5  The impact of monetary policy on the growth cycle and inflation

Section 4 showed that changes in domestic financial conditions, including interest rates, have sometimes moderated the magnitude of the growth cycle; at other times they have accentuated the magnitude of the cycle. The purpose of this section is to isolate the impact of monetary policy on the growth cycle and deviations of inflation from trend.

The impact of domestic interest rate shocks included in the contribution of domestic financial shocks to detrended GDP, and presented in Figure 2, do not fully capture the impact of monetary policy. This is because in addition to the direct(non-systematic)impact of interest rate shocks on detrended GDP, there is also an induced (systematic) impact. The direct impact of interest rate shocks are captured by structural shocks to the interest rate equation that can arise from random shifts in interest rates including changes in preferences of the central bank and exogenous changes in portfolio preferences. The induced impact arises when the monetary authority changes interest rates in response to changes in the domestic economic environment. For example, in response to a rise in import prices that is expected to induce a recession, the central bank will lower interest rates to offset the anticipated fall in domestic demand. This induced impact of monetary policy is captured by the reaction function. Therefore, to gauge the impact of monetary policy on detrended GDP (y), both the direct and induced effects need to be measured. The methodology outlined in this section follows Dungey and Pagan (2000, pages 335-336).

The reaction function is derived on the basis that the monetary authority responds to forecasts of deviations in domestic demand and the domestic price level from trend. Hence to isolate the policy-induced impact of changes in the interest rate, the decomposition for detrended domestic GDP is computed when the monetary authority’s reaction function is suppressed. That is, the historical decomposition for detrended GDP (y) is computed where the coefficients on forecast inflation and GNE in the domestic interest rate equation are set equal to zero and all other coefficients in the model remain the same. This idea is represented by equation (10).

(10)  y_t^* = initial conditions

where θ^* are the impulse response functions and  y_t^* is detrended GDP when the monetary authority’s reaction function has been suppressed.

The total effect of monetary policy is then found by subtracting equation (10) from equation (9), and then adding back the direct effect of interest rate shocks (that is, the second term in equation (10) corresponding to ( u_10,t-i ) to gain the monetary policy index (MPI):

(11)    MPI_t =

The MPI, represented by equation (11), measures how much monetary policy is adding or subtracting to detrended GDP at each point in time. The first component of equation (11) captures the non-systematic reaction of monetary policy and the second component captures the systematic reaction of monetary policy to shocks (such as an import price shock). The profile for detrended GDP in the absence of a monetary policy response is calculated by subtracting the MPI from detrended GDP (y).[15]

Figure 3 shows detrended GDP (y), the monetary policy index (MPI) and detrended GDP without the impact of monetary policy (y*). When the MPI is above zero, monetary policy is acting to raise detrended GDP. When the MPI is below zero it indicates that monetary policy is depressing detrended GDP (y).

One way to evaluate the impact of monetary policy on the growth cycle is to gauge the degree to which monetary policy has been ‘countercyclical’. There are two dimensions to this. One is to see how closely in time the MPI and detrended GDP without monetary policy (y*) cut the horizontal axis with opposite slope. Ideally the MPI would move above zero when detrended GDP without the influence of monetary policy (y*) moves below zero and vice-versa.

The second dimension is the magnitude to which the MPI moves above zero when detrended GDP without monetary policy (y*) is below zero and vice-versa. If monetary policy had been perfectly countercyclical, the amplitude and frequency of the MPI and detrended GDP without monetary policy (y*) cycles would have been the same but opposite in sign. In this case, detrended GDP (y) would have been at zero (that is, monetary policy would have kept GDP at its trend level). Clearly, this is a stringent criterion by which to evaluate the effectiveness of monetary policy in stabilising GDP (y).

Figure 3 – Monetary policy impact on the growth cycle

From Figure 3 it is apparent there have been periods when monetary policy has been countercyclical and periods when it has been procyclical. Until late 1993 the MPI was generally positive when detrended GDP without monetary policy (y*)was negative and vice-versa, although at times the MPI and detrended GDP without monetary policy (y*)cut the horizontal axis at different points in time. For instance, during the 1991 to 1993 recession monetary policy moderated the depth of the recession. Moreover, at times these offsetting effects are quite large. In 1988 and in 1993 these offsetting effects of monetary policy on the deviation of real GDP from trend were as large as 1 percent per quarter.

