Inflation targeting in Indonesia, 1999–2012: an ex-post review

Abstract

Around the end of 1999, Bank Indonesia (BI) adopted inflation targeting as part of its approach to monetary policy. This article reviews the experience up to 2012, examines BI's performance in hitting its inflation targets and considers certain broader indicators of success. Overall, inflation targeting in Indonesia has been a messy, evolutionary process, and BI's implementation record compares unfavourably with that of its peers. Yet Indonesia recorded a significant downward trend in inflation during this period and maintained strong economic growth. Also, almost all of its inflation-targeting arrangements are now in line with common international practice. Looking ahead, this article offers suggestions for sustaining progress in inflation targeting, such as setting more ambitious targets in the outer years and implementing strong policies to reduce inflation further, including after large administrative price shocks.

Pada akhir 1999, Bank Indonesia (BI) mengadopsi kerangka kerja penetapan sasaran inflasi (inflation targeting) sebagai bagian dari pendekatan BI terhadap kebijakan moneternya. Tulisan ini mengulas pengalaman BI sampai tahun 2012, termasuk kinerja BI dalam mengejar target inflasinya serta indicator-indikator kesuksesan lainnya. Secara umum, penargetan inflasi di Indonesia adalah sebuah proses evolusioner yang kurang mulus. Kinerja BI juga masih kalah dibandingkan dengan bank sentral yang lain. Namun, data memperlihatkan penurunan inflasi yang signifikan serta pertumbuhan ekonomi yang kuat di Indonesia pada periode tersebut. Juga, hamper semua kebijakan terkait penetapan sasaran inflasi di BI kini telah selaras dengan praktik internasional. Tulisan ini juga merekomendasikan beberapa kebijakan untuk meraih pencapaian yang lebih berkelanjutan, misalnya dengan menetapkan sasaran yang lebih ambisius di masa yang akan datang serta mengimplementasikan kebijakan yang lebih tegas dalam mengurangi inflasi, khususnya setelah terjadi lonjakan harga-harga administratif.

Keywords:

• inflation targeting
• monetary policy

Acknowledgements

The author wishes to acknowledge the many helpful comments of two anonymous referees. Their suggestions significantly improved the quality of this article.

Notes

¹ Source: <http://databank.worldbank.org/ddp/home.do>.

² Article 7 states that BI's objective ‘is to achieve and maintain stability of the rupiah value’. At the time, it was unclear whether this was intended to mean low inflation or stability in the exchange rate. BI's mission and vision statements clearly prioritise low inflation (see the preface to any of BI's annual or economic reports from 2001 to 2011).

³ See BI (2008: 120; 2009: xvii, 169; 2010: xx, xxi, xxv, 3). Table 11.9 in BI (2002: 215) presents an indicative, linear target path for 2003–06, which is consistent with this article's analysis.

⁴ Namely, ‘Any change in the monetary policy stance is undertaken after evaluating whether future developments in inflation are on track with the established inflation target …. At the operational level, the monetary policy stance is reflected in the setting of the policy rate (BI Rate) with the expectation of influencing money market rates and in turn the deposit rates and lending rates in the banking system. Changes in these rates will ultimately influence output and inflation.’ See <http://www.bi.go.id/web/en/Moneter/Kerangka+Kebijakan+Moneter/>. See also the discussion in the text, under ‘A new beginning: 2004–06’.

⁵ This is consistent with Alamsyah et. al. (2001).

⁶ BI research (BI 2000: 160) shows that monetary policy needs more than one year to affect the real economy. Seasonal or cyclical factors also need time to unwind, including those (such as Ramadan) that are not fully captured in a 12-month calculation of inflation.

⁷ The best answer was found in Goeltom (2007: 146) – namely, that a tight target specified for a short, one-year time horizon would help to restore BI's credibility, provided that the target could be achieved (italics added). In light of actual inflationary developments, Goeltom argued that BI was constrained by both supply shocks (oil and electricity prices) and demand shocks (civil-service wage increases). This is consistent with the BI governor 's explanation noted in the text.

⁸ The cautious wording is notable: ‘In the next five years Bank Indonesia has committed to gradually lowering the inflation rate to around 6%–7%’ (author 's translation from the Bahasa Indonesia version, BI 2001: ix).

