identifying the relevant channels through which the impact of monetary policy is transmitted to the real economy has been one of the most challenging and crucial tasks in the conduct of monetary policy. A clear understanding of the monetary transmission channel can help in ensuring an effective and successful implementation of monetary policy. Besides, the post-Keynesian economists have emphasized the role of banking credits as the key transmission mechanism, although this paper is not directly testing that hypothesis. This aspect has yet been seriously studied, and the only study that exists of a developing country is that of Turkey. Monetary policy affects the real economy via several channels, but the main two that are often mentioned in the literature are the “money channel” and “credit channel”. The money channel, which emphasises the role of central bank’s money in the monetary transmission process, has secured a strong footing in the economic literature (see, for example, Brunner and Meltzer, 1972, 1988, 1990; McCallum, 1990, 1999, 2001, 2004; McCallum and Nelson, 2005; Feldstein and Stock, 1993; Hess and Porter, 1993; Dotsey and Otrok, 1994; Meltzer, 1969, 1970, 2001; Christiano and Rostagno, 2001; and Nelson, 2003a, 2003b). According to this view, an expansionary monetary policy causes bank deposit and money supply to expand, which results in real interest rate to decline, thus it is also known as the interest rate channel. Studies that are supportive of the existence and importance of the bank lending channel in general find that, bank loan behave in a predictable manner when a policy shock is exerted on the economy. When a central bank conducts open market operations to reduce liquidity in the system, bank’s deposit and liquidity are reduced, thereby restricting the ability of banks to supply new loans. Credit is less available in the economy, resulting in firms to cut down on investment and production, thus lower aggregate output in the economy. On the other hand, studies that do not support the importance of the bank lending channel generally find that banks offset the decline in liquidity following a tight monetary policy by adjusting their balance sheet items. Such ability to do so renders monetary policy ineffective in influencing supply of bank loan remaining relatively unchanged even after the implementation of the tight policy stance. For example, Suzuki (2001) employs the VAR analysis to examine the bank lending channel using the Australian data and report that bank loans contracted following a tight monetary policy, the contraction. Studies by Morris and Sellon (1995) which used US data for 1974-1990 also show that bank loan is not an important channel for monetary policy since bank loan supply does not decline in response to tight monetary policy. Earlier studies such as Scholnick (1996) and Tan (1997) on Malaysian situation adopted the credit-rationing model and estimated the disequilibrium model of lending rate. Both studies support the presence of credit rationing as the lending rate adjusted very slowly and did not clear the loan market. More recent studies such as Ibrahim (2005), Azali (2003), Tang (2000, 2002) and Vaithilingam et al. (2003) employed more recent investigation techniques such as the VAR that enable more detailed and conclusive findings. In general, these studies are supportive of the important role played by bank loan in the transmission of monetary policy. The need to understand the interaction between bank loan and bank deposit with monetary policy indicator is critical, hence this study is motivated to address this issue. Bank lending accounted for more than 100 percent of the country’s total gross domestic product (GDP) and is expected to remain high due to the moderate economic performance. Indeed, the important role played by bank lending in achieving the overall economic objective is clearly acknowledged by Bank Negara Malaysia. “As the nation’s monetary authority, the Central Bank is responsible to the Government for promoting monetary stability and a sound financial structure, and for influencing the credit situation to help achieve the nation’s overall economic objectives.” (1994, p. 91). An aspect of novelty of this study is that not only that it uses the latest empirical methodology, it also considers two sub-sample periods following the crisis in 1997/1998. The rest of this paper is organized as follows: the next section provides some background information on the monetary transmission process. Section 3 discusses the data and empirical framework. Section 4 presents the empirical findings and the discussions of the results. Lastly, Section 5 concludes this paper. 2. An Overview of Related Literature Empirical studies analyzing the importance of the banking sector in the monetary transmission process focus on the relationship and impact of disturbances in the banking sector on the aggregate economic activity: see, for example Friedman,1984; Bernanke and Blinder, 1988 and 1992; Bernanke and Gertler, 1995; Bernanke and Mihov, 1998; Dale and Haldane, 1998; Domac and Ferri, 1998; Garretsen and Swank, 1998; Guender, 1998; Gertler and Gilchrist, 1993; Walsh and Wilcox, 1995 and Suzuki, 2001. In general, these studies measure the responses of selected macroeconomic variables such as output, unemployment rate and inflation and the major bank balance sheet variables, which are deposit (representing bank liability) and loan (representing bank asset) to changes in the monetary policy indicator such as interest rates. However, it is often shown that the macroeconomic responses to policyinduced interest rate changes are larger than that implied by the money view. This gives rise to the lending channel, which suggests that bank loans may provide an additional channel for the monetary authority to affect the real economy. An influential study by Bernanke (1983) finds that the output decline in the US during the 1930-1933 Great Depression cannot be fully explained by monetary disturbances as described by the money view. Using the federal funds rate as an indicator for monetary policy, that study analyzed the responses of selected macroeconomic variables such as unemployment rate and inflation, and the bank balance sheet items to changes in the fed funds rate. It was found that in the short run, banks offset the decline in liquidity following a contractionary monetary policy by selling-off their securities holdings, thereby leaving the supply of loan in the economy virtually unchanged. However, in the longer term, the tight monetary policy causes banks to terminate the existing loans and stop making new loans. The brunt of the tight monetary policy is especially felt by borrowers who are dependent on bank lending. Consequently, the tighter monetary policy results in a decline in the aggregate economic output. Until now, the reduction in loan supply has often been quoted as one of the reasons for the depth and persistence of the Great Depression. As lessons learned from that crisis, the central banks the world over