Abstract
This Selected Issues paper for Indonesia uses a small structural macroeconomics model of the Indonesian economy to analyze the inflation outlook and monetary policy challenges. The Bank of Indonesia (BI) introduced its Inflation Targeting Framework in July 2005 with the goal to reduce inflation in the medium term to 3 percent. BI’s official mandate is stability of the rupiah, both internal and external, and BI views the inflation targeting regime with a floating exchange rate as the best strategy to fulfill that mandate.
I. Inflation Outlook and Monetary Policy Challenges: A Model–Based Analysis1
A. Introduction
1. Rising inflationary pressures are posing a major challenge for monetary policy in Indonesia. Driven by rising food prices, strong economic activity, declining real interest rates, and a slightly weaker rupiah, inflationary pressures built up since mid-2007. While inflation remained within the target range at end-2007, it has been accelerating in 2008. Meanwhile, rising oil prices increased pressures to adjust domestic fuel prices, and the resulting 29 percent adjustment in May further raised inflation which reached 11.0 percent in June. As a result, Bank Indonesia (BI) has been gradually tightening monetary policy since May, but uncertainty about the global environment and the mix between demand and supply factors as sources of inflationary pressures pose significant challenges to monetary policy making at this juncture.
2. This chapter uses a small structural macro model of the Indonesian economy to analyze the inflation outlook and monetary policy challenges. The model is a version of the IMF’s forecasting and policy analysis system (FPAS) model which has been modified to fit key features of the Indonesian economy. In analyzing the inflation outlook and policy options, emphasis is given to specific risks arising from the external environment as well as possible domestic shocks. The remainder of the chapter is structured as follows: Section B provides background and outlines current conditions, Section C briefly describes the model, Section D discusses the baseline forecast and alternative policy and risk scenarios, and Section E concludes.
B. Background and Current Conditions
3. BI introduced its Inflation Targeting Framework in July 2005 with the goal to reduce inflation in the medium term to 3 percent. BI’s official mandate is stability of the rupiah, both internal and external, and BI views the inflation targeting regime with a floating exchange rate as the best strategy to fulfill that mandate. Under this framework, the primary objective is achieving the inflation target. In practice, however, BI also aims to support economic growth and at times dampen the volatility of the exchange rate. Specific annual targets are set by the government in coordination with BI. The most recent targets are 5±1 percent for 2008, 4.5±1 percent for 2009, and 4±1 percent for 2010, although the 2009–10 targets might be adjusted following 2008 overshooting. At this point BI does not publish official multi-year inflation forecasts.
4. The framework has been very successful in reducing inflation following the fuel price shock in October 2005. Sharp rate hikes in 2005 helped limit second round effects of the fuel price increases.2 Inflation stabilized after the initial spike and fell below the (revised) target range by end-2006 (Figure 1). This allowed BI to gradually ease monetary policy which helped stimulate credit growth and domestic demand. After falling to 5.5 percent in 2006, GDP growth accelerated again reaching 6.3 percent in 2007 and Q1 2008—the highest annual rate since the Asian crisis.
Inflation and Interest Rate Developments
Citation: IMF Staff Country Reports 2008, 298; 10.5089/9781451818420.002.A001
Inflation and Interest Rate Developments
Citation: IMF Staff Country Reports 2008, 298; 10.5089/9781451818420.002.A001
Inflation and Interest Rate Developments
Citation: IMF Staff Country Reports 2008, 298; 10.5089/9781451818420.002.A001
5. However, inflation started increasing again in mid-2007. High food prices played an important role given their large weight in the CPI index (36 percent in the recently rebased basket), but this has been largely offset by stable administered prices, especially fuel prices, which remained unchanged between October 2005 and May 2008. Nevertheless, broader price pressures started to build, as evidenced in the gradual increase of core inflation.3 Several factors contributed to this development: (1) strong growth and falling unemployment supported by low real interest rates and accelerating credit growth, (2) the relative weakness of the rupiah compared with other regional currencies (resulting in an 11 percent depreciation of the nominal effective exchange rate between June 2007 and April 2008), and (3) some pass-through of high commodity and food prices. BI reacted by keeping rates on hold for most of the second half of 2007. However, with inflation close to the end-2007 target, BI cut interest rates one more time in December by 25 basis points.
