Abstract
This Selected Issues paper takes stock of Indonesia’s performance against the original macroeconomic objectives under the IMF’s extended arrangement. The paper compares the performance of the Indonesian economy in the post-crisis period with that of the other major “crisis” countries in the region. It reviews the background to the current extended arrangement and describes the core macroeconomic objectives of the program. The paper also considers Indonesia’s performance against objectives for growth, inflation, the balance of payments, and improving Indonesia’s debt sustainability.
II. Inflation Developments and the Monetary Framework1
A. Introduction
1. After falling to a record low in early 2000, inflation in Indonesia has increased significantly to a range of 12–15 percent, exceeding the targets under the government-s economic program. This increase, which has contrasted markedly with the experience elsewhere in the region, has come amid a sustained overshooting of the central bank’s targets for base money growth, and persistent weakness in the exchange rate. This has led to questions about the effectiveness of the current monetary framework, which emphasizes operating targets for base money growth as one of the factors used in guiding the stance of monetary policy. In particular, questions have arisen as to whether the demand for base money is sufficiently stable to be used as a guide for monetary policy.
2. This chapter seeks to address these questions by examining the evidence related to the stability of money demand. The chapter is organized as follows. Section B briefly reviews past inflation developments, with particular emphasis on the most recent experience. Section C examines trends in monetary aggregates, highlighting developments in the postcrisis period. Section D considers the extent to which a standard money demand framework can account for the actual developments in the monetary aggregates. Section E provides a summary and conclusion.
B. Inflation Developments
3. From 1985 until 1997, just prior to the economic crisis, Indonesia’s headline inflation rate (as measured by the overall CPI) generally fluctuated in the 5–10 percent range. While this was somewhat higher than in other economies in the region, it was down from an average of 14 percent in the preceding decade, when high inflation was associated with periodic large depreciations of the rupiah (Figure 1).
Indonesia: Inflation 1975-2002
(year-on-year)
Citation: IMF Staff Country Reports 2002, 154; 10.5089/9781451818239.002.A002
Indonesia: Inflation 1975-2002
(year-on-year)
Citation: IMF Staff Country Reports 2002, 154; 10.5089/9781451818239.002.A002
Indonesia: Inflation 1975-2002
(year-on-year)
Citation: IMF Staff Country Reports 2002, 154; 10.5089/9781451818239.002.A002
4. Inflation increased sharply following the onset of the crisis in 1997–98. After the abandonment of the exchange rate peg, the rupiah depreciated by nearly 85 percent from June 1997 to June 1998. In addition, monetary control was lost as Bank Indonesia (BI), acting in its role as lender-of-last-resort, channeled massive liquidity support to banks, which was not sterilized. Supply shortages and problems with the distribution system in the aftermath of the crisis, particularly in the agricultural sector, also contributed to the increase in prices. Accordingly, inflation rose from 5 percent in June 1997 to a peak of over 80 percent in September 1998.
5. BI regained monetary control in mid-1998, adopting a program that was centered on tight control of base money growth. As a result, the rupiah recovered by about 40 percent in the last quarter of 1998, and monthly inflation slowed sharply, with the annual inflation rate falling back to single digits by the second half of 1999.
6. Headline inflation slowed further in early 2000, with prices actually falling in the first quarter. From the third quarter of 1999 until mid-2000, headline inflation was below 2 percent. To a large extent, this outcome reflected sharp declines in raw food prices, which comprise over 20 percent of the consumption basket (Figure 2). However, the increase in prices of other goods also slowed, reflecting the large output gap and appreciating currency.
Indonesia: Growth of CPI components
(year-on-year)
Citation: IMF Staff Country Reports 2002, 154; 10.5089/9781451818239.002.A002
Indonesia: Growth of CPI components
(year-on-year)
Citation: IMF Staff Country Reports 2002, 154; 10.5089/9781451818239.002.A002
Indonesia: Growth of CPI components
(year-on-year)
Citation: IMF Staff Country Reports 2002, 154; 10.5089/9781451818239.002.A002
7. Inflation has since accelerated again, fluctuating in the range of 12-15 percent since mid-2001. This increase has been broad-based, observed across all categories of goods and services. Large adjustments in the prices of administered goods, including fuel and electricity, have also contributed. Nevertheless, even abstracting from such temporary effects, staff estimates suggest that the core rate of inflation has also risen significantly, to about 12-13 percent. Box 1 discusses alternative measures of underlying inflation.
8. The recent increase in inflation coincides with a period of renewed weakness in the exchange rate (Figure 3). The rupiah depreciated through most of 2000 and 2001, in response to political upheaval and slow implementation of reforms. A sharp recovery associated with the change of government in mid-2001 proved short-lived, and the rupiah weakened again in the latter part of 2001, before a partial recovery in early 2002. As discussed further below, base money growth accelerated sharply in 2000 and most of 2001.