⁹ This is surprisingly long, for several reasons. First, the jump from a one to a five-year time horizon is large. Second, five years is considerably longer than the time required for monetary policy to have an impact (generally understood to be one to two years). And, third, few countries operate with a time horizon as long as five years, although several use a ‘medium-term approach’ (interpreted as two years or more) or have an ‘at all times’ perspective (Hammond 2012: 9).

¹⁰ BI (2002) ‘sets the inflation target for 2003 at 9% with a 1% error margin’ (p. 18). It is unclear whether this means 8.5%–9.5% or 8%–10%. The former looks more plausible, because it would have been consistent with the 2006 target of 6%–7% (the target presented in figure 1). BI (2003: 53) says that the target at the beginning of 2003 was 9% with a deviation of +/– 1%. This is strange, however, because it implies a narrowing target width from 2003 to 2006. In the event, it matters little – actual inflation remained well below both targets until they were superseded, in September 2004 (see later in the text).

¹¹ Neither the amendment nor the elucidation is specific about the means that BI must use to achieve this goal. They refer both to monetary targets and to other means, such as open-market operations, discount rates, bank reserve requirements and credit regulations. Elucidation 3 explicitly maintains the dichotomy (see footnote 2) that BI's responsibility for maintaining the rupiah's value includes both stable prices and a stable exchange rate.

¹² The MOU provides wide leeway for certain details of the targets. They can be set in terms of headline or core CPI inflation, and they can be a range, a point or a point with deviation.

¹³ The Ministry of Finance Decree of 6 September 2004 included, in article 3, an explicit provision for accountability: ‘Bank Indonesia and the government will openly explain to parliament and the public the causes of failing to achieve the inflation target’ (author 's translation).

¹⁴ The TPI seems to have started operating in 2005. In due course, members included BI, the Ministry of Finance, the Coordinating Ministry for Economic Affairs, the Ministry of Trade, the Ministry of Communications, the Ministry of Energy and Mineral Resources, the National Development Planning Agency, the Ministry of Agriculture, and the Ministry of Labour and Transmigration (BI 2010: 39).

¹⁵ At first, TPIDs were set up in only a few regions across Indonesia; by 2010, they were operating in 53 regions (BI 2010: 39), apparently with the aim of covering the 66 urban areas included in CPI calculations.

¹⁶ Given that there was no clear announcement for 2004, figure 1 does not include a formal target for that year. Appendix table A1 notes the informal target of 5.5% +/– 1%.

¹⁷ Previously, the target for 2006 had been 6%–7% (appendix table A1).

¹⁸ There was no mention of the old 2006 target of 6%–7%. The report does refer to a 2004 target of 6%–7% for core inflation (BI 2004: 51), but this looks more like a projection than a target. It seems coincidental that it is the same as the old target for 2006. Sarwono (2007) does not cite an official 2004 target but indicates a target range of 4.5%–6.5% in one of his charts.

¹⁹ Regulatory policy reduced banks’ net open positions and strictly limited derivative transactions (BI 2005: 70–2), mainly to protect the exchange rate. Also, the TPI was introduced (footnote 14), and there was ‘measured intervention’ in the foreign-exchange market (BI 2005: 5).

²⁰ For a discussion of the considerations in choosing between base money and interest rates, see, for example, BI (2004: 21) and Hossain (2012: chapter 7).

²¹ As envisaged (BI 2005: 102), the BI rate would be set quarterly (in January, April, July and October), unless adjusted, in special circumstances, at other monthly BI board meetings. For all practical purposes, the BI rate has been set monthly.

²² The BI rate is a ‘signal’, as mentioned in the text; FASBI (Fasilitas Simpanan Bank Indonesia) is BI's key short-term (overnight) policy rate; and JIBOR (Jakarta Interbank Overnight Rate) is a sensitive indicator of liquidity conditions in the interbank money market. The overnight FASBI was inactive between April and September 2005, which accounts for the gap in figure 4 (BI 2005: chapter 5). Figure 4 also uses the one-month Bank Indonesia Certificate (Sertifikat Bank Indonesia, SBI) rate before July 2005 (when the BI rate was introduced), in line with BI's convention (BI 2005: chart 5.17).

²³ In June, July and August 2010, BI press releases (see <http://www.bi.go.id/web/en/Ruang+Media/Siaran+Pers/>) mention a target of 4%–6% for 2010 and 2011. These press releases pre-date the Ministry of Finance decree of 24 August 2010.

