Abstract
Indonesia’s external borrowing spreads increased by more than 1000 bps from mid-2007 to late 2008, before subsiding in recent months. The large increase in spreads prompted questions about whether the spreads adequately reflect the improvements in fundamentals made over the past few years. This Selected Issues paper examines the determinants of Indonesia’s spreads, and finds that fundamentals can explain both the level of and the increase in spreads. It uses a cross-country panel regression model of emerging market sovereign spreads to yield valuable insights into the pattern of Indonesia’s sovereign spreads.
III. External Shocks and Macroeconomic Performance in Indonesia1
Indonesia’s macroeconomic performance prior to the crisis was strong, in part owing to benign external conditions. As these conditions have deteriorated over the past year, however, Indonesia’s performance too has deteriorated. This paper uses vector autoregressions to quantitatively relate the external factors to domestic macroeconomic performance. It finds that global risk aversion, the terms of trade, and global growth matter importantly for Indonesia’s growth; forecasts for these external variables point to average growth of 3–4 percent in 2009. It also finds that monetary transmission in Indonesia takes 2–4 quarters.
A. Introduction
1. External factors appear to be important determinants of Indonesia’s macroeconomic performance. Against the backdrop of generally favorable global financial and economic conditions in 2004-07, Indonesia’s sovereign spreads declined, growth averaged 5½ percent annually, the current account surplus increased, and the real effective exchange rate (REER) appreciated. As global conditions deteriorated in late 2008, however, Indonesia’s performance worsened as well—trade collapsed, the balance of payments came under some stress, and growth slowed.
2. This paper quantitatively assesses the impact of external factors on Indonesia’s performance. Specifically, it examines the impact of growth in Indonesia’s partner countries (or “global” growth), global risk aversion (as measured by the VIX, which is the implied volatility of S&P 500 index options), and terms of trade on Indonesia’s real GDP growth. It also examines their relationship—and that of Indonesia’s growth—with domestic inflation, interest rates, sovereign spreads, and the REER.
3. The assessment is undertaken using an unrestricted vector autoregression (VAR). This approach not only allows for a decomposition of the factors accounting for each of the macroeconomic variables, but also facilitates making short-term forecasts. It complements modeling work done during the 2008 Article IV consultation (Reichold, 2008) as well as work underway at Bank Indonesia and the IMF (Andrle and others, 2009). The next section presents simple graphical relationships between variables, followed by the VAR analysis and forecasts. A final section concludes.
B. External Factors and Domestic Macroeconomic Variables: Graphical Analysis
4. Growth in Indonesia has tended to track growth in its partner countries. Prepared by Rishi Goyal. Quarterly seasonally-adjusted data is used starting in 2000, so as to exclude the 1997–98 crisis period as well as the immediate post- crisis recovery. A graphical analysis suggests a fairly close relationship between the two variables: as partner country growth has slowed or accelerated, growth in Indonesia has also tended to slow or pick up. There are no discernible breaks in the data over this sample period.
5. Similarly, Indonesia’s sovereign spreads have tended to track the VIX. As discussed in Goyal and Ruiz-Arranz (2009), Indonesia’s EMBI spreads have co-moved closely with global financial variables such as the VIX. Given that data on Indonesia’s spreads are available from mid 2004 onwards and given the high level of accuracy of the spread calculated by Hartelius (2006) and Hartelius and others (2008), the actual EMBI is used from mid 2004 onwards whereas the calculated spread is used from 2000 through mid 2004.
