This Selected Issues paper reviews developments during 2001–02 in the Belarusian pension system, and discusses potential short- and medium-term reforms needed to put pension system finances on a sustainable path. The paper presents a short overview of demographic trends and developments in Belarus and their impact on the pension system. It discusses the financial difficulties of early 2002 and the policy measures the authorities took to deal with them. A critical review of the methodology used in preparing annual Social Protection Fund budgets is also presented.

Abstract

This Selected Issues paper reviews developments during 2001–02 in the Belarusian pension system, and discusses potential short- and medium-term reforms needed to put pension system finances on a sustainable path. The paper presents a short overview of demographic trends and developments in Belarus and their impact on the pension system. It discusses the financial difficulties of early 2002 and the policy measures the authorities took to deal with them. A critical review of the methodology used in preparing annual Social Protection Fund budgets is also presented.

III. Core Inflation in Belarus1

Executive Summary

In most economies, the short-term dynamics of inflation are strongly influenced by non-monetary factors such as sudden and large changes in the prices of food and energy (“supply shocks”).2 In a transition economy like Belarus, these supply shocks play an even more prominent role. Increases in utility tariffs for households frequently put pressure on the overall price level. Moreover, the weight of food items in the consumption basket is high (about 64 percent), which exacerbates the seasonal component of CPI inflation.

In this paper, we construct a measure of inflation that removes supply shocks from the headline CPI inflation. We confirm the common intuition that supply shocks, in particular those related to hikes in utility prices, have been quite strong in Belarus, especially during 2002. The combined impact of deregulations and other shocks is estimated at 12 percent last year, compared with annual inflation of 35 percent.

To examine the short-term impact of supply shocks on inflation, we construct a simple model of Belarusian inflation and estimate it over the recent period of relative monetary stability (2000-2002). We find that the estimated model shares many features with standard inflation models estimated on data from advanced economies. Inflation is positively correlated with output and the underlying inflation rate is highly persistent. We also find that the supply shocks only have temporary effects on inflation. At the same time, they have a highly persistent effect on the price level. This finding can be interpreted in the sense that the monetary authority accommodates supply shocks to minimize any output effects arising from deregulation.

A. The Concept of Core Inflation: Motivation

1. It is important to estimate core inflation because it removes sector-specific factors from the headline inflation rate. Thus, it better reflects the underlying monetary component of price increases.3 The most important sources of supply shocks in Belarus have been (i) gradual price liberalization/administrative hikes of prices in various sectors (particularly utilities), and (ii) seasonal fluctuations in food prices.

2. The main benefits of computing core inflation in Belarus are:

  • It enables a more subtle interpretation of historical episodes of high inflation. In certain periods such as early and late 2002, it has not been apparent what is the relative weight of monetary and non-monetary factors in the overall inflation rate.

  • Since supply shocks are shown in this paper to be of a transitory nature, core inflation is a convenient way of distinguishing between temporary (“supply shocks”) and persistent (“core”) components of inflation process. In this way, core inflation can be useful for short-term inflation forecasting.

3. A number of countries have computed and closely follow measures of core inflation. In the U.S., the closely watched measure is inflation without food and energy prices. Seeking a more general, and empirically more successful, alternative to this measure, researchers associated with the Cleveland Federal Reserve Bank have developed the methodology of a weighted-median inflation rate (Federal Reserve Bank of Cleveland, 2003). The Central Bank of Russia has recently started monitoring core inflation and announced an indicative target for this inflation rate (see Box 1 below).

4. The most common measure of core inflation simply excludes food and energy prices from the headline inflation rate, because most of the rapid swings in prices occur in these two sectors. However, this measure may not be suitable for a transition economy. Food items typically make up about half of the consumption basket and excluding them (together with energy) would lead to a measure of inflation that is arguably no longer representative. Excluding food prices alone is also not satisfactory, as many transition countries go through a period of gradual energy price increases that are sources of sizable inflation shocks.

