A. Small Open Economy GPM Model

3. The analysis is conducted with the use of a modified version of the small “New Keynesian” macroeconomic model of Berg, Karam, and Laxton (2006). The model is a stripped-down version of a stochastic general equilibrium (DSGE) model with rational expectations. By virtue of their relatively simple structure, small New Keynesian models have been used for forecasting and policy analysis purposes in central banks and by IMF country desks. A number of inflation forecast targeting (IFT) central banks have used similar models as integral parts of their forecasting and policy analysis system (see Laxton and others, 2009).

4. To capture the commodity dependence and importance of the banking system in Indonesia, the baseline model is extended to incorporate oil prices and macrofinancial linkages through a credit conditions variable (Carabenciov and others, 2008). The model features a small open economy including forward-looking aggregate supply and demand with microfoundations and with stylized (realistic) lags in the different monetary transmission channels. External shocks from the rest of the world are captured here by U.S. growth. Output developments in the rest of the world feed directly into the small economy as they influence foreign demand for Indonesian products. Changes in foreign inflation and interest rates affect the exchange rate and, subsequently, demand and inflation in Indonesia.

5. The model is estimated using Bayesian techniques based on prior distributions for the parameters from cross-country work and assumptions about the Indonesian economy. Bayesian estimation in a situation of a relatively small sample size helps ameliorate the problems of classical econometric estimation, which often gives macro model results that are inconsistent and faced with simultaneity challenges. This is a particularly important aspect for Indonesia where there was a structural change in 2005 when the Bank Indonesia (BI) switched to an IFT framework, necessitating a subsample estimation from 2005 to 2012 to confirm the robustness of the full sample estimates. The model is estimated based on quarterly data from 2000 to 2012 using prior empirical knowledge about the parameters of interest for Indonesia or cross-country studies on emerging markets (see Anand, Ding, and Peiris, 2011; and Bathaluddin and Waluyo 2010). All variables are seasonally adjusted using the X12 filter, with the exception of the interest rate and the exchange rate, and expressed in “gap” terms, defined as deviations from a Hodrick-Prescott time trend or a multivariate filter (see Benes and others, 2010) in the case of the output gap.

6. The parameter estimates shed new insight into the monetary transmission mechanism in Indonesia, the role of domestic and global shocks, and the weights placed on inflation, the output gap, and the exchange rate in an open-economy Taylor-rule. The model has four behavioral equations: (1) an aggregate demand or IS curve that relates the level of real activity to expected and past real activity, the real interest rate, the real exchange rate, foreign demand, and financial conditions; (2) a price setting or Phillips curve that relates inflation to past and expected inflation, the output gap, and the exchange rate; (3) an uncovered interest parity condition for the exchange rate, with some allowance for backward looking expectations; and (4) an open-economy Taylor-rule for setting the policy interest rate as a function of the output gap, expected inflation, and the exchange rate.

(1) The aggregate demand equation and results are as follows:

${ygap}_t={\beta }_{ld}{ygap}_{t+1}+{\beta }_{lag}{ygap}_{t-1}-{\beta }_{RRgap}{RRgap}_{t-1}+{\beta }_{zgap}{zgap}_{t-1}+{\beta }_{RWygap}{ygap}_t^{RW}+{\beta }_{BL}{\eta }_t+{\epsilon }_t^{ygap}$

where ygap is the output gap, RRgap the real interest rate gap, zgap the real exchange rate gap, ygap^RW the output gap in the United States, η a measure of lending conditions based on the spread between market and policy rates,² and the, β a series of parameters attached to these variables, and ε^ygap an error term that captures other temporary exogenous demand shocks (details of the extension to the model to include lending conditions is in Appendix 1).

• Berg, Karam, and Laxton (2006) suggest that the value of β_lag will lie between 0.5 and 0.9, with a lower value for less advanced economies more susceptible to volatility. The coefficient of 0.47 for β_lag is comparable to other emerging markets. The lead of the output gap (β_ld) is typically small, between 0.05 and 0.2, and the estimated value for Indonesia is at the high end of that range. The parameter β_RRgap indicates the effectiveness of the monetary transmission mechanism, while β_zgap and β_RWygap depend on the importance of the exchange rate channel and the degree of openness. The posterior estimates of β_RRgap and β_zgap suggest that the interest rate effect on aggregate demand is stronger than the exchange rate affect, possibly reflecting the importance of factors beyond the exchange rate for competitiveness (e.g., costs of doing business). However, lending conditions have a slightly (0.15) stronger influence on aggregate demand than policy do rates. The value for β_RWygap is 0.15, in line with BI ARIMBI model of Bathaluddin and Waluyo 2010.