6. Inflationary pressures accelerated further in 2008. Food prices rose to new highs, while economic activity remained strong and credit growth continued to accelerate. With headline and core inflation moving further away from the end-2008 target, BI reversed its easing bias, started raising rates in May by 25 bps. The 29 percent increase in fuel prices in late May further added to headline inflation (which reached 11.0 percent in June) and BI raised rates by another 25 bps in June.4
7. The key challenge for BI is now to bring inflation back on a declining path towards the medium-term inflation target. With growth likely close to potential, credit growth exceeding 30 percent annually, and real interest rates in negative territory, this will require rate increases, and early action could help to anchor inflationary expectations. However, BI needs to strike a careful balance. Risks to growth are slanted to the downside considering the global slowdown and continuing global financial market problems. In addition, the high volatility of food and energy prices and continuing pressures to reduce fuel subsidies over the medium term are posing additional challenges.
C. Brief Description of the Model
8. The FPAS model is a small system of forward-looking structural equations that describe the relation between a several key macro variables.5 The model allows for rational, as well as adaptive expectations. It is a “two-country” model, with Indonesia as the home country and the rest of the world proxied by the U.S.6 At the heart of the model are four equations that determine endogenously the main macro variables output, inflation, exchange rate, and interest rates:
IS curve: or aggregate demand curve that relates the level of real activity to expected and past real activity, the real interest rate, and the real exchange rate.
Phillips curve: a price-setting curve that relates inflation to past and expected inflation, the output gap, and the exchange rate.
Exchange equation: an uncovered interest parity condition, with some allowance for backward-looking expectations and incomplete adjustment of the exchange rate to interest rate differentials
Monetary policy rule: a Taylor rule-type equation that sets the policy interest rate as a function of the output gap and expected inflation, with some interest rate smoothing.
9. Several extensions have been made to the basic model to better fit the Indonesian economy. Considering the importance of regulated fuel prices in Indonesia, an equation has been added that describes the evolution of domestic fuel prices, assuming periodic adjustments with a view to keep the subsidy from exceeding a certain target level. Separate Phillips curves are introduced for headline and core inflation with fuel prices directly affecting headline, and some pass-through from headline to core. The real interest rate affecting aggregate demand is based on core inflation, while monetary policy is assumed to target an average of headline and core.7
10. Despite the parsimonious structure, the model captures the main monetary transmission channels. Monetary policy affects inflation in three distinct ways. First, interest rates directly affect aggregate demand which in turn affects inflationary pressures. While not explicitly modeled in this paper, this could work through borrowing costs, the availability of financing, or asset prices. The second key channel is through the exchange rate. Interest rates directly influence the exchange rate through the yield differential, which then has a pass-through effect on inflation. And finally, monetary policy—in particular the policy reaction function and future inflation targets—can affect inflationary expectations which have a direct effect on contemporaneous inflation.
11. However, due to the parsimonious structure, several important variables have to be determined exogenously, outside of the model. The key variables are expressed in terms of their deviations from equilibrium, in other words in “gap” terms. The model itself does not attempt to explain movements in equilibrium real output, the real exchange rate, or the real interest rate, or in the inflation target. Rather, these are taken as given from various sources employing filtering methodologies or using judgment and views about these equilibrium values.
