Determinants of commercial bank reserve behavior in an open economy: Kenya, 1976-1991

In studying the money supply process and monetary policy in Kenya, free liquid reserves (net excess reserves) of commercial banks are of central importance. This study investigates the causes of movements in the free liquid reserve position of Kenyan commercial banks during the 1976-1991 period. During this period, the liquidity of commercial banks as measured by the ratio of free reserves to total deposit liabilities, $\beta,$ was the major source of variation in the money multiplier, frustrating the CBK's attempts to regulate M2. Thus modeling $\beta$ could provide information about whether the desired free reserve ratio of banks is a function of the yield on the secondary reserve assets, the penalty rate, the loan-deposit rate spread, and transitory deposit potential. Following Richter and Teigen (1982), two behavioral variables are included in the general model that may affect the desired free reserve ratio: (1) changes in net government deposits at the Central Bank of Kenya (CBK) relative to the commercial bank deposit liabilities (g) and, (2) changes in foreign exchange position of the CBK relative to the level of commercial bank deposit liabilities (fx). The empirical results indicate that, contrary to previous studies, changes in transitory deposits induced banks to change their desired free liquid position and presumably in the supply of credit and money. Thus, an activist policy targeted in affecting the transitory deposits will cause changes in desired positions. Second, the results indicate the T-bill rate and loan-deposit rate spread inversely affected banks free reserve ratio. Hence, changes in deposit and loan rates (e.g., explicit credit controls) could be an effective control on Kenyan banks liquid reserves. Third, the deviation between the desired and actual free reserves has been influenced by the behavior of the CBK (fx) and the government (g). The negative and highly significant coefficient estimate for fx suggests that the CBK was able to partially offset the effects on banks liquid reserves of the much larger changes in its foreign exchange position that took place. Furthermore, the positive and highly significant coefficient for g suggest that the CBK overcompensated at least in the very short-run. Finally, the outcome from the TTT (or Hendry's) econometric methodology yielded an improved model over the traditional OLS methodology.