Parameter	Prior	Posterior
βld	0.6	0.479
βlag	0.2	0.183
βRRgap	0.15	0.129
βzgap	0.05	0.038
βRWygap	0.15	0.123
βBL	0.15	0.145

View Table

Parameter	Prior	Posterior
βld	0.6	0.479
βlag	0.2	0.183
βRRgap	0.15	0.129
βzgap	0.05	0.038
βRWygap	0.15	0.123
βBL	0.15	0.145

(2) The Philips curve equation and results are as follows:

${\pi }_t={\alpha }_{\pi ld}{\pi 4}_{t+1}+(1-{\alpha }_{\pi ld}){\pi 4}_{t-1}+{\alpha }_{ygap}{ygap}_{t-1}+{\alpha }_{zgap\left(z_{t-}{z}_{t-1}\right)+{\epsilon }_t^{\pi }}$

where ${\pi 4}_{t+1}$ is the four-quarter ahead inflation rate (year/year), ${\pi 4}_{t-1}$

View Expanded

View Expanded

• The α_πld parameter in the headline inflation equation determines the forward-looking component of inflation (while its inverse 1 −α_πld determines the backward-looking component). The parameter value can be interpreted as depending in part on the credibility of the central bank and in part on institutional arrangements regarding wage indexation and other price-setting mechanisms. A high value of α_πld, close to 1, would suggest that small changes in monetary policy cause large changes in price expectations. The α_ygap parameter depends on the extent to which output responds to price changes and, conversely, how much inflation is influenced by real demand pressures, and is typically between 0.25–0.50. This parameter ultimately depends on the “sacrifice ratio” (the loss of output necessary to bring down inflation) and is estimated to be 0.25. The α_zgap parameter represents the short-term passthrough of (real) exchange rate movements into prices, and depends on trade openness, price competition, and monetary policy credibility. The exchange rate passthrough coefficient is estimated to be relatively high in the Indonesia.

Parameter	Prior	Posterior
απld	0.25	0.288
αygap	0.25	0.245
αzgap	0.15	0.171

View Table

Parameter	Prior	Posterior
απld	0.25	0.288
αygap	0.25	0.245
αzgap	0.15	0.171

As an extension, the following equation for core inflation is added:

${\pi }_{c,t}={\alpha }_{c,\pi ld}{\pi 4}_{c,t+1}+(1-{\alpha }_{c,\pi ld}){\pi 4}_{c,t-1}+{\alpha }_{c,ygap}{ygap}_{t-1}+{\alpha }_{c,zgap}\left(z_{t-}{z}_{t-1}\right)+{\alpha }_c({\pi 4}_{t-1}-{\pi 4}_{c,t-1})+{\epsilon }_{c,t}^{\pi }$

• where the term $({\pi 4}_{t-1}-{\pi 4}_{c,t-1})$ has been added to the simple canonical inflation equation to allow for the possibility of relative price and real wage resistance (second-round effect); or more precisely that workers and other price setters may try to partially keep their prices from rising in pace with past movements in headline CPI. If the parameter αc, is zero, commodity price shocks that raise headline inflation, for example, will have no effect on core inflation and may not necessitate an increase in interest rates. However, to the extent that higher commodity prices are an important input into the production costs of many consumer goods, or if workers resist the reduction in their real wages in response to an increase in headline inflation, there could be a role for headline inflation to play in monetary policymaking. The estimated coefficient of αc of 0.49 is significant and should be taken into account, although there are a number of reasons to target core inflation in an economy prone to commodity price swings like Indonesia (see IMF 2011).

Parameter	Prior	Posterior
αc	0.4	0.494

View Table

Parameter	Prior	Posterior
αc	0.4	0.494

(3) The uncovered interest parity equation and results are as follows:

$z_t={\delta }_z{z}_{t+1}+(1-{\delta }_z)z_{t-1}-[{RR}_t-{RR}_t^{RW}-{\rho }^{*}]/4+{\epsilon }_t^z$

where z_t is the real exchange rate (an increase represents a depreciation), RR_t the real interest rate, ${RR}_t^{RW}$ the U.S. real interest rate, ρ^* the historical average risk premium on the domestic currency, δ_z the smoothness parameter, and ${\varepsilon }_t^z$ an error term. This equation, an uncovered interest rate parity condition, posits that the real exchange rate is a function of the expected real exchange rate (the first two terms), the real interest rate differential (the currency risk premium), and a disturbance term.

• The δ parameter in the real exchange rate equation determines the relative importance of forward- and backward-looking real exchange rate expectations. If δ is equal to 1, the exchange rate behaves as in the Dornbusch overshooting model (the real exchange rate is a function of the future sum of all real interest rate differentials). The estimated coefficient of 0.51 makes monetary policy potentially a more effective tool.

