Information about Asia and the Pacific Asia y el Pacífico

CHAPTER 2 Inflation Uncertainty and the Term Premium

Thomas Rumbaugh
Published Date:
January 2012
  • ShareShare
Information about Asia and the Pacific Asia y el Pacífico
Show Summary Details

Based on the term structure model for determining nominal bond yields, this chapter identifies the impact on the cost of borrowing of Indonesia’s relatively higher inflation level and volatility relative to its peers. The higher inflation volatility in Indonesia creates greater uncertainty in forecasting inflation, resulting in a relatively higher inflation risk premium.

Indonesia’s consumer price inflation level and volatility have been historically higher than some of its peer emerging-market economies.

  • Annual inflation in Indonesia, as measured by the consumer price index (CPI), has averaged nearly 12 percent since 1997 and 8.5 percent since the formal adoption of the inflation-targeting framework in July 2005 (Figure 2.1). By comparison, inflation rates for some of Indonesia’s Asian peers, such as Malaysia, Thailand, and the Philippines, have averaged about 3–6 percent since July 2005.
  • Average inflation volatility in Indonesia has also been significantly higher than that of its peers (Figure 2.2). The spikes in Indonesia’s inflation volatility are correlated with administrative price adjustments (Table 2.1). Even core inflation in Indonesia has been highly volatile, as second-round effects from administered energy price increases pass through to the broader economy (correlation coefficient between core and energy inflation = 0.75). Deviations of the inflation outcome relative to annual inflation targets—which have typically been adjusted in anticipation of administrative price increases—are higher on average than those of the comparator group (Figure 2.3).

Figure 2.1Consumer Price Inflation (Year-on-year)

Source: IMF staff estimates.

Figure 2.2Consumer Price Inflation Volatility

Source: IMF staff estimates.

Note: Twelve-month rolling volatility based on annual average inflation.

TABLE 2.1Administrative Price Adjustments (percent)
Fuel typeFebruary 2005October 2005June 2008December 2008January 2009
Auto diesel27.3104.827.9-6.4-6.8
Source: IMF staff estimates.
Source: IMF staff estimates.

Figure 2.3Annual Inflation Deviation from Inflation Target

Source: IMF staff estimates.

Note: Malaysia and India do not follow an inflation target.

The historical volatility of Indonesia’s inflation appears to contribute to uncertainty about estimates of its future inflation. The dispersion of CPI survey forecasts can be used as a proxy for uncertainty about these forecasts (among others, Durham, 2006; and Wright, 2009). Based on data from Consensus Forecasts, the 12-month moving average standard deviation of forecasts for Indonesia’s year-ahead CPI historically has been much higher than those for Malaysia and Thailand (Figure 2.4). 1

Figure 2.4Standard Deviation of Year-Ahead CPI Forecasts (12-month rolling average)

Source: IMF staff estimates.

Both theoretical and empirical evidence show that high volatility and unpredictability of inflation creates economic costs. Studies have identified a negative relationship between both the inflation level and its volatility relative to income growth (see, for example, Judson and Orphanides, 1999). Among the channels through which high and volatile inflation creates economic costs is a higher cost of capital. 2 Indeed, Indonesia’s domestic (and international) borrowing costs have been higher than those of comparable emerging-market economies (Figure 2.5). 3

Figure 2.5Ten-Year Government Local Currency Debt Yield

Source: IMF staff estimates.

Against this background, this chapter examines the term premium on Indonesia’s domestic government yields relative to that of peers to illustrate the impact of inflation uncertainty on borrowing costs. The term premium—that is, the nominal premium sought by investors to compensate for delaying consumption (real term premium) and for inflation uncertainty (inflation risk premium) as explained in more detail in the next section—is calculated using two methodologies for Indonesia relative to other countries. The analysis finds Indonesia’s distant-horizon forward rates (which abstract from the near-term monetary policy stance) are consistently above those of its peers. The findings suggest that Indonesia’s term premium is on average higher than those of its peers, as would be expected as a result of higher inflation uncertainty. The results provide useful information for policymakers given that enhanced monetary policy credibility has been found to lower term premiums on developed-country government yield curves and, by extension, borrowing costs to the wider economy.


A domestic economy’s benchmark borrowing cost is usually determined by government borrowing rates. Government bond yields comprise an average expected future real short-term interest rate over the length of the bond, expected inflation over the length of the bond, and a nominal term premium. The nominal term premium is made up of a real term premium and an inflation risk premium. The real term premium is what investors demand for tying up their funds and delaying consumption. The inflation risk premium is what they demand as additional compensation for uncertainty about expected inflation.


