Republic of Belarus: Selected Issues

The analysis of inflation developments in Belarus is hampered by widespread price controls. Persistence of common inflation is generally higher than that of actual inflation. Factor analysis assumes that covariation among time series can be explained by a few unobserved shocks (factors). The dependent variable in our estimations is growth in real GDP per capita in purchasing power parity (PPP) terms, while the explanatory variable of interest is the annual rate of change in the terms of trade.

Abstract

The analysis of inflation developments in Belarus is hampered by widespread price controls. Persistence of common inflation is generally higher than that of actual inflation. Factor analysis assumes that covariation among time series can be explained by a few unobserved shocks (factors). The dependent variable in our estimations is growth in real GDP per capita in purchasing power parity (PPP) terms, while the explanatory variable of interest is the annual rate of change in the terms of trade.

I. Common and Idiosyncratic Determinants of Inflation1

A. Introduction

1. Inflation has fallen across many Eastern European and transition countries over the past few years (Figure 1). While starting from different levels, by end-2005, seasonally-adjusted monthly inflation had dropped on average by half in countries such as Belarus, Czech Republic, Estonia, Hungary, Poland, Romania, Russia, and the Slovak Republic. By contrast, inflation remains higher in Bulgaria, Latvia, Lithuania, and Ukraine when compared to 2001, with an average increase in seasonally-adjusted monthly inflation rates of about 65 percent. With regard to Belarus, a large output gap has allowed strong noninflationary growth; however, pervasive price controls are likely to have played a role in explaining inflation developments.

Figure 1.
Figure 1.

Inflation in 12 Eastern European and Transition Countries, 2002-05

(Year-on-year change, in percent)

Citation: IMF Staff Country Reports 2006, 315; 10.5089/9781451805253.002.A001

Source: Eurostat, national authorities; and Fund staff estimates.

2. Several factors can explain inflation over time. Money growth is uncontroversially one of these factors. In transition economies, it fueled inflation early on when fiscal obligations were monetized and lack of a credible fiscal stance contributed to the deterioration of market confidence, thus increasing velocity. Wage growth, beyond productivity gains, impacts prices not only directly by increasing costs, but also indirectly by raising domestic demand. This is particularly relevant in countries in which the share of wages in household disposable income is relatively high, and household expenditure is biased toward basic items, which usually comprise the largest component in the CPI basket. The output gap would affect the likelihood that bottlenecks put upward pressures on prices in specific sectors. Real exchange rate appreciation, owing to Balassa-Samuelson effects and/or simply to surges in domestic absorption stemming from higher levels of income, would also trigger inflation in cases in which these pressures are not accommodated through nominal appreciation. Inflationary pressures could also result from relative price adjustments with downward price rigidity while structural reforms are being implemented and both supply and demand adjust during transition.

3. Some factors are common across countries or sectors. These factors are unobserved shocks that drive the underlying inflation process and are common (correlated) across countries or sectors, although their impact depends on their individual “load” and differences in economic structures and policies. In countries in which significant economic convergence has been achieved, a limited number of similar forces are likely to drive inflation at a certain point in time: an oil shock could be an example of the same exogenous force that affects countries/sectors. Some factors are country- or sector-specific. At any point in time, inflation may be driven by shocks that impact a single country—for instance, a correction in energy prices relative to international market prices—and by policies—monetary, fiscal and exchange rate policies—and/or conditions—price controls, relative price adjustments, degree of competition—that are specific to each country or sector. These idiosyncratic factors, by definition, are uncorrelated with common factors.

4. The analysis of inflation developments in Belarus is hampered by widespread price controls. The share of goods and services subject to direct price control reached over 35 percent in 2005, and included food items, and communal and transportation services. In 2005, the Council of Ministers issued a decree that set monthly marginal price increases (0.6–0.8 percent) for all goods and services produced and sold in Belarus, with very limited exceptions. All enterprises seeking higher than marginal price increases need to apply to the local authorities, who may reject the application. Violation of the price registration mechanism may result in fines, sanctions, and business closure. In addition, the government continues to limit profit margins on socially important goods and the majority of food articles in the consumer basket. This would clearly be a feature that is specific to Belarus. In these circumstances, actual inflation may not fully reflect underlying inflationary pressures. Therefore, distinguishing among common and idiosyncratic determinants of inflation appears a promising starting point for analyzing the inflation process.

The chapter is organized as follows. Section B briefly presents the data and the generalized dynamic factor model methodology. Section C reports the results. Section D focuses on inflation forecasts for Belarus, while Section E concludes.