Indonesia: Inflation and exchange rate developments
(yoy change)
Citation: IMF Staff Country Reports 2002, 154; 10.5089/9781451818239.002.A002
Indonesia: Inflation and exchange rate developments
(yoy change)
Citation: IMF Staff Country Reports 2002, 154; 10.5089/9781451818239.002.A002
Indonesia: Inflation and exchange rate developments
(yoy change)
Citation: IMF Staff Country Reports 2002, 154; 10.5089/9781451818239.002.A002
9. Previous research finds evidence supporting the link between inflation outcomes, the exchange rate, and money growth in Indonesia. Ramakrishnan and Vamvakidis (2002) investigate the determinants of inflation for the period 1980 to 2000. They find that movements in the exchange rate, foreign inflation, and base money growth are significant variables in explaining inflation developments in Indonesia. Basri and others (2002) confirm these results, and finds that exchange rate movements are particularly important for explaining short-run inflation outcomes, while money growth helps explain inflation at horizons of a year and longer. McLeod (1997) examines the behavior of prices and monetary aggregates for the precrisis period 1989 to 1995. He finds that the consistent overshooting of the inflation target can be explained by excessive base money growth, resulting from unsterilized capital inflows. He maintains that base money demand is stable, and argues that BI should focus exclusively on base money to achieve its inflation goal.
Alternate Measures of Inflation
Overall inflation in Indonesia is measured by the CPI index, based on surveys of consumer prices during the reference period in each month. Some components such as raw food exhibit volatile month-to-month movements, while components with administered prices face large but infrequent adjustments. To disentangle such effects, a number of measures of inflation have been developed. The analysis in the current chapter is based on headline inflation.
Headline inflation This is the increase in prices of the overall CPI index. BI’s inflation target for 2002 refers to this definition.
Nonfood inflation Raw food accounts for 22 percent of the CPI basket. Raw food prices move sharply in response to agricultural conditions and shocks, and are not reliable indicators of inflation pressures. In the past, IMF staff have used this as a measure of underlying inflation.
Nonfood, nonadministered prices (“Core”) Many components of the CPI basket have prices which are set administratively. For example, about 8 percent of the CPI index consists of petroleum items and utilities (telephone and electricity). In the face of recent large adjustments in the prices of these items, IMF staff have constructed a measure of core inflation excluding this subset of administered prices, as well as raw food prices (Figure 1).
Excluding volatile prices (“Exclusion”) The logic for excluding food prices on the basis of volatility can be applied to all components of the CPI. BI constructs a measure of core inflation which excludes excessively volatile components, mostly food items and administered prices. The components of the “exclusion CPI” represent about 60 percent of the full CPI basket.
Excluding volatile prices (“Trim”) Instead of permanently dropping items from the measure of underlying inflation, the trim method drops items whose monthly movements exceed some preset level. Thus, the composition of the “trim CPI” changes over time. BI’s measure of “trim CPI” drops the 30 items with the highest monthly price change, and the 102 items with the lowest monthly price change. This approach typically excludes 20–30 percent of the CPI basket.
Indonesia: Headline and Core inflation
(year-on-year; core inflation defined as nonfood, nonadministrated prices)
Citation: IMF Staff Country Reports 2002, 154; 10.5089/9781451818239.002.A002
Indonesia: Headline and Core inflation
(year-on-year; core inflation defined as nonfood, nonadministrated prices)
Citation: IMF Staff Country Reports 2002, 154; 10.5089/9781451818239.002.A002
Indonesia: Headline and Core inflation
(year-on-year; core inflation defined as nonfood, nonadministrated prices)
Citation: IMF Staff Country Reports 2002, 154; 10.5089/9781451818239.002.A002
C. Monetary Developments
10. In the period leading up to the crisis, monetary policy sought to balance multiple objectives, including high growth, a sustainable balance of payments, and low inflation. In particular, policy-makers were faced with the competing objectives of seeking to resist a real appreciation in the face of strong capital inflows, while maintaining growth of monetary aggregates consistent with low inflation. Operationally, BI announced an inflation goal, and targeted growth rates for various monetary aggregates—base money, broad money, and bank lending. In practice, the crawling depreciation of the exchange rate served as the nominal anchor. In line with this objective, capital inflows were not sterilized, and the monetary growth targets were generally exceeded.