Recent studies have used developed-country yield curves to estimate term premiums and explain downward shifts in long-term borrowing costs. These studies have grown out of the “conundrum” question as to why long-term interest rates in the United States and euro area countries underwent a sustained decline in the middle of the last decade. Kim and Wright (2005) show that much of the decline in long-term U.S. Treasury yields to 2004 can be explained by a decline in term premiums. Among the factors they suggest as possibly leading to a drop in the term premium are increased attractiveness of longer-maturity obligations resulting from better anchored inflation expectations and a decline in the volatility of real activity, foreign official reserve purchases of developed-country government debt, regulations that encourage pension funds to better match assets and liabilities, reduced home bias of foreign investors, and demographic trends. Likewise, in a cross-country study of developed-country yields, Wright (2009) finds that those countries that reduced inflation uncertainty saw a decline in term premiums.

These studies, however, note the difficulty of isolating the factor—either the real term premium or the inflation risk premium—driving down the nominal term premium. For countries that issue inflation-indexed bonds, it is theoretically possible to isolate the inflation risk premium from the term premium, although the relative liquidity of nominal versus inflation-indexed notes is a major factor distorting estimates. The difference between the distant-horizon forward rate derived from yields on a nominal government bond curve and a similarly derived forward real rate from the inflation-indexed government bond curve comprises expected inflation, a (forward) inflation risk premium, minus a liquidity premium that investors charge to purchase less-liquid inflation-indexed securities. 4 Using well-developed surveys of inflation expectations for these countries, it would be possible to identify the value of the inflation risk premium minus the liquidity premium. However, given the lack of inflation-indexed bonds in a majority of emerging markets, so far such studies are limited.


The analysis in this chapter is based on a term structure model for determining nominal bond yields. The basic relationship is defined as follows: the nominal forward interest rate (i) derived from government bond yields equals the sum of the expected short-term real rate (re), expected inflation (πe), real term premium (rTP), and inflation risk premium (πRP). The sum of real term premium and inflation risk premium (rTP + πTP) equals the nominal term premium. Thus, for a 1-year forward rate:

As discussed above, the real term premium compensates investors for delaying consumption for one additional year, and the inflation risk premium is the additional premium investors demand to compensate them for inflation uncertainty.

Rearranging equation (2.1) gives a simple measure of the term premium:

The term premium is estimated as the n-year forward rate less the expected future short rate less expected inflation. The advantage of using distant-horizon forward rates is that they abstract from the current monetary policy stance and near-term monetary policy expectations.

For the countries in the study, a number of data approximations were made to estimate the term premium. 5 Accordingly, two methodologies are applied to extract term premium estimates from nominal bond yields. In both methodologies, distant-horizon nominal forward rates are calculated using local currency government debt yields. Given data limitations, the findings are best interpreted as a relative measure—that is, the level of Indonesia’s term premium relative to other comparators—rather than an absolute estimate of the term premium for each country (see also footnote 7).

  • Methodology I: The term premium identified in equation (2.2) is estimated using monthly data as follows.
    • ◦ The distant-horizon forward rate is calculated as the 1-year rate, 9 years forward, and is called the “1-year forward rate” for simplicity. It is calculated using 9- and 10-year government debt yields. 6 The forward rate formula is
      where fm,n is the forward rate between m- and n- period bonds, Dn is the duration of the n- period bond, Dm is the duration of the m-period bond, Yn is the yield on the n- period bond, and Ym is the yield on the m- period bond7. For this study, the maturities used were n = 10 years and m = 9 years.8 As already noted, a distant-horizon 1-year forward rate, rather than a 1-year government bond yield, is used because distant-horizon rates abstract from the short-term monetary policy stance relative to the cyclical position. If the bond yield under consideration were to include short-term monetary policy expectations, isolating the term premium would be rendered even more difficult.9
    • ◦ The short-term expected real interest rate is proxied by a time-invariant rate that reflects the underlying real interest rate in the economy. To calculate this rate, the 1-month central bank bill rate and actual annual core inflation are used.10,11 The monthly real rate is then averaged for the period January 2001 to December 2010.12
    • ◦ As a proxy for expected inflation, it is assumed that investors have perfect foresight; that is, expected inflation was assumed to equal the 12-month-ahead actual core inflation.
  • Methodology II: This method offers an alternative estimate of Indonesia’s term premium relative to its peers. For this method, it is assumed that the expected real short-term interest rate is the same across comparator countries. 13 Thus, the following equation gives the difference in the five-year term premiums between Indonesia and a comparator country. The equation does not give the level of term premium for each country.
    where YieldIDN is the nominal five-year rate five years forward (or the five-year forward rate, to simplify) for Indonesia, YieldCountry Y is the five-year forward rate for the comparator country, and E(Π) is expected inflation 5 to 10 years ahead as reported in Consensus Forecast survey results. 14 The sample period runs from 2003 to early 2011.


