BNM's estimated reaction function is found to be considerably dierent for the three sub-periods that are considered. The results are presented in Table 3.2 below.
3.4.1 Pre-capital controls and xed exchange rate regime
period: 1975Q1 to 1986Q4 and 1987Q1 to 1998Q2
There are signicant dierences in the size of the estimated coecients for βπ, Θy and
ρ during the two sub-periods. It is found that the feedback coecient for the interest rate smoothing (ρ) and the feedback coecient on the output gap (Θy) for the 1987-
1998 period decrease notably compared with the estimates for the 1975-1986 period. In contrast, the feedback coecient for ination (βπ) increases considerably during the
Table 3.2: BNM's Estimated Reaction Function: Sub-sample Periods Closed Economy Version of HMT Rule
Contemporaneous Specication 1975Q1-1986Q4 βπ Θy ρ α π∗ Adj. R2 SEE 0.470 0.845 0.858 5.549 4.805 0.824 1.050 (0.405) (0.252) (0.062) (2.352) std. err. (0.253) (0.002) (0.000) (0.023) p-value 1987Q1-1998Q2 βπ Θy ρ α π∗ Adj. R2 SEE 1.491 0.656 0.740 1.692 2.667 0.896 0.635 (0.304) (0.339) (0.109) (0.983) std. err. (0.000) (0.057) (0.000) (0.089) p-value 1987Q1-2005Q2 βπ Θy ρ α ∂βπ ∂Θy ∂ρ ∂α π∗ Adj. R2 SEE 1.491 0.656 0.740 1.692 -0.005 -0.149 -0.445 0.721 1.211 0.936 0.569 (0.304) (0.339) (0.109) (0.983) (0.203) (0.042) (0.214) (0.217) std. err. (0.000) (0.057) (0.000) (0.089) (0.980) (0.001) (0.041) (0.002) p-value Forward-Looking Specication 1975Q1-1986Q4 βπ Θy ρ α π∗ Adj. R2 SEE 0.351 1.125 0.852 6.171 4.883 0.819 1.063 (0.380) (0.463) (0.062) (2.044) std. err. (0.361) (0.019) (0.000) (0.004) p-value 1987Q1-1998Q2 βπ Θy ρ α π∗ Adj. R2 SEE 1.880 0.568 0.674 1.028 3.440 0.913 0.583 (0.391) (0.160) (0.067) (1.411) std. err. (0.000) (0.000) (0.000) (0.984) p-value 1987Q1-2005Q2 βπ Θy ρ α ∂βπ ∂Θy ∂ρ ∂α π∗ Adj. R2 SEE 1.888 0.546 0.679 1.070 -0.057 -0.279 -0.210 1.084 1.732 0.941 0.547 (0.380) (0.177) (0.075) (1.432) (0.210) (0.075) (0.064) (0.484) std. err. (0.000) (0.000) (0.000) (0.978) (0.393) (0.000) (0.015) (0.040) p-value
the movement of interest rates during the 1987-1998 period. The dierence is also seen on the policymakers' implicit ination target. During the later period, monetary policy aimed to achieve a lower ination target.
Lower interest rate smoothing behaviour
The lower estimate for the smoothing coecient (ρ) for the 1987-1998 period suggests that monetary policy had been used more aggressively as the demand management tool during this period. The main reason to explain this outcome is the improvement in the domestic nancial landscape and infrastructure. This condition provided BNM with greater exibility to maneuver monetary policy, with less concern about the negative impact of such action on the overall stability of the nancial system. For example, the successful deregulation of the interest rate regime for the banking institutions in the late 1970s, provided the Malaysian banking system with greater exibility and resilience. Together with the progression and advancement of the Malaysian nancial system that permits a more ecient working of the money and capital markets to allocate nancial resources, the importance of monetary policy has also become more important and dominant over time. This translated into a lower interest rate smoothing behaviour.
Lower feedback coecient on the output gap
While there is evidence of more aggressive use of monetary policy during the later period, the estimated reaction function also indicates that BNM was less responsive to the development in the output gap (Θy). This indicates that, as Malaysia's economy
became more developed, BNM was more cautious in its action responding to the output gap.
BNM's cautious behaviour could be attributed to the increasing diculty for the policymaker to assess the true state of the Malaysian economy. With the rapid and sustained expansion of GDP experienced during the post-1987 period, together with the continuing evolution in the economic and nancial condition of the Malaysian economy, the policymakers' task to get the clear picture on the economic conditions becomes more challenging. To give a snapshot on the dynamic changes of the Malaysian economy, Malaysia experienced a large increase in foreign direct investment (FDI).