12. The model parameters have been estimated using Bayesian techniques. Estimation of the model is complicated by the major structural changes associated with the Asian crisis. To ensure that model parameters are not affected by the crisis period a relatively short sample is used (Q1 2000 to Q1 2008). Potential output growth is assumed to be 6.3 percent currently and rise to 6.7 percent over the medium term as a result of higher investment rates and improvements in the investment climate. The historic output gap and equilibrium real exchange rate are derived by HP filter. Implied inflation targets for the pre IT period (2000–04) are based on inflation projections from past IMF staff reports. Priors are based on similar models estimated for other countries, as well as on other empirical work on Indonesia and judgment. Key results of the estimation are: a neutral real interest rate slightly above 3 percent (despite a lower prior of 2.5 percent), a sacrifice ratio of about 0.9 percent (indicating that a permanent reduction in inflation by 1 percentage point will require a cumulative loss of output of 0.9 percent of annual GDP), and an exchange rate pass-through somewhat below 10 percent after one year assuming no fuel price adjustment and endogenous monetary tightening. 8, 9
D. Simulation Results10
Model Forecast
13. The baseline model forecast suggests that substantial rate increases will be needed to bring inflation back on a declining path toward the medium-term target (Figure 2). A tightening cycle, with rates peaking in Q2 2009, would result in significantly positive real interest rates (forward-looking) and help contain the pass-through from headline to core inflation. The exchange rate should appreciate, in line with contemporaneous and expected future interest differentials, contributing to the dampening trend on inflation. Core inflation would decline gradually over the medium term, reaching about 4 percent by end-2012. Assuming no further fuel price adjustments until 2010, headline inflation would fall below core inflation in 2009 and end the year at about 6.5 percent. However, this assumes no additional price pressures from food prices.11
Model Forecast
Citation: IMF Staff Country Reports 2008, 298; 10.5089/9781451818420.002.A001
Model Forecast
Citation: IMF Staff Country Reports 2008, 298; 10.5089/9781451818420.002.A001
Model Forecast
Citation: IMF Staff Country Reports 2008, 298; 10.5089/9781451818420.002.A001
14. Higher real interest rates and a more appreciated exchange rate would cause output growth to slow. The output gap would open gradually, keeping growth below the assumed potential growth rate of 6.3 percent in 2008. In this simulation, growth would average about 5.7 percent in 2009. Over the medium term, with a gradually closing output gap, growth would accelerate again. However, as this medium-term scenario envisages substantial disinflation, output would have to remain below potential for extended period.
15. Despite tight monetary policy, however, headline inflation could remain above 6 percent for quite some time. The assumption of gradual fuel price adjustments starting in 2010 would result in some upward pressure on headline inflation, with some pass-through to core. Moreover, this baseline scenario assumes a policy reaction broadly in line with BI’s past behavior, suggesting that moderate deviations from the target would likely be accepted, if needed to prevent sharper declines in output.
Scenario 1: Slower Policy Response
16. A slower pace of monetary tightening would support growth in the near-term, but result in significantly higher inflation over the medium term (Figure 3). This scenario assumes that BI is less ambitious regarding its inflation objective, with the goal to prevent a significant drop in growth. With the policy rate rising only to 9.5 percent by end-2008, output would stay at or above 6 percent. However, real rates would remain low and the rupiah would fail to appreciate as in the baseline scenario. The result would be higher inflation in the medium term. However, interest rates would still have to be raised significantly in order to maintain real rates consistent with a gradually declining inflation.
Scenario 1: Slower Policy Response
Citation: IMF Staff Country Reports 2008, 298; 10.5089/9781451818420.002.A001
Scenario 1: Slower Policy Response
Citation: IMF Staff Country Reports 2008, 298; 10.5089/9781451818420.002.A001
Scenario 1: Slower Policy Response
Citation: IMF Staff Country Reports 2008, 298; 10.5089/9781451818420.002.A001
17. However, these results are sensitive to the estimate of the equilibrium real interest rate. This is the real rate that is neutral in the sense that it neither leads to an acceleration of inflation nor a deceleration. The estimation yielded a value slightly above 3 percent, despite a lower prior. However, estimating this rate is difficult and it may in fact change over time. If the equilibrium real rate were 1 percentage point lower than in the baseline simulation, the policy rate could also remain 1 percentage point lower, while achieving the same results with respect to inflation, output, and the exchange rate. In this case, a slower pace of tightening would be sufficient.