Analysis of forward rates based on Methodology I illustrates that Indonesia has a relatively higher term premium than its peers. For the period June 2005 to February 2011, Indonesia’s term premium has, on average, been higher than those for Malaysia, the Philippines, and Thailand. Focusing on broad trends, although the term premiums for all the countries, including Indonesia, were trending down until about late 2007, Indonesia’s term premium subsequently rose substantially more than those of other countries and has stayed at a somewhat elevated level relative to the comparator group (Figure 2.6). Despite an increase in inflation volatility for all the sampled countries after the second half of 2008—when they were struck by the global food and fuel price shock of 2008 and the financial crisis in 2009—implying generally uniform shocks to all countries, Indonesia’s term premium increase has remained persistently higher, with the exception of a brief period in the second half of 2010 when Indonesia’s forward rates were briefly lower than those of some of its peers. 15 A simple regression of the term premium on core inflation volatility, with controls for seasonal movements, indicates that nearly two-thirds of the change in term premium during the selected time period arises from inflation volatility, suggesting that a higher inflation risk premium could be driving Indonesia’s higher term premiums.

Figure 2.6Term Premiums by Country

Source: IMF staff estimates.

Analysis of forward rates based on Methodology II also suggests that, on average, Indonesia has a higher term premium than its peer countries. Through the period examined, Indonesia almost always had higher long-term inflation expectations than the peer group (Figure 2.7). Indonesia also had higher forward rates than comparator countries (Figure 2.8).

Figure 2.7Long-Term Inflation Expectations: Annual Average of Expected Inflation 5–10 Years Ahead

Source: Consensus Forecasts.

Figure 2.8Five-Year Forward Rates

Source: IMF staff estimates.

The results from Methodology II also illustrate the extent to which higher expected inflation rates alone do not explain Indonesia’s higher forward rates. The additional returns that investors perpetually require in Indonesia in excess of the higher expected inflation (Πe IDN) relative to each peer country (Πe Country Y) are reflected in Indonesia’s higher term premium relative to comparator countries, with the exception of a few months in 2010 when Indonesia’s relative term premium was lower than that in the Philippines (Figure 2.9).

Figure 2.9Indonesia’s Relative 5- to 10-Year Nominal Term Premium (Difference between Indonesia and other countries)

Source: IMF staff estimates.


Indonesia’s perpetually higher term premium illustrates the cost to the government, and by extension to the wider economy, of investor uncertainty about inflation risk. The higher term premium does not arise simply because investors expect higher inflation in Indonesia (estimating the term premium already accounts for the higher expected inflation using the actual 12-month-ahead inflation in Methodology I, and survey expectations in Methodology II). Term premium estimates quantify the compensation investors require, on top of their expectations for inflation, for their relative inability to predict inflation, which poses an additional risk to their real returns.

The term premium imbedded in the yield curve could be useful for judging the extent to which monetary policy is anchoring inflation expectations. Large and persistent inflation fluctuations increase investor uncertainty about future inflation, and investors thus demand a higher premium as compensation for this risk. If a government is paying a large term premium because of a high inflation risk premium, financing costs could be lowered by issuing inflation-indexed bonds. 16

  • The relatively higher inflation volatility for Indonesia and larger deviation of actual inflation from forecasts compared with other countries suggest that investors have a higher degree of inflation uncertainty for Indonesia. In addition, the dispersion of survey forecasts indicates that survey participants are more uncertain about their forecasts of inflation in Indonesia than they are for the comparator countries.
  • An explanation for higher inflation uncertainty in Indonesia is that monetary policy has not anchored inflation expectations as successfully as monetary policy in the peer group has. More specifically, ahead of the inflation bout in 2005–06—when one round of administrative price adjustments occurred—policy rates were low compared with Taylor rule estimates (Figure 2.10). 17 This stance may have exacerbated the subsequent inflation pressures rising from the administrative price hikes, leading to a large miss relative to the inflation target. In 2008, although policy accommodation was likely appropriate given external conditions, another large miss of inflation relative to target occurred, when an administrative price hike happened in tandem with the global food and fuel price shock (Figure 2.11). Notwithstanding the limitations of the estimated Taylor rule, these two episodes, combined with ongoing political discussions about the timing and extent of future administrative price hikes, could be contributing to higher perceived inflation risks. Even during periods of low global and domestic inflation, Indonesia’s term premium remains higher than that in comparator countries. This difference is likely related to investors’ continued uncertainty about the likelihood that an appropriate level of inflation will be realized on average over time.