Indonesia: Growth rate of monetary aggregates
Citation: IMF Staff Country Reports 2002, 154; 10.5089/9781451818239.002.A002
Indonesia: Growth rate of monetary aggregates
Citation: IMF Staff Country Reports 2002, 154; 10.5089/9781451818239.002.A002
Indonesia: Growth rate of monetary aggregates
Citation: IMF Staff Country Reports 2002, 154; 10.5089/9781451818239.002.A002
11. There were notable differences in the behavior of the monetary aggregates in the precrisis period (Figure 4). Rupiah broad money grew at a relatively rapid but stable rate, while currency growth fluctuated sharply. The ratio of broad money (M2) to GDP increased gradually through 1997, in line with the expansion of the banking system (Figure 5). The number of banks more man doubled between 1989 and 1994 as entry rules were relaxed. Currency growth increased to near 30 percent by 1994, during a period of declining interest rates and accelerating real growth. As interest rates rose before the crisis, currency growth subsided. Reflecting the high share of currency in base money, developments in the ratio of base money to GDP have broadly followed those in currency, although with stronger end-year fluctuations.2
Indonesia: Monetary Aggregates
(radio to GDP)
Citation: IMF Staff Country Reports 2002, 154; 10.5089/9781451818239.002.A002
Indonesia: Monetary Aggregates
(radio to GDP)
Citation: IMF Staff Country Reports 2002, 154; 10.5089/9781451818239.002.A002
Indonesia: Monetary Aggregates
(radio to GDP)
Citation: IMF Staff Country Reports 2002, 154; 10.5089/9781451818239.002.A002
12. All monetary aggregates experienced rapid growth rates during the crisis. As BI was unable to sterilize the massive liquidity support that was being channeled to banks, base money increased sharply. The growth rate of monetary aggregates peaked in mid-1998, before gradually declining as the government’s stabilization program took hold. By end-1998, the ratios of the monetary aggregates to GDP had broadly returned to their precrisis levels.
13. There have been significant divergences in the behavior of the monetary aggregates in the postcrisis period. Broad money growth has been subdued, and the ratio of M2-to-GDP has declined gradually. This experience reflects the continued weaknesses in the banking sector, and the slow pace of corporate restructuring and legal reform, which have impeded a recovery in credit creation. In contrast, currency growth has increased sharply to around 20 percent since early 2000, before slowing somewhat in early 2002.
14. More generally, movements in the currency-to-deposit ratio reflect the impact of secular change. The currency-to-deposit ratio experienced a steep decline from a level of around 30 percent in the mid -1980s, to about 12 percent in recent years (Figure 6). The steepest decline occurred in the period through 1990, coinciding with the sharp increase in deposits as a result of liberalization of the financial system, particularly the lifting of some of the controls on deposit rates. After increasing during the crisis, the currency-to-deposit ratio returned to its precrisis level by end-1998. Since then, the ratio has risen steadily to its current level of about 12 percent. These recent trends in the currency-to-deposit ratio reflect the increase in currency growth noted above.
Indonesia: Currency-deposit ratio, 1986-2001
Citation: IMF Staff Country Reports 2002, 154; 10.5089/9781451818239.002.A002
Indonesia: Currency-deposit ratio, 1986-2001
Citation: IMF Staff Country Reports 2002, 154; 10.5089/9781451818239.002.A002
Indonesia: Currency-deposit ratio, 1986-2001
Citation: IMF Staff Country Reports 2002, 154; 10.5089/9781451818239.002.A002
15. The monetary policy regime has changed dramatically compared with the precrisis period. Most importantly, the authorities have adopted a floating exchange rate regime. To deal with the crisis and regain control over its balance sheet, BI brought the sources of base money growth under control, discontinuing some liquidity facilities, and increasing penalties for the use of its discount window and bank overdraft facilities. Net domestic assets were targeted to stay broadly unchanged, and a floor was set on net international reserves.
16. The base money framework was initially successful. However, in the second half of 2000 and most of 2001, base money and inflation repeatedly exceeded their targets, despite several revisions in these targets (Table 1). The over-run in base money has mainly reflected stronger-than-expected growth in currency in circulation, which comprises almost two-thirds of base money.3 These deviations in base money from the target path have raised questions about whether this reflects primarily incipient inflationary pressures or instability in the demand for base money. To better understand this experience, it is useful to assess in more depth the factors underlying the demand for money, and the extent to which actual monetary developments can be explained by a standard money demand framework.
Indonesia: Inflation Outturn Relative to Targets
Figures report amount by which the actual outturn at the end of the quarter exceeded the predicted value for that date. The residual column is calculated by imposing the quantity equation (negative residuals imply an underestimate of the change in velocity).
Indonesia: Inflation Outturn Relative to Targets
| Inflation | Base money | Growth | Residual | |
|---|---|---|---|---|
| Deviation from path expected in November 20001/ | ||||
| Mar-01 | 2.1 | 2.2 | 0.0 | −0.2 |
| Jun-01 | 4.4 | 7.9 | −2.8 | −6.3 |
| Sep-01 | 6.2 | 10.4 | −3.1 | −7.3 |
| Dec-01 | 7.4 | −2.0 | −3.6 | 5.8 |
| Deviation from path expected in April 20011/ | ||||
| Jun-01 | 1.4 | 5.1 | 1.6 | −2.0 |
| Sep-01 | 2.0 | 6.4 | −2.0 | −6.4 |
| Dec-01 | 4.7 | 5.0 | −2.0 | −2.3 |
| Deviation from path expected in July 20011/ | ||||
| Sep-01 | 0.8 | 4.9 | −1.3 | −5.4 |
| Dec-01 | 0.0 | 0.0 | −0.7 | −0.7 |
Figures report amount by which the actual outturn at the end of the quarter exceeded the predicted value for that date. The residual column is calculated by imposing the quantity equation (negative residuals imply an underestimate of the change in velocity).