B. Methodology and Data

5. Factor analysis assumes that covariation among time series can be explained by a few unobserved shocks (factors). In factors models, therefore, a large number of covarying series are transformed into a smaller number of unobserved orthogonal series (common components) so as that each additional factor (component) explains as much as possible of the remaining variation in the observed series. The observed series is then represented as the sum of the common component, which can be interpreted as underlying inflation, and of a disturbance term (idiosyncratic component), which is uncorrelated with the common component.

6. The analysis in this chapter is based on an application of the generalized dynamic factor model (GDFM) proposed by Forni and others (2000 and 2001).2 This is a statistical approach that extends principal component analysis and Stock and Watson’s (1989) coincident and leading indicator approach. The basic framework is that of a dynamic factor model in which the assumption of mutually orthogonal idiosyncratic components is relaxed to allow for some mild cross correlation. Underlying inflation is therefore assumed to be proxied by a common component, which is driven by a small number of common factors. These factors are the same across the countries, although potentially impacting inflation differently in each country (different coefficients or “loads”).

7. The dataset comprises a panel of 12 countries, and 223 monthly series of CPI indices and their components over the period 2001–05. Factor models can accommodate large panels and overcome the problem inherent in multivariate analysis when the time dimension is smaller than the cross-country dimension. The data set contains seasonally adjusted monthly inflation from January 2001 through December 2005, both for headline CPI inflation and for its components, for 12 Eastern European and transition countries, with over 13,000 data points.3 The sources are the Harmonized Index of Consumer Prices (HICP) and national statistics. All 223 series were tested for unit roots; 15 CPI components in various countries turned out to be nonstationary and were dropped from the dataset for the estimation.

8. The first step in the analysis is to determine the number of common factors. A principal component analysis of the spectral density matrices of the data (Figure 2) shows the share of the cumulative variance (cumulative eigenvalues) of the series that is explained by each successive principal components (eigenvector). Different thresholds can be set to identify the number of common factors (components): here, this is chosen by stopping at the factor (eigenvalue) that improves upon the explained cumulative data variability by less than 10 percent at all frequencies. This yields four dynamic common components, which explain bout 75 percent of the total data variability.

Figure 2.
Figure 2.

Cumulative Data Variability Explained by the First Ten Common Factors

(In percent)

Citation: IMF Staff Country Reports 2006, 315; 10.5089/9781451805253.002.A001

Source: Fund staff estimates

9. The next step is to determine the number of static factors. The relation among static and common factors, and lags is given by:

number of static factors=number of common factors * (1+number of lags).4

With 2 common factors and 12 as the number of lags (in light of the monthly frequency), the number of static factors are set at 26.

C. Generalized Dynamic Factor Model Results

10. Applying the GDFM to our dataset decomposes inflation in the 12 countries in the sample into common and idiosyncratic components.5 Figure 3 plots headline CPI common (underlying) and actual inflation for each country. The common component of each country’s inflation—that is, that part of inflation that is explained by shocks that are shared across countries and sectors—tracks movements in headline inflation while smoothing it by eliminating cross-section and cross-country disturbances. Common inflation explain over 35 percent of the variability of actual inflation for the whole panel (Table 1). Across countries, however, common components tend to account for a somewhat larger share of actual inflation variability in Belarus, Russia, and Ukraine (Group A), indicating that idiosyncratic shocks are relatively less important in these countries.

Figure 3.
Figure 3.
Figure 3.

Measures of Headline CPI Inflation, 2001-05

(Monthly seasonally-adjusted, in percent)

Citation: IMF Staff Country Reports 2006, 315; 10.5089/9781451805253.002.A001

Source: Eurostat, national authorities; and Fund staff estimates._____________ common components------------------ actual
Table 1.

Inflation Variance

article image
Source: Eurostat, national authorities; and Fund staff calculations.

11. Persistence of common inflation is generally higher than that of actual inflation (Table 2). Persistence is proxied by the half life of a unit shock, which indicates the length of time necessary to halve the magnitude of the original shock to inflation. It is calculated as:

Halflife=|log12logβ|

where, β is derived by estimating a simple regression of the monthly seasonally-adjusted headline CPI inflation on its lag and a constant: πt = α + βπt−1 + εt. The fact that common inflation shows higher persistence suggests that relative price adjustments, Balassa-Samuelson effects, and, more generally, structural transformation may not have fully run their course.6

Table 2.

Inflation Persistence: Half Life

article image
Source: Eurostat, national authorities; and Fund staff calculations.