Figure 2.10Indonesia: Monetary Policy Stance

Source: IMF staff estimates.

Figure 2.11Inflation and Interest Rate

Source: IMF staff estimates.

Note: SBI = Bank Indonesia certificate.

How can inflation expectations be anchored and the inflation risk premium lowered?

  • As discussed in the literature, countries that established higher levels of monetary policy credibility saw a decline in the term premium on their domestic government debt. A relatively aggressive monetary policy response to emerging inflation pressures has a near-term cost to the economy because it dampens growth. However, in the long run, well-anchored inflation expectations will help depress the nominal cost of capital by lowering both expected inflation and the inflation risk premium, supporting long-term growth. Greater monetary policy credibility will be established with a track record of meeting inflation targets.
  • In addition, effective communication with market participants about how inflation targets will be set and met is also necessary to better anchor expectations. In particular, Bank Indonesia (BI) could improve its signaling of adjustments to the monetary stance, thus clarifying the monetary policy reaction function. Communicating that BI is committed to meeting the middle of the inflation target band on average, over time, would be effective in dampening expected inflation volatility and the inflation risk premium. Targeting progressively lower levels of inflation going forward, in line with its trade partners, could help lower volatility and reduce Indonesia’s borrowing costs.
  • Gaining policy credibility also requires that monetary operations be consistent with the announced monetary stance. Consistency and transparency of monetary operations in line with the announced stance are necessary to achieve policy credibility. 18 In this regard, the measures announced in 2010 for liquidity management and interbank market development are steps in the right direction to help improve monetary operations. 19

    AcemogluD.S.JohnsonJ.Robinson and Y.Thaichar oen2003Institutional Causes, Macroeconomic Symptoms: Volatility, Crises and Growth,Journal of Monetary Economics Vol. 50January pp. 49123.

    • Search Google Scholar
    • Export Citation

    CampbellJ.A.W.Lo and A.C.MacKinlay1996The Econometrics of Financial Markets (PrincetonNew Jersey: Princeton University Press).

    • Search Google Scholar
    • Export Citation

    DurhamJ. Benson2006An Estimate of the Inflation Risk Premium Using a Three-Factor Affine Term Structure Model,FEDS Working Paper No. 2006–42. Available via the Internet:

    • Search Google Scholar
    • Export Citation

    GoyalR. and M.Ruiz-Arranz2009Explaining Indonesia’s Sovereign Spreads,” in “Indonesia—Selected Issues,IMF Country Report No. 09/231 (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation

    HordahlP.2008The Inflation Risk Premium in the Term Structure of Interest Rates,BIS Quarterly Review (September) pp. 2338.

    • Search Google Scholar
    • Export Citation

    JudsonRuth and AthanasiosOrphanides1999Inflation, Volatility and Growth,International Finance Vol. 2 No. 1 (April) pp. 11738.

    • Search Google Scholar
    • Export Citation

    KimD.H. and J.H.Wright2005An Arbitrage-Free Three-Factor Term Structure Model and the Recent Behavior of Long-Term Yields and Distant-Horizon Forward Rates,Finance and Economics Discussion Series 2005–33 (Washington: Board of Governors of the Federal Reserve System).

    • Search Google Scholar
    • Export Citation

    PoirsonHelene2009Monetary Policy Communication and Transparency,Chapter 10 in India: Managing Financial Globalisation and Growth ed. by KalpanaKochhar and CharlesF. Kramer (New Delhi: Business Standard Books).

    • Search Google Scholar
    • Export Citation

    TaylorJohn B.1993Discretion versus Policy Rules in Practice,Carnegie-Rochester Conference Series on Public Policy Vol. 39 pp.195214.

    • Search Google Scholar
    • Export Citation

    WrightJonathan H.2009Term Premiums and Inflation Uncertainty: Empirical Evidence from an International Panel Dataset,Finance and Economics Discussion Series 2008–25 (Washington: Board of Governors of the Federal Reserve System).