Indonesia: Inflation Outturn Relative to Targets
| Inflation | Base money | Growth | Residual | |
|---|---|---|---|---|
| Deviation from path expected in November 20001/ | ||||
| Mar-01 | 2.1 | 2.2 | 0.0 | −0.2 |
| Jun-01 | 4.4 | 7.9 | −2.8 | −6.3 |
| Sep-01 | 6.2 | 10.4 | −3.1 | −7.3 |
| Dec-01 | 7.4 | −2.0 | −3.6 | 5.8 |
| Deviation from path expected in April 20011/ | ||||
| Jun-01 | 1.4 | 5.1 | 1.6 | −2.0 |
| Sep-01 | 2.0 | 6.4 | −2.0 | −6.4 |
| Dec-01 | 4.7 | 5.0 | −2.0 | −2.3 |
| Deviation from path expected in July 20011/ | ||||
| Sep-01 | 0.8 | 4.9 | −1.3 | −5.4 |
| Dec-01 | 0.0 | 0.0 | −0.7 | −0.7 |
Figures report amount by which the actual outturn at the end of the quarter exceeded the predicted value for that date. The residual column is calculated by imposing the quantity equation (negative residuals imply an underestimate of the change in velocity).
D. Money Demand in Indonesia
Approach
17. Following a standard framework, real money demand is specified as a function of real output, the opportunity cost of money, and other factors:
m = f(y;r;x),
where m is real money, y is real output, r represents opportunity cost variables, and x represents other variables which are believed to affect money demand.
18. This chapter considers the demand for a variety of monetary aggregates, with particular emphasis on currency demand. Monthly real broad and narrow money aggregates are calculated by dividing end-month values by the CPI level. Monthly real output is imputed from the official quarterly real output series. All specifications are estimated in logarithmic form.
19. The definition of the opportunity cost variable differs depending on the measure of money under consideration. For the broader aggregates, a real interest rate is constructed based on ex post inflation outcomes (both year-on-year and quarter-on-quarter). Following the standard approach, the nominal deposit rate is used when analyzing the demand for narrower monetary aggregates. This approach presumes that the opportunity cost of holding currency is the foregone interest on deposits. Another alternative to holding rupiah currency is holding foreign currency, which would earn the expected rate of depreciation. The actual ex post rate of depreciation is used to capture this phenomenon.
20. Money demand was estimated using monthly data from January 1990 through December 2001. A key consideration is the treatment of the crisis. The postcrisis period differs from the precrisis period in many important ways, including the exchange rate regime, the instruments of monetary policy, and the structure of the financial system. To capture possible changes in the income elasticity of money demand since the crisis, an interaction term is included, defined as a dummy variable times real income, with the dummy variable taking the value of one starting in July 1997 through the end of the sample period. In addition, the specifications are considered separately for the pre- and postcrisis periods. A general shortcoming of such exercises for Indonesia is that only a limited amount of data is available for the postcrisis period.
21. To account for serial correlation, a lagged-adjustment model is employed, with lagged real money included in the specification. Elasticities can be calculated from the estimated coefficients.4 In addition, the specifications include a seasonal dummy variable to control for the temporary increase in currency holdings during the end-year festivities.5
22. Economic theory suggests that the estimated income elasticity of money demand should be near one for currency, and possibly higher for the broader aggregates. Typically, as income rises and the financial and banking systems become more developed, demand for broad money should increase, while demand for currency would be expected to moderate. As financial intermediation deepens, the need to transact using currency would decline. In the Indonesian case, recent difficulties in the banking system could retard the growth of broad money and lead to a continued reliance on currency. The interest semi-elasticity should be less than zero for currency. Although the interest semi-elasticity for broad money could theoretically be positive, it is generally expected to be less than zero.
23. Studies for developing countries typically find elasticities consistent with theory. Sriram (2001) reports on recent studies for a large number of countries, and finds that the income elasticity of broad money demand in developing countries is often above one, while the income elasticity of currency demand is close to one in most of these countries. By contrast, Dekle and Pradhan (1997, 1999) find that, for the precrisis period, the income elasticity of broad money demand in Indonesia is near one, and the income elasticity of demand for Ml around 1.5.6 However, more recent research by BI based on quarterly data from 1984 to 2001 finds that the income elasticity of currency demand is around 1.1, and the income elasticity of broad money demand is near 2 (BI Research Department, 2001). The results in both studies are highly sensitive to the time period considered, and the definition of the opportunity cost variable. Sriram (2001) also reports that the interest semi-elasticity is negative for most countries and monetary aggregates.