Scenario 2: Continued Rise in Food Prices
18. Food prices continue to pose a significant risk to the near-term inflation outlook. This scenario considers a further increase in food prices, similar to what has been observed over the past year. Specifically, it assumes a 10 percent rise in the relative price of volatile food items over the course of 2 quarters. While the model does not explicitly include food prices, this shock can be simulated as a shock to headline inflation. The cumulative shock to headline inflation would be 2 percentage points. With an approximately 50 bps increase in policy rates, the pass-through to core inflation should remain limited to about 50 bps and the effect on headline inflation should be relatively short-lived. However, the slowdown in output would be somewhat more pronounced.
Scenario 2: Continued Rise in Food Prices
Citation: IMF Staff Country Reports 2008, 298; 10.5089/9781451818420.002.A001
Scenario 2: Continued Rise in Food Prices
Citation: IMF Staff Country Reports 2008, 298; 10.5089/9781451818420.002.A001
Scenario 2: Continued Rise in Food Prices
Citation: IMF Staff Country Reports 2008, 298; 10.5089/9781451818420.002.A001
Scenario 3: Increased Risk Premium and Exchange Rate Depreciation
19. The key result of this adverse shock scenario is that a significant depreciation of the rupiah resulting from capital outflows would require an aggressive policy response. This scenario assumes a sharp increase in the required risk premium, resulting in a 15 percent depreciation relative to the baseline. To limit the pass-through to inflation and prevent sharper depreciation, policy rates would be raised by an additional 250 bps. The decisive policy action would limit the impact on inflation. However, inflation would still run about 1½ to 2 percentage points higher compared to the baseline after about one year. Output would be slightly more volatile than under the baseline, with the initial expansionary effect of the depreciation being offset later by higher interest rates.
Scenario 3: Surge in Risk Premium—Exchange Rate Depreciation
Citation: IMF Staff Country Reports 2008, 298; 10.5089/9781451818420.002.A001
Scenario 3: Surge in Risk Premium—Exchange Rate Depreciation
Citation: IMF Staff Country Reports 2008, 298; 10.5089/9781451818420.002.A001
Scenario 3: Surge in Risk Premium—Exchange Rate Depreciation
Citation: IMF Staff Country Reports 2008, 298; 10.5089/9781451818420.002.A001
E. Conclusions
20. While the results need to be interpreted carefully, the model-based analysis can provide valuable inputs for inflation forecasting and monetary policy making. It provides a consistent quantitative framework to analyze policy options and risks, which takes into account dynamic interactions between shocks, policy reaction, and forward-looking expectations. Moreover, it does so in a simple structural framework that can be easily interpreted. However, simplification also means that some important factors are not considered within the framework. This includes in particular a more detailed analysis of supply-side factors, the external, financial, and fiscal sectors.
21. The baseline simulation suggests that significant monetary tightening is needed to contain inflationary pressures. Expectations about future monetary policy play a key role in the transmission of monetary policy, both for anchoring inflation expectations and for determining the exchange rate. While there is uncertainty about the magnitude of the needed rate hikes, the model simulations suggest an increase in the policy rate to at least 10½ percent at the peak of the cycle. This would be consistent with the estimated policy reaction function and should ensure that inflation is put back on a firmly declining path. Of course, the path and the extent of the rate increases will have to be reconsidered in line with future inflation and economic developments.
22. Decisive action early on would have significant medium-term benefits for inflation and would create room to lower rates at an earlier stage. An excessively slow pace of rate increases would likely result in a loss of credibility and would make achievement of the medium-term target more difficult. Moreover, it would only delay but not prevent the need for further rate hikes. On the other hand, the risk of too aggressive rate hikes seems small at this point, considering that significant rate hikes would be needed even under more benign assumptions.