    • Search Google Scholar
    • Export Citation

Dispersion for the Philippines is not shown owing to lack of a long enough time series in Consensus Forecasts.


Although high levels of inflation and volatility can be welfare reducing, low inflation and price stability are not sufficient conditions to achieve higher growth, especially if the supporting economic and institutional environment is weak (see Acemoglu and others, 2003).


See Goyal and Ruiz-Arranz (2009) for factors determining Indonesia’s sovereign external spreads relative to its peers.


See Durham (2006) and Hordahl (2008) and references therein for a review of previous studies.


In addition to Indonesia, this analysis estimates term premiums for Malaysia, India, the Philippines, Thailand, and Mexico. Methodology I is not applied to India because of the absence of bond yield data of contiguous maturities necessary to estimate a 1-year forward yield.


Generic government yield time series are used as constructed by Bloomberg (i.e., each benchmark 10-year bond yield rolls into the new issue).


In the absence of zero-coupon yields, it is assumed that duration equals maturity, that is, D(n) = nand D(m) = m,for the 9- and 10-year bonds. A test was done using precise duration calculations for several data points and the magnitude of the difference in the forward rates was small. Because this simplification is applied across all the countries in the study, the comparative findings have greater meaning than absolute estimates of the term premium. For a detailed derivation of this formula, see Campbell, Lo, and MacKinlay (1997).


In the absence of a complete data series on the 9- and 10-year government yields for the Philippines, the implied forward rate is calculated on the 4- and 5-year government yields.


For example, the 1-year government bond yield will comprise the actual current short-term rate (say 1-month) and average expected short-term rates out to one year plus a term premium. The 1-month rate will be determined almost entirely by the current monetary stance, and expected future short-term interest rates will be determined almost entirely by expected monetary policy moves. The distant-horizon forward rate abstracts from monetary policy expectations and comprises a real return to capital (the time-invariant real rate used here), expected inflation, and a term premium.


Using headline inflation is likely to bias inflation expectations upward and real rates downward because of the large spikes arising from the administered price changes. Using core inflation corrects this problem. Moreover, because the analysis uses distant-horizon forward rates, future changes in core inflation are a better approximation of expected inflation over time.


A time-variant real rate was not used because sporadic negative real rates during periods of high inflation distort the underlying long-term economic real interest rate. In addition, the real return to capital adjusts slowly based on the capital-to-labor ratio and thus a long-term average is more appropriate than are monthly observations.


In line with the explanation in footnote 14, the sporadic negative real rates are removed because they would otherwise bias the real rate downward.


Such an assumption is suitable for emerging-market countries at broadly similar stages of economic and market development.


The benefit of using long-term inflation expectations (e.g., average expected inflation 5–10 years ahead) is that these abstract from near-term factors that impact inflation expectations, such as administered price increases and commodity price pass throughs. Instead, long-term inflation expectations get to the level of inflation expected to be targeted or managed by monetary policy on average, over time. Five-year forward rates are used instead of 1-year forward rates in Methodology II because the 5-year forward rate matches up with the 5–10 years ahead annual inflation expectations as reported by Consensus Forecasts. Forward rates are calculated as described in Methodology I.


This was also when the second-round inflationary effects related to Indonesia’s 2008 administrative price increase took hold.


The caveats are, however, that governments will have to pay out relatively more on inflation-indexed bonds if actual inflation ends up higher than inflation expectations imbedded in nominal bond yields; also, the liquidity premium demanded by investors to buy less-liquid inflation-indexed bonds may erode savings from eliminating the inflation risk premium. Of course, issuing a greater proportion of short-term debt would lessen the term premium the government pays, but doing so would raise rollover risks, reduce liquidity in remaining longer-dated issues, and eliminate an important benchmark for private sector long-term borrowing.


Taylor rule estimates are derived using potential output measures based on the Hodrick-Prescott filtering technique, and Bank Indonesia’s annual inflation targets. The Taylor rule provides a framework for evaluating the stance of monetary policy and the level of the nominal interest rate (see Taylor, 1993).


Poirson (2008) delves into these issues in the discussion of monetary policy communication for India.


On June 16, 2010, BI announced measures that include a 1-month minimum holding period on Bank Indonesia certificates (SBIs); lengthening of SBI maturity, including the introduction of 9- and 12-month bills; widening of the interest rate corridor by 100 basis points to 200 basis points; a 1-month term deposit facility; and an initiative to facilitate triparty repo trading.

    Other Resources Citing This Publication