The total amount of FDI inow during the 1975-1986 period was USD9.048 billion and in 1987-1998, this amount increased signicantly to USD47.072 billion.10 The large
inow of FDI during this short period of time contributed to the rapid expansion of the Malaysian economy. During this period, Malaysian GDP grew at the average rate of 9% annually. Hence, the dynamic economic progress during this short span of time gave the central bank a great challenge to evaluate the true state of the economy before making monetary policy decisions.
The result of the estimated reaction function suggests that BNM has reacted to this increasing uncertainty by putting smaller weight on the output gap. Given that the structure of the Malaysian economy changed considerably and experienced rapid growth during this period, BNM's cautious behaviour as inferred from this result is sensible and consistent with the famous proposition made by Brainard (1967). Faced with multiplicative uncertainty, in deciding policy actions, the central bank should be on the cautious side. As Alan Blinder, former member of Federal Reserve Board, put it on this issue:
. . . .estimate how much you need to tighten or loosen monetary policy to `get it right'. Then do less. Blinder (1998)
The results of the estimated reaction function suggests BNM did just that from 1987 onwards.
The observation that in the face of increasing diculty to asses the true economic condition, BNM reacted less forcefully to the output gap (↓ Θy) but showed less
tendency to smooth interest rates (↓ ρ), is quite interesting. Normally, we would expect, faced with increase policy uncertainty, BNM to become more cautious in its overall policy action and this would translate into a higher interest rate smoothing. This turns out not to be the case. During the 1987-1998 period, BNM demonstrates its precautionary policy stance by responding less to the output gap but not to the willingness to change interest rate. There could be two reasons for this policy beha- viour. One, as we mentioned before, from 1987 onwards BNM operated in a more resilient banking system that allows it to use monetary policy more eectively as one of the tools in the macroeconomic demand management. Second, this could reect the
outcome from policymakers' strategy at that time to address the inationary pressure in Malaysia. During that period, inationary pressure was dominantly emanated from the supply constraint to keep up with the increase demand. The continuous expansion of the economy from 1987 onwards resulted the Malaysian economy to operate at the full employment level by 1992 (Bank Negara Malaysia (1994b)). Instead of focusing its policy action to slowdown the demand side, Malaysia's policymakers (BNM and other Government agencies like the Treasury and the Economic Planning Unit) took a dierent approach to tackle this issue. They tried to address the supply constraint by encouraging more capacity building. Thus, to encourage investment, particularly from the private sector, monetary policy (and interest rate) was made more accommodative when responding to the development in the output gap. As we will see in Chapter 4, this policy action is consistent with BNM's policy preferences. Compared to the 1975- 1986 period, BNM increases its policy preference to stabilize the output gap during the later period.
Higher feedback coecient on the ination outlook
The estimation results reveal that the coecient for βπ, (the feedback coecient for
ination) during the period of 1975-1986 is small and is not statistically signicant. Interestingly, BNM's estimated reaction function for this period was not really consis- tent with the behaviour predicted by the HMT-type interest rate rule in the way the central bank articulates its interest rate decision. Even though the simple HMT rule ts the data fairly well (with the adjusted R2 for 1975-1986 period is around 0.8), the estimation results indicate that the ination factor was not statistically signicant in inuencing the movement of interest rates during this period. Stabilizing the output gap was the dominant factor that contributed to BNM's action to change the interest rate.
Does this result imply that the simple HMT-rule fails to represent BNM's reaction function for this period? While this is one of the possibilities, the fact that the simple HMT rule is able to track the actual data well suggests the high possibility for this rule to be a valid representation of BNM's reaction functions for the 1975-1986 period. Hence, the result could indicate possible inconsistencies on the part of BNM's behaviour at that time, in particular its failure in setting the interest rates to be in line with the Taylor's principle (i.e. to set βπ >1). On this note, the simple HMT-rule ts the data
Figure 3.1: Malaysia's GDP Growth 1975Q1-2005Q2 (quarterly, seasonally adjusted, %)
-12 -8 -4 0 4 8 12 1975 1980 1985 1990 1995 2000 2005 % 1975Q1 to 1986:Q4 1987Q1 to 1998:Q2 1998Q3 to 2005:Q2 Commodity Price Collapse
1997 Asian Financial Crisis
post- Sept. 11, 2001 events
Aftermath of First Oil Price Shock 1973-74
1975Q1-1986Q4 1987Q1-1998Q2 1998Q3-2005Q2 1975Q1-2005Q2
Mean 5.91 9.09 4.77 6.45
Std. Dev. 4.49 2.43 5.42 4.65
much better during the later periods, with the estimated coecients for βπ well above
1 and statistically signicant.