23. Key risks emanate from food prices and the external environment. Food prices continue to pose a significant risk in the near term. However, effects should generally be short-lived and the policy reaction should thus be less aggressive. On the other hand, a sharp loss of confidence, resulting in a significantly weaker exchange rate would require a much more aggressive response.
Appendix: Model Equations and Parameters
Key Model Equations:
Phillips Curve:
Phillips Curve (Core Inflation):
IS Curve:
Monetary Policy Reaction Function:
Interest Parity Condition:
Fuel Price Adjustment:
Variables:
| ygap | Output gap |
| ygapus | US output gap |
| i | Nominal interest rate |
| π | Headline inflation (one quarter at annual rate) |
| πc | Core inflation (one quarter at annual rate) |
| π4 | Headline inflation (past 4 quarters) |
| πc4 | Core inflation (past 4 quarters) |
| π* | Inflation target |
| r | Real interest rate |
| rus | US real interest rate |
| π_rp_fuel | Rate of change of the relative price of domestic fuel |
| z | Real exchange rate (in logs) |
| ze | Expected real exchange rate |
| ρ* | Risk premium |
| pfuel | Domestic fuel price (in logs) |
| poil | Global oil price in domestic currency (in logs) |
| sub* | Targeted fuel price subsidy (log difference between global and domestic price) |
| ygap | Output gap |
| ygapus | US output gap |
| i | Nominal interest rate |
| π | Headline inflation (one quarter at annual rate) |
| πc | Core inflation (one quarter at annual rate) |
| π4 | Headline inflation (past 4 quarters) |
| πc4 | Core inflation (past 4 quarters) |
| π* | Inflation target |
| r | Real interest rate |
| rus | US real interest rate |
| π_rp_fuel | Rate of change of the relative price of domestic fuel |
| z | Real exchange rate (in logs) |
| ze | Expected real exchange rate |
| ρ* | Risk premium |
| pfuel | Domestic fuel price (in logs) |
| poil | Global oil price in domestic currency (in logs) |
| sub* | Targeted fuel price subsidy (log difference between global and domestic price) |
Key Parameters Values:
| απ | 0.1841 |
| αygap | 0.2851 |
| αzgap | 0.1937 |
| αfuel1 | 0.1173 |
| αfuel2 | 0.0223 |
| αheadline | 0.4110 |
| γlag | 0.6004 |
| γπ | 1.1791 |
| γygap | 0.3960 |
| φfuel | 0.925 |
| βld | 0.0761 |
| βlag | 0.6014 |
| βUS | 0.1234 |
| βr | 0.0827 |
| βz | 0.0175 |
| βfuel1 | 0.0040 |
| βfuel2 | 0.0019 |
| θ | 0.4286 |
| δld | 0.5173 |
| r* | 3.15 |
| απ | 0.1841 |
| αygap | 0.2851 |
| αzgap | 0.1937 |
| αfuel1 | 0.1173 |
| αfuel2 | 0.0223 |
| αheadline | 0.4110 |
| γlag | 0.6004 |
| γπ | 1.1791 |
| γygap | 0.3960 |
| φfuel | 0.925 |
| βld | 0.0761 |
| βlag | 0.6014 |
| βUS | 0.1234 |
| βr | 0.0827 |
| βz | 0.0175 |
| βfuel1 | 0.0040 |
| βfuel2 | 0.0019 |
| θ | 0.4286 |
| δld | 0.5173 |
| r* | 3.15 |
References
Berg, A., P. Karam and D. Laxton, 2006a, “A Practical Model-Based Approach to Monetary Policy Analysis: Overview,” IMF Working Paper No. 06/80 (Washington: International Monetary Fund).
Berg, A., P. Karam and D. Laxton, 2006b, “A Practical Model-Based Approach to Monetary Policy Analysis: An How-To Guide,” IMF Working Paper No. 06/81 (Washington: International Monetary Fund).