However, there are also times during the mid and late 1990s when monetary policy was pro-cyclical. From late 1993 until 1997 monetary policy raised detrended GDP (y), even though y* was positive. This pro-cyclical impact became smaller after 1995 but the impact of monetary policy was not reversed sufficiently quickly to offset the other shocks to GDP between 1993 and 1999. For example, in 1998 the economy was experiencing a growth recession and monetary policy was accentuating the below trend decline in real GDP. This was the period during which the Reserve Bank of New Zealand adopted the Monetary Conditions Index (MCI) as an operating procedure for monetary policy. It was also the period when New Zealand was impacted by a severe drought. However, the impact of monetary policy on detrended GDP during the 1998 growth recession was small compared to that arising from the droughts.

This analysis is an ex post assessment of the impact of monetary policy on the business cycle. It is not an evaluation of the optimality of monetary policy which would require a comparison of the consequences of the monetary policy implementation procedures used by the Reserve Bank with the consequences of using alternative procedures based on ‘real time’ data and judged against the monetary authority’s loss function. The results nevertheless provide support for both the Reserve Bank of New Zealand’s (2000b) and Svensson’s (2001) assessment that monetary policy was “a little slow to recognise the pace of acceleration in 1992/93, and a little slow to recognise the joint impact of the Asian crisis and the first drought through late 1997 and early 1998” (Svensson, p.26). The outcome in 1998 may also reflect problems associated with operating monetary policy using the monetary conditions index that have been raised by Stevens (1998), Guender (2001), Guender and Matheson (2002) and Engelbrecht and Loomes (2002).

Although the structural VAR model captures the deviation of the log of output and the log of the price level from trend, with some modifications the same procedure can be applied to derive the impact of monetary policy on deviations of inflation from trend.[16] The impact of monetary policy on deviations of the domestic price level from trend can be obtained by deriving equation (11) for the domestic price level. The impact of monetary policy on deviations in domestic inflation from trend can also be isolated by recognising the following relationship:

(12)    

where is actual annual inflation in the year to quarter t, is the log of the actual price level, is the log of the trend value of the log of price level and is the deviation of the log of the price level from its trend value.

Recognising that the deviations from trend component comprises two parts, one part generated by monetary policy, denoted, and the other part generated by all other influences, denoted , equation (12) can be rearranged as

(13)     

(14)

The left hand side of equation (14) is the deviation of actual from trend inflation (whereis trend inflation), and the right hand side is the sum of the deviation from trend inflation generated by monetary policy,, and by all other influences in the year to quarter t, .

From equation (11) corresponding to the impact of monetary policy on deviations of the domestic price level from trend,

(15)     

Accordingly, the model generates estimates for the impact of monetary policy on the deviations of actual inflation from trend () and the deviations of inflation from trend generated by all other influences ().

Figure 4 shows the time series plots for the three components of equation (14): the deviation of inflation from its trend rate, the component arising from the effect of monetary policy , and the component arising from all other shocks ( ). The interpretation is similar to Figure 3. When the observations for and are on the same side of the horizontal zero line the interpretation is that monetary policy accentuates inflation variability around trend. When the observations for and are on the opposite sides of the horizontal zero line the interpretation is that monetary policy reduces inflation variability around trend.

Figure 4 – Monetary policy impact on inflation deviations from trend

Figure 4 shows that prior to the introduction of formal inflation targeting there are periods when monetary policy was adding to inflation variability arising from other shocks (for instance during 1984, most of 1987 and 1988). There are also periods when monetary policy was reducing the impact of other shocks on inflation variability (for instance from 1985 to late 1986). The pattern appears to change after the introduction of the 1989 Act until the late 1990s. From 1990 until 1999 the predominant impact of monetary policy was to reduce the impact on inflation variability arising from other shocks. Exceptions were from December 1993 to late 1994, in late 1996, and during the 1998 recession. These periods closely correspond to the periods when monetary policy was also adding to output variability and, in the case of 1998, when the Reserve Bank was using the Monetary Conditions Index as an operating procedure.