Indonesia and Partner Country Growth, 2000Q1–2009Q1
(In percent)
Citation: IMF Staff Country Reports 2009, 231; 10.5089/9781451818451.002.A003
Indonesia and Partner Country Growth, 2000Q1–2009Q1
(In percent)
Citation: IMF Staff Country Reports 2009, 231; 10.5089/9781451818451.002.A003
Indonesia and Partner Country Growth, 2000Q1–2009Q1
(In percent)
Citation: IMF Staff Country Reports 2009, 231; 10.5089/9781451818451.002.A003
VIX and Indonesia’s EMBI Spread, 2000Q1–2009Q1
(In basis points)
Citation: IMF Staff Country Reports 2009, 231; 10.5089/9781451818451.002.A003
VIX and Indonesia’s EMBI Spread, 2000Q1–2009Q1
(In basis points)
Citation: IMF Staff Country Reports 2009, 231; 10.5089/9781451818451.002.A003
VIX and Indonesia’s EMBI Spread, 2000Q1–2009Q1
(In basis points)
Citation: IMF Staff Country Reports 2009, 231; 10.5089/9781451818451.002.A003
6. The real exchange rate has co-moved with the terms of trade, though the relationship has not always been strong. As Indonesia’s export commodity prices have risen (fallen) and its terms of trade have strengthened (weakened), particularly since 2002-03, the REER has tended to strengthen (weaken). Of the domestic nominal variables, the policy interest rate appears to have responded to quarter-on-quarter, annualized inflation.
Indonesia: Terms of Trade and Real Effective Exchange Rate, 2000Q1–2009Q1
Index, 2002/03 = 100)
Citation: IMF Staff Country Reports 2009, 231; 10.5089/9781451818451.002.A003
Indonesia: Terms of Trade and Real Effective Exchange Rate, 2000Q1–2009Q1
Index, 2002/03 = 100)
Citation: IMF Staff Country Reports 2009, 231; 10.5089/9781451818451.002.A003
Indonesia: Terms of Trade and Real Effective Exchange Rate, 2000Q1–2009Q1
Index, 2002/03 = 100)
Citation: IMF Staff Country Reports 2009, 231; 10.5089/9781451818451.002.A003
Indonesia: CPI Inflation and Policy Interest Rate, 2000Q1–2009Q1
(In percent)
Citation: IMF Staff Country Reports 2009, 231; 10.5089/9781451818451.002.A003
Indonesia: CPI Inflation and Policy Interest Rate, 2000Q1–2009Q1
(In percent)
Citation: IMF Staff Country Reports 2009, 231; 10.5089/9781451818451.002.A003
Indonesia: CPI Inflation and Policy Interest Rate, 2000Q1–2009Q1
(In percent)
Citation: IMF Staff Country Reports 2009, 231; 10.5089/9781451818451.002.A003
C. Vector Autoregression Analysis
7. Putting all the variables together in a single VAR can yield insights into the overall macroeconomic relationships. A VAR is estimated with 2 lags, using global growth, VIX, terms of trade, domestic growth, inflation, interest rates, EMBI spreads, and the REER. In other words, the external variables—global growth, VIX, and terms of trade—are ordered first, given that they are exogenous to Indonesia-specific variables. These global variables are followed by a parsimonious structure for the Indonesian macroeconomy, namely, growth, inflation, interest rates, spreads, and the REER.
8. As a first step, the plausibility of the model estimates is established on the basis of impulse response functions. The model produces impulse response functions in line with expectations: all else equal, an increase in global growth translates into an increase in Indonesian growth, a decline in sovereign spreads, and an appreciation of the REER. An increase in global risk aversion, on the other hand, leads to a deterioration in sovereign spreads and higher domestic interest rates and depresses Indonesian growth. An exogenous deterioration in sovereign spreads (EMBI) also results in a weaker REER and higher domestic interest rates as well as lower growth. Finally, an improvement in the terms of trade positively impacts the REER, resulting in downward pressures on prices and interest rates; subsequently, however, the REER depreciates with upward price responses and interest rate action, though the impact is somewhat insignificant quantitatively.2
9. Having established the plausibility of the model, a variance decomposition exercise permits an analysis of the key factors that account for the variation in the main macroeconomic variables. The real variables are considered first, followed by the nominal variables:
Growth: While the strength of domestic demand in Indonesia has helped support growth more than in many other emerging markets, external factors can still have a significant impact on performance. Over a 2-year horizon, the decomposition shows that external factors account for nearly ½ of the variation in Indonesia’s growth. Global risk aversion (VIX) accounts for nearly ⅓ of the variation in growth. Global growth helps account for over 10 percent of the variation on impact, but this contribution declines over a 1 year horizon. Terms of trade account for less than 15 percent of the variation.