²⁴ BI's target is cited in the governor 's foreword to the BI's Annual Report for 2009, dated March 2010. The government's announcement is dated 24 August 2010 (appendix table A1).

²⁵ For example, Svensson (2009) points out that as the time horizon lengthens, the likelihood increases of some change in circumstances rendering the initial policy settings inappropriate.

²⁶ Still, it is the time horizon against which BI reports regularly; see <http://www.bi.go.id/web/en/Moneter/Inflasi/Bank+Indonesia+dan+Inflasi/penetapan.htm>.

²⁷ Assessments of two and three-year targets were not included if the initial target was superseded in any appreciable way by subsequent changes to the second or third-year target. For example, the three-year target for 2005-07 was not included in table 3, because the target for 2007 was raised by one percentage point in August 2007. See appendix table A1.

²⁸ These data for Indonesia begin in July 2005 (Roger 2009: 6), which implies they consider only the government's targets, and not those announced by BI.

²⁹ These comparisons are slightly less unfavourable (at 33% and 2.4%, respectively) for Indonesia when compared with low-income countries in a ‘disinflation phase’.

³⁰ For their sample, in 2002–10, Indonesia's inflation rate was inside its targets 13% of the time.

³¹ To evaluate such a monetary-policy regime, Sevensson (2009) suggests analysing deviations between inflation gaps and output gaps, and taking into account the information available when the policy decision was made. This would likely be a very challenging task for Indonesia, in that detailed data would be needed on the information (and its quality) available at BI board meetings. Grenville and Ito (2010) attempt a simplified version of this for Thailand, by estimating a Taylor-rule reaction function.

³² The t-statistic on the trend coefficient is 3.7.

³³ The t-statistic rises to 7.2, and the implied drop-off in inflation is almost 0.65 per annum.

³⁴ BI staff consistently distinguish among core inflation (which they believe is influenced by monetary policy and accounts for about two-thirds of the CPI); volatile food inflation (which, by definition, is considered to be temporary, with a weight of about 15% in the total CPI); and administered inflation (which is determined by governmental policy decisions, at least on timing, with a weight of less than 20% in the CPI).

³⁵ BI reports regularly to parliament; it publishes quarterly reports on monetary policy and monthly reviews of monetary policy; and senior BI staff often speak with the press.

³⁶ Indeed, Indonesia's target mid-point is lower than that of only Ghana (8.7%) and Turkey (5%). Among regional comparators, Thailand and Korea are at 3% and the Philippines is at 4% (Hammond 2012: 9, including for the average mid-point in the text).

³⁷ This is consistent with Roger 's (2009) tentative conclusion that inflation-targeters were better than non-targeters at coping with the financial shocks of 2007–09.

³⁸ Subsequent analysis by Filardo and Genberg (2010: 259, 271) found a drastic reduction in the mode of private-sector inflation forecasts for inflation-targeters in the region but less evidence of reduced dispersion among the forecasters, including in Indonesia.

³⁹ By way of example, an anonymous referee questioned two of the targets in an early draft of this article. One of them looks like a translation error (see the following footnote and appendix table A1 for 2012–13). The other is correct, as evidenced by a statement in the governor 's foreword to BI's Economic Report for 2009.

⁴⁰ This article does not include this target in its analysis, because the Bahasa Indonesia version of BI (2011) does not refer to it as a target such (sasaran). There appears to have been a translation error in the English version of the report. See appendix table A1.

⁴¹ Friedman (1969) goes further, arguing that central banks should keep nominal interest rates at zero by pursuing a rate of deflation equal to the interest rate on government (riskless) bonds.

⁴² The situation is even less clear for developing countries. For example, Chowdhury and Ham (2009) find exploratory evidence for Indonesia that the threshold is between 8.5% and 11.0%. But Hossain (2012: 165) suggests that BI should consider price stability (italics add- ed) as its method of reducing inflation and inflationary uncertainty. For their part, Sugema and Bakhtiar (2010) find that monetary policy in Indonesia can effect inflation in the short run but not in the long run, which undermines the credibility of setting any inflation target.

⁴³ See <http://www.tradingeconomics.com/>.

⁴⁴ The former BI governor 's speech is available at <http://www.bi.go.id/web/en/Ruang+Media/Pidato+Dewan+Gubernur/speech_darmin_bd_2010.htm>.

⁴⁵ See, for instance, BI's Monetary Policy Report: Quarter IV 2012, or BI (2011: xxiii).