Excess Reserve Holdings of Jordan Commercial Banks : VARX Implementation

This study tends to fill a gap in the literature as it considers a small open developing country in a region that takes the attention of the world. The main aim of this study is to examine the factors that determine commercial banks excess reserve holdings in Jordan banking system. Vector Autoregressive with explanatory variables (VARX) technique is used. The empirical results suggest that excess reserve in Jordan banking system tend to reach its long term level after four quarters and the correction process occurs every six months. Another finding, as banks customers' prefer to hold currency banks prefer to hold less excess reserve. Moreover, banks tend to hold more excess reserve in booming economies. Finally, monetary policy tools seem to be affecting factors in banks excess reserve holding preferences. However, fiscal policy does not seem to be within the factors that affect excess reserve holdings.


Research Problem and Importance
Central banks supervise and control banks using direct and indirect tools in order to achieve stability in prices, keep interest rate in the desired level, and stabilize exchange rate in country currency.On the other hand, Banks hold reserve with central bank in order to meet their liabilities.The main objective that monetary authority controlling and supervising commercial banks is to limit their daily transaction and to make sure that commercial banks do not extend the granting facilities, because unreasonable increase in lending leads to high inflation which on the other hand causes a shortage in banks' liquidity.Banks usually has two types of reserve, one is obligatory required by central banks and another is optional.The excess reserve is the deposited amount into a central bank account that exceeds the required reserve.The required reserve usually earns no interest while the excess reserve may earn little or no interest.
Central bank of Jordan (CBJ) is the monetary authority that takes the responsibility to monitor and supervise the banking system in Jordan, which was established in 1964 and officially started in 1965.Like all banking system in the world, CBJ has imposed the required reserve on banks various types of deposits in 1967 in order to control banks when monetary policy implemented and to provide liquidity when shortage occurs.Required reserve is imposed on commercial bank deposits, the ratio of required reserve changes according to monetary policy objectives.CBJ should hold at least 5% of the bank's deposits and not more than 35% (Note 1).
According to the CBJ standards, the excess reserves of commercial banks includes reserves on: net issue of governments security that issued by the Ministry of Finance plus the interest rate that accrued on this issues, the amount of banks borrowing from each other, certificate of deposits (CD) which is issued by CBJ as direct tool to absorbing liquidity, repurchase agreement between banks (overnight Repo), and the window amount (what left from banks reserves at the end of daily transaction) (Note 2).
The objective of this study is to analyze the excess reserve holding in Jordan banking system and to investigate the determinants of commercial banks excess reserve holding.It also aims to estimate the marginal effect of each of the determinants of banks' excess reserve holding.In order to achieve the purpose of this study and following previous researchers time series techniques are employed.
The importance of this study comes from the timing of its investigation as the sample considers observations that cover the period before and after the 2006/2007 financial crises.Moreover, it considers Jordan banking system which is a Middle Eastern country that is surrounded with political conflicts.To my knowledge, there is no previous study covers similar country in similar region during the financial crises.Therefore this study tends to fill a gap in the literature as it considers a small open developing country in a region that takes the attention of the world.
The rest of the study is organized as follows; the next section provide a brief review of the literature, section three describes each type of reserves in Jordan banking system and their implementation trend.Section four explains specifications of empirical models and variables definitions, source of data and sample selection follows in section five.The following section introduces the empirical results and section seven concludes.