REER: External factors account for over 60 percent of the variation in Indonesia’s real effective exchange rate. The terms of trade and global growth, in particular, are important. Among the domestic variables, inflation plays an important role in accounting for the REER.
Interpretation. These decompositions illustrate that external conditions, particularly related to risk aversion, can impact Indonesian growth. As long as global financial conditions remain strained, the impact of a pickup in regional or global growth on Indonesia’s growth may be constrained. Furthermore, improved terms of trade should boost growth, unless accompanied by increased financial volatility; so, if a depreciation in the U.S. dollar boosts commodity prices and the terms of trade but is accompanied by greater uncertainty and financial market turmoil, then the growth impact of the higher terms of trade may be offset. It should also be noted, however, that domestic variables, such as lagged growth, continue to significantly affect growth. Rising domestic demand, therefore, owing for instance to countercyclical fiscal policy, can help to buffer growth against adverse external shocks.
10. The behavior of inflation and interest rates can also be assessed, thereby permitting an analysis of Indonesia’s monetary transmission mechanism.
Impulse response. An increase in interest rates, all else equal, results in lower growth in 3–4 quarters, and an appreciation in the REER in about 1 quarter. The impact on inflation is noted in 2 quarters.3
Variance decomposition. The variance decomposition suggests that domestic factors account for over 60 percent of the variation in inflation and the bulk of the variation in interest rates. Lagged inflation remains the key explanatory variable for both inflation and interest rates. Lagged interest rates also matter importantly for accounting for interest rate variation. This close relationship of inflation and interest rates suggests that stabilization of inflation can occur even against the backdrop of a strained external environment.
Variance Decomposition of Indonesian Real GDP
In percent
Citation: IMF Staff Country Reports 2009, 231; 10.5089/9781451818451.002.A003
Variance Decomposition of Indonesian Real GDP
In percent
Citation: IMF Staff Country Reports 2009, 231; 10.5089/9781451818451.002.A003
Variance Decomposition of Indonesian Real GDP
In percent
Citation: IMF Staff Country Reports 2009, 231; 10.5089/9781451818451.002.A003
Variance Decomposition of REER
In percent
Citation: IMF Staff Country Reports 2009, 231; 10.5089/9781451818451.002.A003
Variance Decomposition of REER
In percent
Citation: IMF Staff Country Reports 2009, 231; 10.5089/9781451818451.002.A003
Variance Decomposition of REER
In percent
Citation: IMF Staff Country Reports 2009, 231; 10.5089/9781451818451.002.A003
Variance Decomposition of Inflation
In percent
Citation: IMF Staff Country Reports 2009, 231; 10.5089/9781451818451.002.A003
Variance Decomposition of Inflation
In percent
Citation: IMF Staff Country Reports 2009, 231; 10.5089/9781451818451.002.A003
Variance Decomposition of Inflation
In percent
Citation: IMF Staff Country Reports 2009, 231; 10.5089/9781451818451.002.A003
Variance Decomposition of Interest Rates
In percent
Citation: IMF Staff Country Reports 2009, 231; 10.5089/9781451818451.002.A003
Variance Decomposition of Interest Rates
In percent
Citation: IMF Staff Country Reports 2009, 231; 10.5089/9781451818451.002.A003
Variance Decomposition of Interest Rates
In percent
Citation: IMF Staff Country Reports 2009, 231; 10.5089/9781451818451.002.A003
D. Forecasts
11. Near-term conditional forecasts can be made using the model. Given the path for partner country (or global) growth and commodity prices (hence, terms of trade) from the World Economic Outlook, and assuming a broadly constant path for global risk aversion (VIX), a baseline projection can be made for 2009 Q2-Q4. These projections may be modified on the basis of some upside risks and downside risks to ascertain upside and downside forecasts.