5. A useful measure of core inflation is based on the median price change. Core inflation is in this case defined as a median price change in the cross-section of commodities that are contained in the CPI basket. Since core inflation thus measured represents a price change of a typical commodity, it is unlikely to be influenced by sectoral shocks. The intuition for median inflation can be acquired from Figure 1, where the distribution of price changes in December 2002 is plotted. The headline inflation rate of 3.2 percent was strongly influenced by increases in the prices of vegetables (up by 20 percent), water (up by 28 percent) and energy (up by 6 percent). After calculating a weighted-median (rather than a weighted-average) price change, it turns out that inflation would have been just 1.41 percent without these and other outliers.4

Figure 1.
Figure 1.

Histogram of Price Changes

(December 2002)

Citation: IMF Staff Country Reports 2003, 119; 10.5089/9781451805123.002.A003

Source: Ministry of statistics.

B. Core Inflation in Belarus

6. The ministry of statistics constructs one price index that approximates core inflation: inflation without rent, water, fuel and power prices. However, seasonal fluctuations and liberalization measures in the food sector would make this measure a rather imprecise measure of core inflation. For example, vegetables contributed more than one percentage point to December 2002 inflation of 3.2.

7. The staff has estimated core inflation in Belarus. The median price change was derived using detailed price data provided by the ministry of statistics from January 2000 to December 2002. Figure 2 compares the CPI inflation series with core inflation thus estimated. The core inflation series is consistent with popular stories about the development of inflation in Belarus. For example, increases of energy prices carried out at the beginning and at the end of 2002 led to only temporary increases in inflation. Both favorable and unfavorable food shocks that occur in the summer and winter seasons are removed too. Median inflation is highly correlated with inflation excluding rent, water and energy (the correlation is 75 percent in first differences), but is more sensible as it excludes sharp fluctuations caused by food and other items. Moreover, inflation excluding rent, water, and energy does not leave out occasional large changes in prices of public transportation.

Figure 2.
Figure 2.

Headline and Core Inflation in Belarus

(January 2000-December 2002)

Citation: IMF Staff Country Reports 2003, 119; 10.5089/9781451805123.002.A003

Sources: Ministry of Statistics and IMF staff estimates.

8. Monetary policy is still important. Although the measure of core inflation removes seasonal fluctuations in prices of individual commodities, it cannot (and should not) filter out any seasonal pattern in monetary policy. Both in 2001 and 2002, monetary and fiscal policies were loosened at the end of year, which led to a rise in both core and headline inflation. In this sense, core inflation is not a seasonally adjusted series.

9. Supply shocks have had a significant impact on inflation in recent years. During 2002, the authorities rapidly increased cost recovery in the communal services sector. Hence, while the annual inflation rate reached 35 percent, 12 percent could be attributed to various supply shocks. In 2003, the authorities expect that similar administrative price increases will contribute 5–6 percent to the inflation rate. It needs to be stressed, however, that at annual frequencies inflation is largely a monetary phenomenon, and the overall inflation level is determined by the monetary policy set by the NBB.5

Core Inflation in Belarus and Russia

The Russian State Statistical Office (RSSO) recently started publishing a measure of core inflation. It is defined as inflation excluding energy, administered prices, fruits and vegetables. So far only a limited number of data points are available, which prevents us from comparing effects of supply shocks on inflation in Belarus and Russia.

However, given the increasing level of cooperation between the National Bank of Belarus and the Central Bank of Russia, it is reasonable to ask whether the measure of core inflation employed in Russia would be sensible in Belarus. Figure 3 compares Belarusian headline inflation, median core inflation and a measure of core inflation that was calculated according to the methodology used in Russia.

Figure 3.
Figure 3.

Median Core Inflation and Its Alternative

Citation: IMF Staff Country Reports 2003, 119; 10.5089/9781451805123.002.A003

Sources: Ministry of Statistics and IMF staff estimates.

While the intuition behind removing volatile fruit and vegetable prices is reasonable, this alternative measure of core inflation does not, for example, smooth out shocks to prices of all the other food items. In Belarus, food excluding fruits and vegetables forms over 50 percent of the CPI basket. The difference between the two measures of core inflation was substantial during several months of 2000 and 2001. During 2002, median inflation and core inflation according to the alternative methodology were quite similar as the frequency of shocks from sectors other than fruits, vegetables and energy diminished.

Table 1.

Impact of Supply Shocks on Inflation

(2000-02)

article image
Sources: Ministry of Statistics and IMF staff estimates.

The inflation model that underlies this calculation is estimated in Section C of this paper.