Literature Review
Due to the importance of excess reserve in the banking system continuous research studies were performed to investigate the factors that determine why profit maximizing banks hold excess reserve.The literature in the topic is divided into two groups, a group of literature focus on developing reserve management models (Orr & Mellon, 1961;Poole, 1968;Frost, 1971;Baltensperger, 1974;Baltensperger, 1980;Santomero, 1984, Longworth, 1989).More recently, other studies continue to modeling banks' reserve management (Civcir & Parikh 1998;Allen, 1998;Nautz, 1998;Furfine, 1998;Clouse & Dow, 1999;Selgin, 2001;Clouse & Dow, 2002;Heller & Lengwiler, 2003;and Bindseil, Camba-Mendez, Hinch, & Weller, 2004).On the other hand, literature body has a set of studies that empirically investigate commercial banks' demand for excess reserve (Agenor, Aizenman, & Hoffmaister, 2004;Demiralp & Farley, 2005;Maynard & Moore, 2005;Aikaeli, 2006;Saxegaard, 2006;Murta & Garcia, 2009;Agenor & El Aynaoui, 2010;Talasli, 2010;Anderson-Reid, 2011).A brief explanation of some of the studies which are related to the determinants of excess reserve is introduced.Agenor et. al. (2004) develop a model that sets the precautionary motives as the determinants of excess liquid assets held by commercial banks.They include the required reserve ratio to total deposits, uncertainty in the ratio in cash to deposit, the dispersion of output from trend, the discount rate and the lag of excess reserve.Their work was extended by Saxegaard (2006) by including variables that account for the unintentional build-up of excess reserves using a sample data set from the banking system in sub-Saharan Africa (SSA).Saxegaard commented that banks tend to hold higher excess reserve to protect them against surprising deposits withdrawals.Moreover, he includes the demand deposits to time and savings deposits ratio and find that this variable has a positive impact on the excess reserve holdings.
Another study by Demiralp & Farley (2005) investigates the determinants of reserves and the impact of monetary policy tools on these factors.Their model includes required reserve ratio, an important tool of monetary policy, as an influencing factors on excess reserve.Their investigation also considers change in average daily reserve maintenance, the federal rate, open market operations and the amount of required for each bank.The main conclusion of suggest a positive impact of open market operations on excess reserve.The determinants of demand for excess liquid assets in Barbados banking system during the period of 1974 -2004 are investigated by Maynard & Moore (2005).Their empirical results classify the determinants of excess reserve demand into customer characteristics, economic condition, monetary policies and fiscal policies.Maynard & Moore include the net domestic assets of central bank to account for the effect of fiscal policies.Their findings suggest a positive relationship between cash to deposit ratio, the customers' preferences variable.They also find a significant impact of the monetary measures on the demand for excess reserve.Economics condition and fiscal policy tools are also concluded to affect the level of excess reserve, according to Maynard & Moore.Aikaeli (2006) investigates the long run determinants of excess reserve holdings.His study finds a negative long run effect of required reserve on excess reserve and positive long run effect of cash preferences, bank borrowing rate, and loan return variation on excess reserve bank holdings.Murta & Garcia (2009) discuss that banks hold reserve to manage their liquidity risk.They employ a sample from the banking system in the EURO area to investigate the demand for excess reserve.Their main findings suggest that there is a positive impact of financing cost on excess reserve demand.Moreover, there is no significant impact of uncertainty resulted from the credit crisis on the demand for excess reserve.
Agenor & El Aynaoui (2010) use a simple macroeconomic model with credit market imperfection and find that demand for excess reserve is determined by precautionary variables and the forgone opportunity from holding the reserve.Talasli (2010) investigates the determinants of excess reserve demand in Turkey banking system and analyze the pattern of reserve holding in Turkey using ARMA based model.He concludes that there is a systematic pattern of excess reserve balances within single maintenance period.Moreover, liquidity shocks, size of outflow funds and shortage in interbank borrowing have significant effect on the pattern of holding excess  The dependent variable is the amount of excess reserve held by commercial banks with the central bank.The first independent variable is the commercial banks customer characteristics measured by the currency to deposits ratio ⁄ , this variable represents the ability of the bank to meet commercial banks' customer currency demand fluctuations (Anderson-Reid, 2011).The economic condition is measured by the real gross domestic product (RGDP) and is expected sign may be in both directions, positive and negative, it is positive as in boom condition the economic activities grows and banks will tend to hold more reserve to avoid any overdraft.It also can be negative as boom condition may increase the opportunity cost of holding none earning reserve and therefore will hold no reserve or less reserve.
Fiscal policy is measured by the net domestic assets of central banks assets (NDA), this variable proxy for the government spending and the tax policy in the economy.Expansionary fiscal policy implies higher NDA and thus higher excess reserve.Monetary policy is measured by required reserve (RR) and interest rates (interest rate on treasury bills and bonds (R TB ), interest rate on interbank borrowing (R IBL ) and interest rate on overnight window deposits (R OWD )).The required reserve ratio is expected to negatively affect the excess reserve holding.
When monetary authority increases RR it is meant for the economy to slow down through reducing the percentage of deposits available for commercial banks for loans purposes.As a result the commercial banks will also lower their holding of excess reserve in order to compensate the reduction of loans.In other words, RR is expected to negatively affect the banks holding of ER.
The interest rate on the treasury bills can be viewed as the opportunity cost of holding excess reserve that is the higher the interest on treasury bills the higher the cost of holding excess reserve.Therefore, R TB is expected to negatively affect the holding of excess reserve.The interest rate on interbank loans (R IBL ), interbank loans is the type of loans that commercial banks may refer to fund the financial deficiency.The (R IBL ) is the cost of these loans hence the higher the (R IBL ) the higher the cost of interbank loans.Therefore, the higher the (R IBL ) is the higher the incentives for commercial banks to hold more ER and vise a versa.Finally, the overnight window deposit interest rate (R OWD ) is included, there is a positive relationship expected between ER and (R OWD ).