Parameter	Prior	Posterior
δz	0.6	0.509

View Table

Parameter	Prior	Posterior
δz	0.6	0.509

(4) The open-economy Taylor-rule and results are as follows:

${RS}_t={\gamma }_{RSlag}{RS}_{t-1}+(1-{\gamma }_{RSlag})*({RR}_t^{*}+\pi 4_t+{\gamma }_{\pi }[{\pi 4}_{t+4}-{\pi }_{t+4}^{*}]+{\gamma }_{ygap}{ygap}_t+{\gamma }_{zgap}{zgap}_t)+{\epsilon }_t^{RS}$

• The γ parameters in the monetary policy rule equation depend on the speed and extent to which the monetary authorities adjust the nominal interest rate, and the relative importance of the inflation target versus the real activity target. There is a significant degree of interest rate smoothing in Indonesia but BI does aggressively respond to inflation forecasts (expectations) above the targeted level. It is common for central banks to pay some attention to real activity even in a “pure” inflation targeting framework and, thus, for the γ_ygap coefficient to be greater than zero. This is borne out in the Indonesia data where the weight on real activity is comparable to other emerging markets. γ_zgap reflects the weight on the real exchange rate, which has been observed to be quite significant in emerging markets (see Stone and others, 2009). The estimated coefficient is low and indicates that BI does not appear to place weight on exchange rate developments in conducting monetary policy, although it may reflect the use of foreign exchange intervention to manage the exchange rate instead.

Parameter	Prior	Posterior
γRSlag	0.5	0.494
γπ	1.5	1.509
γygap	0.2	0.195
γzgap	0.05	0.011

View Table

Parameter	Prior	Posterior
γRSlag	0.5	0.494
γπ	1.5	1.509
γygap	0.2	0.195
γzgap	0.05	0.011

7. The model-based forecast suggests that in order for inflation to remain within BI’s target range, lending conditions would need to be tightened over the next two years. The analysis forecasts inflation and other real economy factors (such as the output gap) conditional on the WEO forecast for the U.S. economy, global inflation, and estimated distributions for stochastic shocks including supply-side factors in Indonesia. The simulations based on no tightening of lending conditions would result in core inflation exceeding the authorities’ inflation target, while gradually moving market rates to the policy rate (i.e., closing the spread between them) would help keep core inflation within the target range with broadly unchanged policy rates. The exact magnitudes of the estimated impact have significant uncertainties around them, as they depend on exact model specification, but the direction of change is likely to be instructive.

Core Inflation Forecast (No Tightening)

(Quarter-on-quarter s.a.a.r.)

Citation: IMF Staff Country Reports 2012, 278; 10.5089/9781475510775.002.A001

• Download Figure
• Download figure as PowerPoint slide

View Full Size

Core Inflation Forecast (No Tightening)

(Quarter-on-quarter s.a.a.r.)

Citation: IMF Staff Country Reports 2012, 278; 10.5089/9781475510775.002.A001

• Download Figure
• Download figure as PowerPoint slide

Core Inflation Forecast (No Tightening)

(Quarter-on-quarter s.a.a.r.)

Citation: IMF Staff Country Reports 2012, 278; 10.5089/9781475510775.002.A001

• Download Figure
• Download figure as PowerPoint slide

Core Inflation Forecast (With Tightening LC)

Quarter-on-quarter s.a.a.r.)

Citation: IMF Staff Country Reports 2012, 278; 10.5089/9781475510775.002.A001

• Download Figure
• Download figure as PowerPoint slide

View Full Size

Core Inflation Forecast (With Tightening LC)

Quarter-on-quarter s.a.a.r.)

Citation: IMF Staff Country Reports 2012, 278; 10.5089/9781475510775.002.A001

• Download Figure
• Download figure as PowerPoint slide

Core Inflation Forecast (With Tightening LC)

Quarter-on-quarter s.a.a.r.)

Citation: IMF Staff Country Reports 2012, 278; 10.5089/9781475510775.002.A001

• Download Figure
• Download figure as PowerPoint slide

B. Conclusion

8. The model estimates shed other important insights for monetary policy making. The Bayesian estimates show that the impact of lending conditions on the output gap is 0.15 compared to a coefficient of 0.13 for the policy rate. Thus, evaluating the monetary stance and inflation forecast without taking into account lending conditions could potentially lead to overshooting the inflation target if lending conditions are accommodative. Importantly, however, the model forecasts suggest that inflation could be kept within target by tightening lending conditions (raising market rates to the policy rate) without hiking policy rates. In addition, there are significant second-round effects, with a 1 percentage point increase in the difference in headline from core inflation leading to a 0.45 percentage point increase in core inflation. Therefore, changes to administered fuel prices that have not kept up with global prices can be another source of inflation pressure that may require a policy response.