Estimation Results
24. For broad money (M2 and rupiah M2), the results are not fully consistent with theoretical priors or the findings of other studies (Table 2). Contrary to the findings of other studies, the estimated interest semi-elasticity is positive and statistically significant. In addition, the estimated coefficients vary across specifications and time periods. Alternate definitions of the opportunity cost variable and the inclusion of regressors that seek to account for extreme movements in the immediate postcrisis period do not produce robust results. The estimated income elasticities are high (near two), but not inconsistent with results from other developing countries which have undergone significant financial-sector development. The weak results for the broad money specifications could reflect the breakdowns of the monetary transmission mechanism in the postcrisis period.7
Money Demand Specifications for the Period 1990–2001
For ease of exposition, the table presents actual estimated coefficients multiplied by 1000.
Money Demand Specifications for the Period 1990–2001
| M2 | Rupiah M2 | M1 | Base Money | Currency | |
|---|---|---|---|---|---|
| Constant | −1.94 | −0.33 | −1.84 | −2.55 | −2.79 |
| (0.55) | (0.35) | (0.52) | (0.80) | (0.67) | |
| Interest rate (x 103)1/ | 0.68 | 0.29 | −1.54 | −1.27 | −1.82 |
| (0.12) | (0.09) | (0.03) | (0.06) | (0.04) | |
| Real income | 0.23 | 0.04 | 0.25 | 0.29 | 0.34 |
| (0.06) | (0.04) | (0.06) | (0.08) | (0.07) | |
| Lagged money | 0.87 | 0.98 | 0.78 | 0.79 | 0.70 |
| (0.04) | (0.02) | (0.05) | (0.05) | (0.05) | |
| Crisis*real income (x 103)1/ | 2.37 | −0.34 | 2.24 | 6.33 | 4.86 |
| (0.95) | (0.58) | (0.09) | (2.09) | (1.17) | |
| Memo items: | |||||
| Income elasticity (precrisis) | 1.77 | 2.25 | 1.14 | 1.38 | 1.13 |
| Income elasticity (postcrisis) | 1.79 | 2.23 | 1.15 | 1.41 | 1.15 |
| Interest semi-elasticity (x 103)1/ | 5.25 | 18.10 | −7.00 | −6.05 | −6.07 |
For ease of exposition, the table presents actual estimated coefficients multiplied by 1000.
Money Demand Specifications for the Period 1990–2001
| M2 | Rupiah M2 | M1 | Base Money | Currency | |
|---|---|---|---|---|---|
| Constant | −1.94 | −0.33 | −1.84 | −2.55 | −2.79 |
| (0.55) | (0.35) | (0.52) | (0.80) | (0.67) | |
| Interest rate (x 103)1/ | 0.68 | 0.29 | −1.54 | −1.27 | −1.82 |
| (0.12) | (0.09) | (0.03) | (0.06) | (0.04) | |
| Real income | 0.23 | 0.04 | 0.25 | 0.29 | 0.34 |
| (0.06) | (0.04) | (0.06) | (0.08) | (0.07) | |
| Lagged money | 0.87 | 0.98 | 0.78 | 0.79 | 0.70 |
| (0.04) | (0.02) | (0.05) | (0.05) | (0.05) | |
| Crisis*real income (x 103)1/ | 2.37 | −0.34 | 2.24 | 6.33 | 4.86 |
| (0.95) | (0.58) | (0.09) | (2.09) | (1.17) | |
| Memo items: | |||||
| Income elasticity (precrisis) | 1.77 | 2.25 | 1.14 | 1.38 | 1.13 |
| Income elasticity (postcrisis) | 1.79 | 2.23 | 1.15 | 1.41 | 1.15 |
| Interest semi-elasticity (x 103)1/ | 5.25 | 18.10 | −7.00 | −6.05 | −6.07 |
For ease of exposition, the table presents actual estimated coefficients multiplied by 1000.
25. The specifications for currency and other narrow monetary aggregates yield economically plausible results. Using data for the full 1990–2001 period, real currency demand increases with real output, as expected, with an income elasticity above one. Increases in deposit rates reduce currency demand, although the estimated semi-elasticities are relatively low.8 As the coefficient on the crisis variable shows, the income elasticity of currency demand appears to have experienced a small structural increase from the onset of the crisis. The specifications for base money and Ml yield similar results, reflecting the large component of currency in these aggregates. Given these results, the rest of this section focuses on currency demand in greater detail.
26. For the whole period, the estimated long-run income elasticity of currency demand is greater than one, and broadly stable. For the period 1990–2001 (including the interaction term to account for the crisis), the estimated long-run income elasticity is slightly above 1.1 (Table 2, last column). 9 The estimated income elasticity is relatively unaffected by the precise time period considered. The coefficient on the interaction term is positive but small. As a result, the estimated long-run income elasticity has barely risen from 1.13 in the precrisis period to 1.15 in the postcrisis period.10 This result suggests that underlying demand for currency has been remarkably stable, despite the crisis. The estimated long-run income elasticity is consistent with results from other countries and for research covering different time periods in Indonesia.