12. Average annual growth of 3–4 percent can be expected for 2009. Should growth in partner countries be somewhat higher than expected, with improved terms of trade and lower global risk aversion, an outcome in the upper end of the range can be expected. Similarly, should the external factors be worse than expected—with lower partner country growth, weaker terms of trade, and higher global risk aversion—a growth outcome in the lower end of the range would be expected.
13. On the basis of the monetary stimulus in the pipeline, the model suggests the potential for some inflation pick up in the second half of the year. With lags of 2–4 quarters in the monetary transmission mechanism, the easing cycle that commenced in December 2008 is expected to impact growth and inflation in the second half of 2009. This suggests a cautionary stance of monetary policy going forward, with, at a minimum, a pause in the easing cycle to assess how conditions evolve.
E. Conclusions
14. The empirical analysis presented in this paper suggests that external factors can matter importantly for Indonesia’s macroeconomic performance. Global risk aversion, in particular, is found to have a significant quantitative impact on growth. The terms of trade also are found to affect growth. Growth in partner countries and the terms of trade are found to affect the REER systematically. The analysis points to caution in interpreting the impact of higher terms of trade, if accompanied by higher risk aversion. To the extent that U.S. dollar depreciation would lead to improved terms of trade but, at the same time, is accompanied by greater uncertainty in financial markets and ultimately higher risk aversion, then the growth impact of higher terms of trade could be offset.
15. Monetary transmission is found to take 2–4 quarters. Therefore, the impact of the monetary easing cycle that commenced in December 2008 would be expected to impact macroeconomic variables during the second half of 2009. These are the normal lags in the transmission process. Inflation and interest rates are also found to largely be determined by themselves (or their lagged variables), suggesting that stabilization of inflation could occur even against the backdrop of unsettled external conditions.
16. Model forecasts for growth in 2009 are in the 3–4 percent range. A modestly better external environment would be expected to result in a growth outcome in the upper end of the range, while a weaker-than-expected environment would result in an outcome in the lower end of the range. The model also forecasts that inflation could pick up somewhat in the second half of the year, suggesting the need for a cautionary stance of monetary policy going forward.
References
Andrle, Michal, Charles Freedman, Roberto Garcia-Saltos, Danny Hermawan, Douglas Laxton, and Haris Munandar, 2009, “Adding Indonesia to the Global Projection Model,” mimeo. (March) (Washington: International Monetary Fund).
Goyal, Rishi, and Marta Ruiz-Arranz, 2009, “Explaining Indonesia’s Sovereign Spreads,” Indonesia: Selected Issues, forthcoming (Washington: International Monetary Fund).
Hartelius, Kristian, 2006, “Main Drivers of Emerging Market Bond Spreads: Fundamentals or External Factors?,” Global Financial Stability Report, April, pp. 28– 31 (Washington: International Monetary Fund).
Hartelius, Kristian, Kenichiro Kashiwase, Laura E. Kodres, 2008, “Emerging Market Spread Compression: Is it Real or is it Liquidity?,” IMF Working Paper 08/10 (Washington: International Monetary Fund).
Reichold, Steffen, 2008, “Inflation Outlook and Monetary Policy Challenges: A Model-Based Analysis,” in Indonesia: Selected Issues, IMF Country Report No. 08/298 (Washington: International Monetary Fund).
Prepared by Rishi Goyal.
The low significance could partly be due to measurement error in computing the terms of trade. In the absence of firm estimates from the authorities based on detailed trade data, staff estimates of the terms of trade are used.
As in VARs estimated for other countries, such as the United States, the VAR estimated here demonstrates an “inflation puzzle,” which refers to a seemingly counter-intuitive increase in inflation in the impulse response of an interest rate increase. This can be explained by the pattern in the data whereby interest rates react to inflation increases, and so rates rise when inflation rises.