10. The share of monthly inflation volatility explained by supply shocks has increased as monthly inflation rates have fallen. In November 2002, headline CPI inflation was 3.2 percent, but only 1.3 percent was core inflation. The remainder (1.84 percent) was related to hikes in gas and other utility prices for the households. Similarly, in December 2002, headline inflation was 3.2 percent but the core inflation was just 1.4 percent. Therefore, it is not unusual that much of the inflation in any given month can be attributed to “special,” non-monetary factors, and that the underlying inflation rate is much lower than suggested by the headline.

11. Does the inflation rate quickly fall back to the level of core inflation or does the supply shock persist? The question of how inflation evolves after a supply shock is important. It can be answered using econometric techniques, but Figure 2 provides some useful intuition. Consider for example the beginning of 2002. There were repeated unfavorable inflation shocks, yet the core inflation rate fell slightly at first and then remained constant. This suggests that the persistence of supply shocks is currently low and inflation quickly returns to its core level within a month or two. This conclusion is confirmed by econometric analysis below.

C. Inflation Model and Policy Implications

12. A model of Belarusian inflation is needed in order to analyze the persistence of supply shocks formally. Given the difficulties associated with estimating structural models of transition economies, inflation is modeled here by a simple reduced-form Phillips curve.1 To ensure stability of such a reduced form, it is necessary to estimate the model over a period of stable economic policy (Lucas, 1976). Therefore, data are used for the period from April 2000 to December 2002, which approximately forms a stable policy regime. During this period, monthly inflation rates stabilized below 5 percent and the crawling band exchange rate regime (launched in 2001) provided an anchor for inflation expectations.

13. The Phillips curve is estimated using monthly data on CPI inflation (Πt), year-on-year growth of industrial output (yt) and a measure of supply shocks (Shockt), which is defined as the difference between headline and core inflation rates. The variable εt denotes aggregate shocks such as monetary and wage innovations.2

πt=0.189(0.308)+0.746πt1(0.102)+0.090yt1(0.031)+0.818Shockt(0.175)0.688Shockt1(0.213)+ɛt
R¯2=0.83,s.e.=1.67,LQ(1)=0.76,LQ(4)=0.52

14. This simple model fits the data very well with a small number of regressors. The adjusted R2 is over 80 percent, slope coefficients are highly statistically significant, and residuals have standard properties. The estimated inflation model exhibits several notable features: (i) supply shocks are short-lived, (ii) inflation and output are positively correlated, and (iii) underlying inflation is highly persistent.

15. The persistence of supply shocks is limited. Since the coefficient on the lag of Shock, is nearly the opposite of the coefficient on contemporaneous Shockt supply shocks evaporate from headline inflation within a month or two. As suggested by Figure 2, in the current policy regime, supply shocks (that often represent administrative price hikes) have only short-lived effects on inflation. However, they have a highly persistent effect on the price level,3 which means that the National Bank accommodates these disturbances.

16. The other two results—that output is correlated with inflation and that underlying inflation is persistent—are standard in the business-cycle literature.4 But the finding that the Phillips-curve relationship is estimable using Belarusian data is important. The persistence of underlying inflation can be attributed to accommodative monetary policy and the influence on expectations of the crawling band exchange rate regime. Another factor has likely been a wage policy that has systematically attempted to set real wage growth over productivity growth.

17. The result on the stickiness of underlying inflation is essential (the coefficient on inflation lag is close to one). Aggregate shocks, such as to wages or money, have had highly persistent effects on the core and headline inflation rates. The intuition can again be obtained by looking at figure 2. Following the dramatic wage increase in the middle of 2001, core inflation rose for many periods and only slowly returned to lower levels. Furthermore, core inflation has been stable at 1-1.5 percent a month since May 2002 and shows no sign of decline. It may have even been increasing recently, consistent with the seasonal character of monetary policy in Belarus. Clearly, the high level of underlying inflation has been built into inflation expectations. It may thus prove costly for Belarus to postpone reducing inflation to the level of its mam trading partners.