Sample Selection and Data Sources
Time series data sample for the dependent variable and independent variables is constructed for the period 1994-Q1 to 2012-Q4.The resulted sample contains 76 observations in quarterly frequency format.The natural logarithm is taken to all variables included in the model.The variable of net domestic asset (NDA) has many negative values which make calculating the natural logarithm impossible.Therefore, before calculating the natural logarithm for the NDA variable mathematical transformation is done to all observations according to the following transformation equation NDA t = NDA t -Min{NDA}.

Estimations Results
The nature of macroeconomic variables has high possibility of being trending and non-stationary.Therefore, it is necessary to perform unit root test to figure out the level of integration of each variable.I employ two unit root tests, Augmented Dickey Fuller (ADF) test and Phillips Perron (PP) test, Table (2) reports the results of unit root test employing three model specifications, with constant term, without constant term and with trend.VAR model turns a multiple time-series regression of each endogenous variable on lags of itself and on lags of all the other endogenous variables.VAR also fits a variant of VAR models known as VARX model that contains independent variables.VARX is the model to be employed in this study because I am interested to know the determining factors of excess reserve commercial banks holdings.VARX model can be expressed as follows: Where p is the lag order and X t is a vector of independent variables taken at time t and β is a vector of coefficients for the exogenous variables.Before turning to VARX estimation it is important to determine the appropriate lag order of the data.Table (3) reports the lag order selection based on different criteria, Likelihood Ratio (LR), Final Prediction Error (FPE), Akaike's Information Criterion (AIC) and Hannan and Quinn Information Criterion (HQIC) (Note 3).Lag order selection tests suggest the fourth lag as the appropriate lag order.This is consistent with Pesaran & Pesaran (1997) who recommend the fourth lag order to be used with quarterly data.4), the lag terms indicate that the excess reserve is affected in the amount of excess reserve two and four quarters in the past.The lag terms indicate the speed of excess reserve correction to the long term level.Coefficients of LnER t-2 and lnER t-4 is significantly and positive, that is the long term level of excess reserve reached after four quarters (one full year), the correction process occurs in two steps (every six months).The positive sign indicate that short term excess reserve is below its long term levels by .40percent (.20 percent each six months).Banks customers' characteristic represented by the natural logarithm of the C/D ratio has significant negative effect on the determination of the excess reserve holdings which is consistent with its expected sign.This relationship indicates that as currency to deposits holding increase excess reserve decrease because banks customers tend to keep their wealth in hand as currency rather than deposit them in the banking system.
The economic condition seems to have its expected effect on excess reserve holing.The variable Ln(RGDP) has significantly positive effect on the holding of excess reserve, that is; the belter the economic condition the higher holding of excess reserve.The impact of required reserve also comes out as expected.The variable lnRR has significant negative effect on excess reserve holdings, the increase in required reserve banks tend to reduce excess reserve to fill in the shortage occurred as a result to the higher required reserve.

Conclusion
This study tends to fill a gap in the literature as it considers a small open developing country in a region that takes the attention of the world.It investigates an important issue during the period before and after the 2006/2007 financial crises.The main aim of this study is to examine the factors that determine commercial banks excess reserve holdings in Jordan banking system.
Time series techniques are used to estimate the excess reserve elastisities according to a set of variables.Vector Autoregressive with explanatory variables (VARX) is used, the dependent variable is the natural logarithm of excess reserve held in CBJ during the period of 1994-Q1 to 2012-Q4.The independent variables are classified into four groups; banks characteristics, economic condition, monetary policy and fiscal policy.
The coefficients of the excess reserve of lag terms indicate that the long term level of excess reserve reached after four quarters (one full year) and the correction process occurs in two steps (every six months).Banks customers' characteristic is significant.The empirical results indicates that as currency to deposits holding increase excess reserve decrease because banks customers tend to keep their wealth as currency in hand rather than deposit them in the banking system.The better the economic condition is the higher holding of excess reserve.Moreover, the monetary policy seems to have significant impact on banks excess reserve holdings.More specifically, the increase in the required reserve tends to reduce banks holding of excess reserve.The higher the interest rate of interbank loans the higher banks holding of excess reserve and the higher the interest of overnight window deposits the higher the excess reserve holdings.Finally, the fiscal policy concluded to be statistically insignificant.
This study can be extended in the future by considering the Middle East and North African countries where different environmental and institutional conditions can be tested.Additional future trend is to consider data at individual banks level data panel data techniques can be used.

Table 2 .
Unit root test of model's variablesThe results indicate that all the included variables are integrated of order zero [I(0)] or integrated of order one [I(1)] that is the variables are stationary at the level form or at the first difference form.The results of the unit root tests indicate that Vector Autoregressive (VAR) technique is suitable to be employed.It worth to mention that it becomes not possible to employ VAR model if there is any of the variables I(2) or higher.

Table 3
VARX estimation of the equation above are reported in table (