27. It is difficult to assess the factors responsible for an income elasticity significantly above one. The ongoing difficulties of the banking system may reduce the attractiveness of holding deposits and undertaking transactions through the financial system, thereby contributing to the continuing importance of currency. Alternatively, it is possible that as economic development reaches more far-flung and poorer parts of the country which lack banking infrastructure, the demand for currency increases, as a replacement for barter arrangements. Finally, currency demand may have increased in the face of increasing political and economic uncertainty. Such developments are not easily captured in the framework, and specifications including time trends to proxy for these ongoing changes produce inconclusive results.
Alternate Specifications
28. Splitting the sample into two periods seems to suggest a marked increase in the income elasticity of currency demand in the postcrisis period (Table 3). Furthermore, a Chow test confirms that the coefficients experience a structural break at the time of the crisis, suggesting that estimates for the whole period which do not account for the crisis may be mis-specified. The behavior of currency in the postcrisis period appears to be highly volatile, however, and the estimated income and interest elasticities vary widely depending on the precise time period considered (last three columns of Table 3). The estimated interest semielasticity varies across specifications and time periods. In all cases, higher deposit rates lower currency demand, although the effect is lower for the postcrisis period. The estimated income elasticity in the postcrisis period is above two, which is well above the precrisis value, and inconsistent with findings from other studies. Accordingly, no strong conclusions about the behavior of money demand can be drawn from specifications based only on the short and unstable postcrisis period.
Indonesia: Currency Demand Specifications Before and After the Crisis
Indonesia: Currency Demand Specifications Before and After the Crisis
| Full period 1990–2001 | Precrisis 1990–97.6 | Postcrisis 1998.1–2001 | Postcrisis 1998.7–2001 | Postcrisis 1999.7–2001 | |
|---|---|---|---|---|---|
| Constant | −2.06 | −2.28 | −21.8 | −22.3 | −28.3 |
| (0.68) | (0.76) | (3.32) | (3.94) | (3.65) | |
| Interest rate (x 103) | −0.93 | −3.52 | −3.84 | −4.00 | −12.6 |
| (0.41) | (1.38) | (0.56) | (0.5) | (3.21) | |
| Real income | 0.22 | 0.29 | 2.09 | 2.11 | 2.65 |
| (0.07) | (0.09) | (0.30) | (0.35) | (0.32) | |
| Lagged currency | 0.85 | 0.73 | 0.10 | 0.14 | −0.04 |
| (0.04) | (0.07) | (0.11) | (0.11) | (0.10) | |
| Memo items: | |||||
| Number of observations | 143 | 89 | 48 | 42 | 30 |
| Income elasticity | 1.50 | 1.08 | 2.32 | 2.44 | 2.56 |
| Interest semi-elasticity (x 103) | −6.22 | −12.84 | −4.27 | −4.57 | −12.1 |
| Chow F-test for pre/post crisis | 4.25 | … | … | … | … |
Indonesia: Currency Demand Specifications Before and After the Crisis
| Full period 1990–2001 | Precrisis 1990–97.6 | Postcrisis 1998.1–2001 | Postcrisis 1998.7–2001 | Postcrisis 1999.7–2001 | |
|---|---|---|---|---|---|
| Constant | −2.06 | −2.28 | −21.8 | −22.3 | −28.3 |
| (0.68) | (0.76) | (3.32) | (3.94) | (3.65) | |
| Interest rate (x 103) | −0.93 | −3.52 | −3.84 | −4.00 | −12.6 |
| (0.41) | (1.38) | (0.56) | (0.5) | (3.21) | |
| Real income | 0.22 | 0.29 | 2.09 | 2.11 | 2.65 |
| (0.07) | (0.09) | (0.30) | (0.35) | (0.32) | |
| Lagged currency | 0.85 | 0.73 | 0.10 | 0.14 | −0.04 |
| (0.04) | (0.07) | (0.11) | (0.11) | (0.10) | |
| Memo items: | |||||
| Number of observations | 143 | 89 | 48 | 42 | 30 |
| Income elasticity | 1.50 | 1.08 | 2.32 | 2.44 | 2.56 |
| Interest semi-elasticity (x 103) | −6.22 | −12.84 | −4.27 | −4.57 | −12.1 |
| Chow F-test for pre/post crisis | 4.25 | … | … | … | … |
29. Specifications including exchange rate depreciation as the opportunity cost of currency do not yield robust results. When exchange rate movements are used instead of the deposit rate, the results suggest that a higher rate of depreciation leads to lower demand for currency,11 although the coefficient is insignificant (Table 4). If both the deposit rate and the exchange rate depreciation are included, higher depreciation increases the demand for currency. These specifications do not yield robust results, as the estimated income and interest elasticities differ significantly across specifications. Similarly, for the broader monetary aggregates, specifications including exchange rate depreciation yield poor results.