are loosening monetary policy since August 2007 in response to the sub-prime loan crisis of 2007/08. In Malaysia, the banking system forms an integral part of a larger financial system which is made up of the financial institutions and the financial markets. The financial institutions can be grouped into the banking system and the nonbank financial intermediaries. The banking system is made up of commercial banks, finance companies and merchant banks, while the non-bank financial intermediaries comprised the provident and pension funds, insurance companies, development financial institutions, savings institutions and other financial intermediaries (such as cooperative societies, leasing and factoring companies, venture capital companies and housing credit institutions). The operations of Malaysia’s banking system fall under the purview of the central bank, the Bank Negara Malaysia (BNM). The financial system experienced a rapid growth in tandem with the robust economic performance during 1976-1996. By the end of 2006, the total assets of the financial system was about equal to the GDP of the country and it was RM2,091.2 or US$ 635 billion compared to RM329.3 billion in 1990 and RM74.2 billion in 1980. The rapid growth financial sector reflects the process of financial deepening due to increased demands for financial products and services in line with the structural transformation of the economy. More importantly, the banking system remained a major contributor to the total assets of the country’s financial system. Of the total assets of the financial system at end-2006, 68 percent were contributed by the banking system while 32 by the non-bank financial intermediaries. In terms of growth, the total assets of the banking system grew by approximately 11.3 percent in the period of 1990-2006, comparable to that of the financial system at 11.5 percent. Apart from its significant contribution to the growth of the country’s financial system, the banking system has also been instrumental in financing the country’s economic growth. This is well-reflected by the high ratio of total loan extended by the banking institutions to total gross domestic product (GDP) at more than 100 percent as at end-2006. One important lesson learned from the 1997/98 Asian financial crisis was that fragility of the banking sector could lead to a loss of confidence which would affect the real economy negatively. In view of this, BNM banking policies in the post-crisis period are directed towards strengthening and increasing the resilience of the banking system as well as to preserve the integrity and stability of financial sector. In this regard, one of the major initiatives taken was to consolidate the banking institutions in order to create a core of strong domestic banking institutions with large capital base. Prior to the consolidation program, as at end-1997, the banking system comprised 35 commercial banks, 39 finance companies and 12 merchant banks. As a result of the consolidation program, as at end-2006, the number of commercial banks has been reduced to 22 operating within ten holding companies, with the operations of the finance companies fully being absorbed by the commercial banks and the merchant banks being re-categorized into the 14 investment banks comprising the merchant banks and discount houses. This led to a strong capital base of risk-weighted capital ratio at 13.5 percent as at end-2006. The non-performing loan ratio declined from above 13.9 percent in 1998 to 3.7 percent. 3. Data and Methodology A. Data The overnight rate, henceforth denoted as ONR is used as the monetary policy indicator, while the objective variables comprised of the bank balance sheet items, namely, bank loans (BL) and bank deposits (BD) and other variables which are consumer price index (CPI) and industrial production index (IPI). The CPI and the IPI are used as the proxies for inflation and level of economic activity, respectively. Given that the economy is a highly open one, the conduct of monetary policy may be influenced by foreign shocks, thus we also include real effective exchange rate (RER) as a control variable. All series are real (adjusted by price index with the year 2000 as base year) and variables are expressed in logarithm, except for the interest rate. The study uses monthly data from January 1989 to December 2006. During this period, the economy experienced a financial crisis. To avoid the disturbance from affecting the results of the analysis, tests are done over separate subperiods and also over the whole period. The pre-crisis period covers the period from January 1989 to December 1996. In this period, the data set excludes the period from 1997 to 1998, which were the heights of the East Asian financial crisis. The post-crisis period runs from January 1999 to December 2006. All data are sourced from BNM’s Monthly Statistical Bulletin, except for the RER which is gathered from International Financial Statistics published by International Monetary Fund. B. Empirical Framework ARDL Bound Testing Approach The long-run dynamic relationships between monetary policy variables and bank balance sheet items are estimated by using the newly proposed ARDL bound testing approach: Pesaran et al. (1996). The ARDL has numerous advantages. First, unlike the most widely used method for testing cointegration, the ARDL approach can be applied regardless of the stationary properties of the variables in the samples and it allows for inferences on long-run estimates, which is not possible under the alternative cointegration procedures. This procedure can be applied irrespective of whether the series are I(0), I(1), or fractionally integrated (Pesaran and Pesaran 1997; and Bahmani-Oskooee and Wing Ng, 2002), thus avoiding problems resulting from non-stationary time series data (Laurenceson and Chai, 2003). Second, the ARDL model takes sufficient numbers of lags to capture the data generating process in a general-to-specific modelling framework (Laurenceson and Chai, 2003). It estimates (p+1)k number of regressions in order to obtain optimal lag-length for each variable, where p is the maximum lag to be used, k is the number of variables in the equation. Finally, the ARDL approach provides robust results for a smaller sample size of cointegration analysis. Since the sample size of our study is small, particularly for pre- and post-crisis subperiods, this provides a motivation for the study to adopt this model. Ideally, for the purpose of this study, we needed to incorporate all the variables in the model, but a VAR model can be poorly estimated in a finite sample, as the addition of a variable will quickly exhaust the degree of freedom. Following Gertler and Gilchrist (1993, 1994), we estimated a series of separate VAR models including INF, IPI, and RER, policy variable (ONR), and the bank balance sheet items. Thus, our models contain only five variables. The ARDL models for the bank balance sheet items used in this study can be written as follows: BDt = į0 + I1ONRt + M2IPIt + J3RERt + ȝ4INFt + €t