This result highlights two important observations on BNM's behaviour in formulating monetary policy in Malaysia. Firstly, it indicates that the formulation of monetary policy during the 1975-1986 period was geared towards achieving output stability, with little consideration to the ination outlook. During this period, the growth of Malaysia's GDP was more volatile (see Figure 3.1 for details). GDP growth was badly aected by the oil price shocks (early and end-1970's) and the collapse of commodity prices in the international market (mid-1980s). Perhaps, these factors prompted BNM to tempo- rarily sacrice its price stability objective and focus more towards stabilizing the output uctuations. Secondly, BNM's failure to comply with the Taylor's principle during this period caused a temporary increase in the level and volatility of the ination rate in Malaysia. During 1975-1986, it is found that the ination rate was much higher and volatile than what was experienced during the later period. While higher domestic ina- tion during the period could be attributed to the eects of supply shock experienced by the Malaysian economy following the rst (1973) and the second (1979) oil price
Figure 3.2: Malaysia's Ination Rate 1975Q1-2005Q2 (quarterly, seasonally adjusted, %)
-1 1 3 5 7 9 11 1975 1980 1985 1990 1995 2000 2005 % 1975Q1 to 1986Q4 1987Q1 to 1998Q2 1998Q3 to 2005Q2 Aftermath of First Oil Price Shock 1973-74
Second Oil Price Shock 1979-80
1997 Asian Financial Crisis 1991-93 Capacity Constraint 1975Q1-1986Q4 1987Q1-1998Q2 1998Q3-2005Q2 1975Q1-2005Q2 Mean 4.28 3.23 2.02 3.37 Std. Dev. 2.62 1.23 1.19 2.08 Minimum -0.29 0 0.87 -0.29 Maximum 10.54 5.73 5.67 10.54
shocks, failure of BNM to set interest rates by responding to the ination outlook could be the amplifying factor contributing to this high ination experience. In particular, BNM's failure to set interest rates by reacting more than one-to-one to ination could have exacerbated the impact of the oil price shocks on the domestic price. See table and graph in Figure 3.2 for details. This outcome is in-line with the theoretical ndings advocated by Taylor (1999b) and Woodford (2001) about the importance of an interest rate rule to fulll Taylor's principle in order to ensure price stability. Failure to do so, leads to the self-fullling inationary process to take eect. In the Malaysian case during this episode, the supply shocks from the higher oil price triggered the upward adjustment on the price of general goods, as well as the inationary expectation among economic agents. These factors contributed to the higher domestic ination. When nominal interest rates failed to adjust more than one-to-one to changes in the ination rate, the reduction in the real interest rates pushed up the aggregate demand through the interest rate channel. The higher aggregate demand put further upward pressure on the general price level. Hence, the price level continued to be adjusted upwards.
As a result, ination during 1975-1986 remained high for an extended period and peaked at 10.5% in Q2 1984.
In contrast, ination rate during the 1987-1998 period was much lower and relatively stable. During the estimation period, the estimated coecient for βπ of BNM's reaction
function is well above 1 and is statistically signicant. Despite the Malaysian economy growing at a rapid rate during this period with GDP expanding at an average annual rate of 9% - the ination rate remained low and stable. Again, the stable ination outcome during this period was attributed to BNM's sound conduct of monetary policy by setting interest rates according to Taylor's principle.
The benet of this sound policy to Malaysia's economy was considerable. Ination during the later period was much lower and more stable. The volatility of the ination rate declined substantially, with its standard deviation reduced to just 1.232 (2.624 for 1975-1986). In addition, the benet from stable ination is prosperous economic growth. During the period, Malaysia experienced the longest economic boom in its history. GDP grew at an average rate of 9% with a much lower volatility. This outcome is consistent with the argument put forward by Clarida, Gali, and Gertler (2000) - aggressive monetary policy to ght ination works to reduce aggregate output volatility by eliminating sunspot equilibria.
Lower ination target
The estimated reaction function across the two time periods also reveals that there is a signicant decline in the BNM's implicit ination target, π. Using the respective estimates for Θy and βπ of the two sub-samples, the ination target during the later
period is derived to be between 2.7-3.4% (about 4.8% for the 1975-1986 period). This decline may indicate the change in BNM's policy objective. In-line with the overall improvement in Malaysia's real-sector economy during the 1987-1998 period, monetary policy was indeed implemented to achieve a lower ination target.