Choudhri, Ehsan U., and Dalia S. Hakura, 2006, Exchange Rate Pass-through to Domestic Prices: Does the Inflationary Environment Matter?, Journal of International Money and Finance, Vol. 25, Issue 4, June 2006, pp. 614–39.
Epstein, N., P. Karam, D. Laxton, and D. Rose, 2006, “A Simple Forecasting and Policy Analysis System for Israel: Structure and Applications,” in Israel: Selected Issues, IMF Staff Country Report No. 06/121 (Washington).
Harjes, T. and L. Ricci, 2007, “A Quantitative Analysis of Inflation Dynamics in South Africa,” in South Africa: Selected Issues, IMF Staff Country Report No. 05/345 (Washington).
Ito, Takatoshi and Kiyotaka Sato, 2007, “Exchange Rate Pass-Through and Domestic Inflation: A Comparison between East Asia and Latin American Countries,” Discussion Paper No. 07040, Research Institute of Economy, Trade and Industry (RIETI).
Laxton, Douglas and Alasdair Scott, 2000, “On Developing a Structured Forecasting and Policy Analysis System Designed to Support Inflation-Forecast-Targeting,” in Inflation Targeting Experiences: England, Finland, Poland, Mexico, Brazil, Chile, (Ankara: The Central Bank of Turkey), pp. 6–63.
Steinberg, C., 2008, “Nigerian Inflation: A Structural Model for Policy Analysis,” in Nigeria: Selected Issues, IMF Staff Country Report No. 08/65 (Washington).
Tiffin, A., 2007, “Modeling Monetary Policy in Romania,” in Romania: Selected Issues, IMF Staff Country Report No. 07/220 (Washington).
Prepared by Steffen Reichold (APD).
Prices for gasoline were raised by 30 percent in March 2005, followed by an 88 percent increase in October. Similar price increases took place for other types of fuel. In addition, subsidies were removed for industrial users.
Indonesia’s measure of core inflation excludes volatile food prices (currently accounting for about 20 percent of the CPI) and administered prices, such as fuel prices, transportation tariffs, electricity, and tobacco among others (currently about 28 percent of the CPI). The numbers refer to the 2002 base year CPI as core inflation measures have not yet been released under the recently rebased CPI.
Under the old CPI weights which were in effect until May 2008, measured inflation would have been significantly higher. While precise numbers are not available, an estimate based on monthly inflation in the 7 main CPI categories results in 12.6 percent y/y.
See Berg, Karam, and Laxton (2006a) and (2006b) for a more detailed description of the basic model.
In the simulations, US variables are taken as exogenously given, based on the latest WEO forecast.
While Indonesia’s official target is on headline inflation, BI is assumed to react less aggressively to shocks to food and administered prices than to broad inflationary shocks reflected in core inflation. This behavior can be modeled by reducing the weight of headline inflation in the policy reaction function, and raising the weight ofcore inflation.
Estimating the equilibrium real interest rate is difficult and estimates depend on the sample period. Moreover, the equilibrium rate may change over time. As real interest rates were relatively high during 2000–02, a shorter sample period would likely result in a lower estimate. Thus the current neutral real interest rate could be somewhat lower than the estimated 3.15 percent, consistent with a lower risk premium compared to the earlier post-crisis years. However, the model projections in this paper are based on the formal estimate over the full 2000–08 sample.
The estimated pass-through is low compared to previous studies that find values around 40 percent after one year (see Choudhri and Hakura (2006) and Ito and Sato (2007)). However, those studies include the crisis years which were characterized by a high exchange rate pass-through and the estimates were affected by periodic fuel price adjustments. Assuming proportionate fuel price adjustments, the pass-through would be significantly higher than 10 percent in the estimated model.
Model simulations start in Q4 2008. Q2 and Q3 are based on a near-term judgment forecast that takes into account a broader set of available data, short-term indicators, and projections.
As food prices are not explicitly included in the model, the implicit assumption is that food inflation is in line with underlying core inflation. In other words, relative prices of food are assumed to remain at the current level.