BLt = į0 + I1ONRt + M2IPIt + J3RERt + ȝ4INFt + İt

The error correction version of ARDL framework pertaining to the variables in the Equations (1.1) and (1.2) can be reproduced as follows: p

'BDt G0 ¦Ti 'BDt i ¦Ii 'ONRt i ¦Mi IPIt i ¦ Ji 'RERt i ¦Pi 'INFt i (2.1) O 1 BD t 1 O 2 ONR t 1 O 3 IPI t 1 O 4 RER t 1 O5 INFt 1 1t p

'BLt G0 ¦Ti 'BLt i ¦Ii 'ONRt i ¦Mi IPIt i ¦ Ji 'RERt i ¦Pi 'INFt i

O 1 BL t 1 O 2 ONR t 1 O 3 IPI t 1 O 4 RER t 1 O5 INFt 1 H1t

In the above equations, the terms with the summation signs represent the error correction dynamics, while the second part (term with λs) correspond to the long-run relationship. The null of no cointegration in the long-run relationship is defined by H0: λ1 ≠ λ2 ≠ λ3 ≠λ4 ≠ λ5 = 0 which is tested against the alternative of H0: λ1 ≠ λ2 ≠ λ3 ≠λ4 ≠ λ5 ≠ 0, by the means of familiar F-test. However, the asymptotic distribution of this F-statistic is non-standard irrespective of whether the variables are I(0) or I(1). Pesaran et al. (1996) have tabulated two sets of appropriate critical values. One set assumes all variables are I(1) and another assumes that they are all I(0). This provides a bound covering all possible classifications of the variables into I(1) and I(0) or even fractionally integrated. If the F-statistic lies above the upper-bound level, the null hypothesis is rejected, which indicates the existence of cointegration. However, if the F-statistic falls below the bound level, the null cannot be rejected, showing that no cointegration exists. If, however, it falls within the band, the result is inconclusive. Finally, in order to determine the optimal lag-length incorporated into the model and select the ARDL model to be estimated, the study employs the Akaike Information Criteria (AIC).