APPENDIX I: Core Inflation: Methodology and Interpretation

Following Bryan and Cecchetti (1991), core inflation is measured as the weighted-median inflation rate. For its computation, we need detailed information on price changes in the individual components of the consumer price index. For any given month, we obtained from the ministry of statistics and analysis price changes for 50-100 commodities along with the appropriate basket weights. We first sorted the individual price changes in ascending order while keeping track of weights. The median inflation rate is at the 50th percentile of this sorted list.

The typical interpretation of core inflation is that it reflects underlying monetary inflation (Bryan and Cecchetti, 1991 and 1994). A preliminary test of whether measures of core inflation and supply shocks calculated in this paper are consistent with this proposition can be made in two steps. First, supply shocks should be independent of current money and secondly, core inflation should be uncorrelated with current supply shocks. Regression results confirmed that the series calculated in this paper satisfy this property.

Another, more recent, definition of core inflation in the literature stipulates that core inflation is the inflation rate that best predicts future headline inflation, conditional on the level of economic activity (Smith, 2003). A partial test of this proposition consists of adding core inflation into the Phillips curve as a separate regressor. It turns out that coefficients on past inflation and past supply shocks become statistically insignificant, which is consistent with this definition of core inflation.

Table 2.

Headline and Core Inflation in Belarus

(2000–02)

article image
Sources: Ministry of Statistics and IMF staff estimates.

References

  • Ball, Laurence, and N. Gregory Mankiw, 1995, “Relative-Price Changes as Aggregate Supply Shocks,Quarterly Journal of Economics, Vol. 110, No. 1, pp. 16193.

    • Search Google Scholar
    • Export Citation
  • Bryan,Michael F., and Stephen G. Cecchetti, 1994, “Measuring Core Inflation,” in Monetary Policy ed. by N. Gregory Mankiw (The University of Chicago Press), pp. 195220.

    • Search Google Scholar
    • Export Citation
  • Bryan, Michael F., and Christopher J. Pike, 1991, “Median Price Changes: An Alternative Approach to Measuring Current Monetary Inflation,” Federal Reserve Bank of Cleveland Economic Commentary (December 1).

    • Search Google Scholar
    • Export Citation
  • Federal Reserve Bank of Cleveland, 2003, “Median CPI Provides Better Measure of Core Inflation.” Available via the Internet: http://www.clev.frb.org/reseaich/daWmcpi.htm.

    • Search Google Scholar
    • Export Citation
  • Lucas, Robert E., 1976, “Econometric Policy Evaluation: A Critique,Journal of Monetary Economics, Vol. 1, pp. 1946.

  • Mankiw, N. Gregory, 2001, “The Inexorable and Mysterious Tradeoff Between Output and Inflation,Economic Journal, Vol. 111, pp. C45C61.

    • Search Google Scholar
    • Export Citation
  • Mankiw, N. Gregory, and Ricardo Reis, 2003, “Sticky Information Versus Sticky Prices: A Proposal to Replace the New Keynesian Phillips Curve,Quarterly Journal of Economics, Vol. 117, No. 4, pp. 12951328.

    • Search Google Scholar
    • Export Citation
  • Smith, Julie, 2003, “Median Inflation: Is This Core Inflation?Journal of Money, Credit and Banking, forthcoming.

1

Prepared by Martin Sommer.

2

See Ball and Mankiw (1995) for review. Ball and Mankiw prefer using term "relative-price shock" rather than "supply shock" because at the microeconomic level, a change in a price of one commodity can be caused by both supply and demand shocks. At the macroeconomic level, however, this distinction is not important.

4

Core inflation cannot be read directly from Figure 1 because the actual calculation takes account of weights in the CPI basket. See Appendix 1 for details on the methodology.

5

The NBB can in principle offset inflationary effects of any supply shocks by sticking to its predetermined monetary targets.

1

The theoretical underpinning for this approach is provided by Mankiw and Reis, 2003.

2

Standard errors of estimated coefficients are reported in parentheses. LQ(1) and LQ(4) are probability values of the Ljung-Box test statistics at the first and fourth lags, respectively.

3

This can easily be seen when we approximate the estimated model as: εt=-εt-i + βyt-i + Shockt - Shocks + εt. Inflation returns to its pre-shock level already after one period but the price level stays permanently higher. Estimates in Table 1 are based on this approximating model.

4

See Mankiw (2001) for a review.

Republic of Belarus: Selected Issues
Author: International Monetary Fund