Indonesia: Currency Demand Specifications with Exchange Rate Depreciation
Indonesia: Currency Demand Specifications with Exchange Rate Depreciation
| Full period 1991–2001 | Full period 1991–2001 | Precrisis 1991–97.6 | Postcrisis 1997.7–2001 | |
|---|---|---|---|---|
| Constant | −3.15 | −4.21 | −4.58 | 0.79 |
| (0.87) | (0.83) | (1.09) | (2.71) | |
| Interest rate (x 103) | … | −3.09 | −5.79 | −4.07 |
| … | (0.63) | (2.29) | (0.99) | |
| Exchange rate depreciation | −0.02 | 0.04 | 1.07 | 0.09 |
| (0.01) | (0.02) | (0.79) | (0.03) | |
| Real income | 0.35 | 0.49 | 0.53 | 0.15 |
| (0.09) | (0.09) | (0.12) | (0.21) | |
| Lagged currency | 0.75 | 0.61 | 0.59 | 0.52 |
| (0.06) | (0.06) | (0.08) | (0.10) | |
| Crisis*real income (x 103) | 2.80 | 5.72 | … | … |
| (1.11) | (1.18) | … | … | |
| Memo items: | ||||
| Number of observations | 132 | 132 | 78 | 54 |
| Income elasticity (precrisis) | 1.41 | 1.25 | 1.29 | 0.31 |
| Income elasticity (postcrisis) | 1.42 | 1.27 | 1.29 | 0.31 |
| Interest semi-elasticity (x 103) | … | −7.83 | −14.03 | −8.44 |
Indonesia: Currency Demand Specifications with Exchange Rate Depreciation
| Full period 1991–2001 | Full period 1991–2001 | Precrisis 1991–97.6 | Postcrisis 1997.7–2001 | |
|---|---|---|---|---|
| Constant | −3.15 | −4.21 | −4.58 | 0.79 |
| (0.87) | (0.83) | (1.09) | (2.71) | |
| Interest rate (x 103) | … | −3.09 | −5.79 | −4.07 |
| … | (0.63) | (2.29) | (0.99) | |
| Exchange rate depreciation | −0.02 | 0.04 | 1.07 | 0.09 |
| (0.01) | (0.02) | (0.79) | (0.03) | |
| Real income | 0.35 | 0.49 | 0.53 | 0.15 |
| (0.09) | (0.09) | (0.12) | (0.21) | |
| Lagged currency | 0.75 | 0.61 | 0.59 | 0.52 |
| (0.06) | (0.06) | (0.08) | (0.10) | |
| Crisis*real income (x 103) | 2.80 | 5.72 | … | … |
| (1.11) | (1.18) | … | … | |
| Memo items: | ||||
| Number of observations | 132 | 132 | 78 | 54 |
| Income elasticity (precrisis) | 1.41 | 1.25 | 1.29 | 0.31 |
| Income elasticity (postcrisis) | 1.42 | 1.27 | 1.29 | 0.31 |
| Interest semi-elasticity (x 103) | … | −7.83 | −14.03 | −8.44 |
Model Predictions
30. The preferred model tracks actual currency developments well. Despite the many shocks over the period, within-sample predicted values fit the actual outcome closely (Figure 7). Although currency is volatile, the simple specification captures the main movements. Historically, deviations of currency from the predicted path tend to be small and corrected relatively quickly. The model does predict a significantly lower level of currency for 2001 than actually occurred (Figure 8). However, since the end of the sample period in 2001, the level of currency has fallen.
Indonesia: Within-sample currency projections
Citation: IMF Staff Country Reports 2002, 154; 10.5089/9781451818239.002.A002
Indonesia: Within-sample currency projections
Citation: IMF Staff Country Reports 2002, 154; 10.5089/9781451818239.002.A002
Indonesia: Within-sample currency projections
Citation: IMF Staff Country Reports 2002, 154; 10.5089/9781451818239.002.A002
Indonesia: Within-sample currency projections
Citation: IMF Staff Country Reports 2002, 154; 10.5089/9781451818239.002.A002
Indonesia: Within-sample currency projections
Citation: IMF Staff Country Reports 2002, 154; 10.5089/9781451818239.002.A002
Indonesia: Within-sample currency projections
Citation: IMF Staff Country Reports 2002, 154; 10.5089/9781451818239.002.A002
31. The observed decline in currency growth in 2002 is consistent with the predictions from the model. Out-of-sample projections (based on estimating the preferred model through mid-2000) fit the actual experience through March 2002, except for end-year spikes and a large sustained deviation from March 2001 through the end of that year (Figure 9). It is too early to conclusively assess the reasons for this deviation. However, the fact that recent outcomes can be explained by the model suggest that the over-run in 2001 may have been the result of a temporary shock—possibly related to increased political uncertainty—and not an indication of a permanent increase in currency demand.
