Residual-based Tests for Cointegration in Panel Data. Journal of Econometrics 90: Khalid, A. M. (1999), Modelling Money Demand in Open Economies: The Case of Selected Asian Countries. Applied Economics 31: Mark, N. C. and D. Sul (2003), Cointegration Vector Estimation by Panel DOLS and Long-run Money Demand. Oxford Bulletin of Economics and Statistics 65: Mundell, R. (1963), Capital Mobility and Stabilization Policy under Fixed and Flexible Exchange Rates. Canadian Journal of Economics 29: Narayan, P. K., S. Narayan, et al. (2008), Estimating Money Demand Functions for South Asian Countries. Empirical Economics Spring. Pedroni, P. (2004), Panel Cointegration: Asymptotic and Finite Sample Properties of Pooled Time Series Tests with an Application to the PPP Hypothesis. Econometric Theory 20: Pradhan, B. K. and A. Subramanian (2003), On the Stability of Demand for Money in a Developing Economy: Some Empirical Issues. Journal of Development Economics 72:

13 Estimation of the Money Demand Function in a Heterogeneous Panel Rajapaksa, R. P. C. R. (2004), The Money Demand Function for Sri Lanka for the Period Social System Studies 8: Ramajo, J. (2001), Time-varying Parameter Error Correction Models: The Demand for Money in Venezuela, 1983:I-1994:IV. Applied Economics 33: Rao, B. B. and S. Kumar (2008), A Panel Data Approach to the Demand for Money and the Effects of Financial Reforms in the Asian Countries. MPRA Siddiki, J. U. (2000), Demand for Money in Bangladesh: A Cointegration Analysis. Applied Economics 32: Weliwita, A. and E. M. Ekanayake (1988), Demand for Money in Sri Lanka during the Post Periods: A Co-integration and Error-Correction Analysis. Applied Economics 30:

14

15