ESTIMATING THE OPTIMAL MONETARY POLICY RULE FOR PAKISTAN

Abstract

A near consensus in the contemporary monetary economics is that monetary policy can achieve its objectives more precisely if it is designed as a rule rather than discretion. Even a well intentioned discretionary monetary policy becomes time inconsistent while consistency is at the heart of rule based policy adopted by an autonomous and transparent central bank. In standard Macroeconomic Models a loss function, defined over output gap and deviation of inflation from its optimal value, is assigned to the central banker who has the autonomy to choose its operating target so that the loss to the society is minimum. Assuming behavioral equations of the private sector as constraint, the first order conditions are derived from loss minimization, which after manipulation give a policy reaction function that is made explicit in an instrument rule while it remains implicit in the targeting rules (Svensson 1997). Empirical literature brings to light the superiority of rule based policy in a variety of Macroeconomic Models and against a bunch of structural shocks that may hit the economy. The underlying thesis sets two objectives regarding monetary policy of Pakistan. The first objective is to estimate monetary policy reaction function. For this purpose, the Taylor type rules and McCallum rules are estimated using quarterly data on Pakistan economy over the period 1993 Q3 to 2010 Q2. Both types of rules have been modified by incorporating exchange rate management and interest rate smoothing as policy objectives. Moreover, we have found recursive estimates of the parameters to sort out policy inconsistency. We have also looked into the issue of nonlinearity of the monetary policy reaction function with regards to output gap and inflation rate assuming asymmetric preferences of monetary authority. The second objective is to estimate loss, defined as sum of variances of output gap and inflation rate, associated with different specifications iv of the rules, which is then compared with that found in the historical data. We find that monetary authority in Pakistan does not follow Taylor rule as coefficient of output gap is negative and statistically insignificant and the coefficient of inflation rate, though statistically significant, is far below the benchmark value suggested by Taylor (1993). State Bank of Pakistan (SBP) is found to involve in exchange rate management and interest rate smoothing and this result is robust to different modifications in the Taylor rule. The parameters of output gap, inflation rate and differenced exchange rate, in the reaction function, are not stable over time and vary over the business cycle and across different inflationary regimes. The variation in the coefficient of output gap is found countercyclical while the coefficient of inflation rate follows the same pattern with respect to inflationary regimes. The coefficients of exchange rate and lagged interest rate remain almost stable. The threshold value of output gap is found 2.5% below which the response of interest rate to output gap fluctuations is positive but above which the response is insignificant. The threshold rate of inflation is found at 6% and coefficient of output gap is found positive only high inflationary regime while the coefficients of inflation rate and exchange rate are significant only in low inflationary regime. Monetary authority responds to currency depreciation more strongly when interest rate is low compared to that when it is high. Moreover, the response of interest rate to output gap is significant only if currency depreciation is below threshold (estimated at 0.68) while response to exchange rate is significant only if there is high speed of depreciation (above threshold). The results are robust to inclusion of fiscal deficit in the Taylor rule. In Pakistan, fiscal deficit negatively affects interest rate which is because of the borrowing of government from State Bank of Pakistan (SBP) for budgetary support. In a modified version of the Taylor rule interest rate is found to negatively respond to changes in growth rate of real GDP. Growth rate of monetary base negatively depends on the difference between nominal GDP growth rate and its average value indicating countercyclical response at the part of monetary authority. Moreover, growth rate of money exhibits strong inertia and is negatively related to currency depreciation. The coefficients in the McCallum rule too are not stable during the sample period. The coefficients of growth rate of nominal GDP and exchange rate are not stable over time, while the parameter capturing inertia is stable over the sample period. The response of monetary growth rate to nominal GDP growth rate and to exchange rate are significant only when nominal GDP is above its threshold value and/or when currency depreciates at higher rate. The simulation analysis confirms that policy consistency can improve welfare significantly and rule based policy, in general, is found superior to the one that has been observed during the sample period. Moreover, the original Taylor (1993) rule is found the best among all specifications and inclusion of exchange rate and lagged interest rate negatively affects welfare and the condition of zero lower bound on nominal interest rate is violated, in most of the cases, when interest rate smoothing is included as a policy objective. Interestingly, making monetary policy reaction function nonlinear does not add to the performance of the rule. The stochastic simulation also confirms that Taylor rule can perform well in a variety of shocks that may hit the economy. Finally, it is found that Taylor rule may perform well even if the fiscal deficit partially dilutes the stance of monetary policy but the performance of the rule is better in a model of monetary dominance.