Indonesia: Out-of-sample currency projections
Citation: IMF Staff Country Reports 2002, 154; 10.5089/9781451818239.002.A002
Indonesia: Out-of-sample currency projections
Citation: IMF Staff Country Reports 2002, 154; 10.5089/9781451818239.002.A002
Indonesia: Out-of-sample currency projections
Citation: IMF Staff Country Reports 2002, 154; 10.5089/9781451818239.002.A002
E. Conclusions
32. The Indonesian economy has undergone significant changes in recent years. Prior to the crisis, the monetary aggregates experienced strong secular movements, partly related to the ongoing development of the financial system. Inflation also fluctuated significantly over this period. As a result of the crisis, the structure of the banking system and the monetary regime changed significantly. In recent years, inflation has risen steadily, while the monetary aggregates have followed divergent paths. Growth in the broad money aggregates has remained subdued, while narrower aggregates, in particular currency, have experienced rapid growth.
33. Despite the many shocks affecting the economy, the evidence presented in this chapter suggests that monetary developments can be at least partly explained using a standard framework. Although the results for broad money are weak, a standard framework explains currency demand well. There is evidence that the income elasticity of currency demand is greater than one, and has increased only modestly in the postcrisis period. With the exception of a period in 2001, the model explains actual currency developments well, both within and outside the sample period. The recent decline in currency—in line with the model’s predictions—suggests that the 2001 outcome may have reflected largely temporary factors.
34. These findings suggest that the base money framework remains a useful anchor for monetary policy in Indonesia. Although currency can experience significant short-term fluctuations, over longer horizons currency demand appears to be relatively stable, with an income elasticity slightly above one. These results appear to imply that part of the reason why inflation has overrun its target is because base money growth has exceeded its programmed path. The findings of this chapter thus highlight the importance of monitoring base money developments, as outturns consistently above the programmed path may provide indications of incipient inflation pressures.
35. Future work could consider the factors which affect base money outcomes at shorter horizons. In particular, it would be useful to determine the extent to which base money movements reflect temporary factors. Also, the transmission mechanism between policy interest rates and base money growth could be explored in greater detail.
References
Alamsyah, H., C. Joseph, J. Agung, and D. Zulverdy, 2001, “Toward Implementation of Inflation Targeting in Indonesia.” Bulletin of Indonesian Economic Studies, 37, 309–324.
Bank Indonesia, 2001, Credit Crunch in Indonesia in the Aftermath of the Crisis: Facts, Causes, and Policy Implications.
Bank Indonesia, 2002, “Demand for Money and Monetary Policy Implication: Preliminary Study,” Mimeo.
Basri, M.C., A. Damajanti, and I.K. Sutisna, 2002, “Sources of Inflation: A Simple VAR Analysis.” Mimeo.
Dekle, R. and M. Pradhan, 1997, “Financial Liberalization and Money Demand in ASEAN Countries: Implications for Monetary Policy.” IMF Working Paper No: 97/36.
Dekle, R. and M. Pradhan, 1999, “Financial Liberalization and Money Demand in the ASEAN Countries,” International Journal of Finance and Economics, Vol. 4, 205–215.
Enoch, C., B. Baldwin, O. Frécault, and A. Kovanen, 2001, “Indonesia: Anatomy of a Banking Crisis. Two Years of Living Dangerously, 1997-99,” IMF Working Paper No: 01/52.
McLeod, R. (1997), “Explaining Chronic Inflation in Indonesia,” Journal of Development Studies, Vol. 33 392–410.
Ramakrishnan, U., and A. Vamvakidis, 2002, “Forecasting Inflation in Indonesia,” IMF Working Paper (forthcoming).
Sriram, S.S., 2001, “A Survey of Recent Empirical Money Demand Studies,” IMF Staff Papers, Vol. 47 334–365.
This chapter was prepared by Alexander Wolfson (APD).
In addition, the ratio of base money-to-GDP experienced permanent increases when BI increased required reserve ratios in February 1996 and April 1997.
The other components of base money, namely cash in vaults and commercial banks’ deposits at BI are smaller and, except for year-end spikes, exhibit less variation.
For example, the income elasticity is calculated by dividing the coefficient on real income by one minus the coefficient on lagged real money.
Specifications using seasonally adjusted data yield similar results.
Their specification includes a linear time trend and year-specific dummy variables to account for ongoing liberalization of the financial system.
See Bank Indonesia (2001) for an examination of the impact of the banking crisis on the monetary transmission mechanism.
The estimated semi-elasticity of -.0018 in the specification for currency implies that a 100 basis point increase in deposit rates reduces currency demand by 0.18 percent.
For the remainder of this chapter, this specification is considered the preferred model.
Specifications which also include a crisis dummy to account for level changes in currency demand yield similar estimates of the income elasticity.
Although somewhat counter-intuitive, this result mirrors the findings of BFs Research Department (mimeo, 2002), and could suggest that depreciation acts as a leading indicator of future inflation, causing individuals to increase their holdings of currency in anticipation of future price rises.