3.4.2 Post-capital controls and xed exchange rate regime
period: 1998Q3 to 2005Q2
From the results in Table 3.2, there is strong evidence to suggest that BNM's reaction function changes considerably during the engagement of this regime. The estimated coecient on the interactive dummy for output gap (∂Θy) and interest rate smoothing
(∂ρ) are found to be statistically signicant, indicating changes in BNM's responses to stabilize output and the tendency to change its policy stance. The estimated coecient on the interactive dummy for ination (∂βπ) is not signicant; suggesting that the
policymakers' concern on the ination outlook in formulating monetary policy remained unchanged when the regime was in place. Nevertheless, there is evidence that BNM reduced its ination target to around 1.2-1.7% (from around 2.7-3.4% previously). The estimated coecient for ∂Θy is negative, suggesting that BNM's feedback reaction
to stabilize the output gap decreases during the period. At rst glance, this result may seem to be counter-intuitive, given the fact that the main aim of the introduction of this regime was to revive the Malaysian economy. With the Malaysian economy at that time operating well below its potential level, it is expected that BNM's concern to stabilize output would have increased. Nevertheless, this result is indeed consistent with the downward trend observed earlier regarding BNM's feedback coecient on output. As found earlier, the estimated coecient for Θy declines in the 1987-1998 period and this
trend continues when the capital control and xed exchange rate regime was introduced. Again, a possible reason to explain the decline in BNM's feedback coecient to stabilize output is the increasing uncertainty on the Malaysian economic structure following the 1997 Asian Financial Crisis and the subsequent introduction of capital controls and xed exchange rate regime. The impact of these unprecedented and short-lived events to the Malaysian economy must have been considerable and perhaps unknown to BNM at that point of time. With the increased uncertainty of the impact of such events to the real sector, the result suggests that BNM acted cautiously by reacting less forcefully to the output gap.
In contrast, the estimated coecient for ∂ρ is also negative, suggesting BNM's reduced tendency to smooth the interest rates. The lower interest rate smoothing coecient on its estimated reaction function indicated the changing behaviour of BNM's during the period to change its policy rate more forcefully than the manner showed before
the regime was introduced. This also suggests that monetary policy was used more aggressively during the period as part of the key strategies to revive the Malaysian economy on the aftermath of the Asian Financial Crisis. This result is not surprising, given the fact that one of the main objectives of capital control that was introduced at this time was to provide BNM with autonomy to conduct independent monetary policy. Despite imposing a xed exchange rate regime by pegging the ringgit to the US dollar, the imposition of capital control allowed BNM to pursue an independent monetary policy without being dictated by the US's monetary policy. The result of BNM's reaction function for this period indicated that BNM took full advantage of the `window opportunity' provided by the capital control, by actually changing its policy rate more aggressively. From 1998Q3 onwards, BNM drastically loosened its monetary stance. Between 1998Q3 to 1999Q4, BNM reduced its policy rate 7 times. Within that 13-months period, interest rates declined from 11% in June 1998 to 5% at end-1999.
3.5 Conclusion
This chapter established a few important features about the way Bank Negara Malaysia (BNM) formulates monetary policy in Malaysia.
First, modeling BNM's reaction function by using a simple HMT-type interest rate rule is able to track the interest rate movement generally well. As it is being commonly used to model the reaction function of central banks in various developed and developing countries, this exercise indicates that a similar approach also works equally well in summarizing the central bank's behaviour in Malaysia.
Second, the estimated reaction function with the backward looking specication performs poorly in tracking BNM's past behaviour. This provides empirical evidence that interest rate movements in Malaysia have been largely inuenced by current and future economic development. Hence, it is very likely that BNM has always adopted a forward-looking approach in its policy making.
Third, the exchange rate factor is not statistically signicant in explaining the movement of BNM's reaction function. Despite the fact that BNM does smooth exchange rate movement, this exercise failed to capture any empirical evidence to esta- blish a relationship between BNM's interest rate policy and the movement of exchange
rates. This indicates BNM does not directly use its interest rate policy to stabilize exchange rates. Instead, this outcome is achieved mainly through BNM inuencing the movement of the nominal exchange rate by adopting a managed oat exchange rate regime. In doing so, BNM intervenes directly in the foreign exchange market and does not directly use interest rates as the instrument.
Fourth, BNM's behaviour in setting interest rates changed considerably over time. The change in behaviour was triggered by the continuous development and progression of the Malaysian economy, as well as the change in the policy landscape following the introduction of the capital control and xed exchange rate regime in September 1998.