12. Table 3 and Figure 4 report the difference between common and actual inflation in Belarus. Several stylized facts are worth mentioning. First, while on average the difference between common and actual inflation is zero over the sample period, this is positive in 2005 and larger than in any of the previous years—the ratio between common and actual inflation shows a similar picture. This is true both for the whole CPI index and for the single CPI components. Second, as it is likely that the difference will revert to its zero mean (and to a ratio of one), actual inflation may pick up. Third, the CPI components that are the most above their past averages are those, such as housing, clothing, and alcoholic beverages, that are subject to extensive price controls.

Table 3.

Belarus: Difference between Common and Actual Inflation, 2001-05

article image
Source: Eurostat, national authorities; and Fund staff calculations.
Figure 4.
Figure 4.
Figure 4.
Figure 4.

Belarus: Measures of Inflation for CPI Components, 2001-05

(Monthly seasonally adjusted, in percent)

Citation: IMF Staff Country Reports 2006, 315; 10.5089/9781451805253.002.A001

Source: Eurostat, national authorities; and Fund staff estimates.____________ common components---------------- actual

D. Forecast

13. The Bai and Ng’s (2001) algorithm is used to determine the optimal number of static factors. The algorithm is maximized at around 60 static factors. For the estimation, the number of static factors is set at 52, resulting from 4 common factors and 12 lags.

14. An inflation forecast would need to incorporate a projection for both underlying (common component) inflation and idiosyncratic inflation. Underlying inflation in Belarus is predicted by using the one-sided predictor proposed by Forni and others (2003). The idiosyncratic component, however, may have an important impact on inflation in the short term. A forecast of the latter is projected both by estimating a classic Box-Jenkins ARIMA model and by applying the same common component analysis to the idiosyncratic component of inflation.7 Figure 5 shows monthly seasonally-adjusted inflation forecasts of both underlying inflation and headline inflation in Belarus, based on the two approaches to forecast the idiosyncratic component. To note is that headline and underlying inflation forecasts—when the forecast of the idiosyncratic component is predicted on the basis of the same framework that is used to predict underlying inflation—tend to increasingly overlap as the forecast horizon increases. This is consistent with the idea that the idiosyncratic component of inflation picks up the short-term impact of specific policy actions.

Figure 5.
Figure 5.

Belarus: Actual Inflation and Inflation Forecasts, 2001-06

(Monthly change, seasonally adjusted)

Citation: IMF Staff Country Reports 2006, 315; 10.5089/9781451805253.002.A001

Source: National authorities; and Fund staff estimates.

E. Conclusions and Areas of Further Analysis

15. Based on the analysis of common and idiosyncratic components of inflation in Belarus, several preliminary conclusions are possible. First, inflationary pressures appear to have mounted in 2005. In fact, underlying inflation appears to be above actual inflation by more than at any time over the last five years. At the same time, price controls may now play a bigger role in masking these inflationary pressures.

16. From an analytical standpoint, it would be interesting to compare underlying inflation, as derived in this paper, with other concepts of underlying inflation such as “core” inflation, trimmed mean, and the median. All of these are useful indicators of inflationary pressures that could inform the NBRB’s monetary policy stance.

17. More generally, it would be interesting to regress underlying inflation, as derived in this paper, on demand and supply side variables and test their explanatory power. For instance, Chernookiy (2004) could be re-estimated with the common component series derived here.

18. Finally, while common components explain roughly a similar proportion of inflation variability across countries in the panel, the levels of common inflation across countries differ. This could reflect the way (load) common factors impact inflation in each country, which ultimately depends on the structure of the economy, including the exchange rate regime, production technologies, relative prices, and other catching up issues. To assess these channels, a series of fixed-effects panel regressions could be performed, which would regress common inflation differentials on a series of explanatory variables that proxies the structural and policy framework.

References

  • Bai, J. and S. Ng, 2000, “Determining the Number of Factors in Approximate Factor Models,mimeo.

  • Chernookiy, V., 2004, “Model of Inflation Processes in the Republic of BelarusNational Bank of the Republic of Belarus Research Papers, No. 1/2004.

    • Search Google Scholar
    • Export Citation
  • Coorey, Sharmini, Mauro Mecagni, and Erik Offerdal, 1996, “Disinflation in Transition Economies: The Role of Relative Price Adjustment,IMF Working Papers, WP/96/138 (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation
  • Forni, M., and others, 2000, “The Generalized Factor Model: Identification and Estimation,Review of Economics and Statistics, Vol. 82, No. 4, pp. 54054.

    • Search Google Scholar
    • Export Citation
  • Forni, M., and others, 2003, “The Generalized Dynamic Factor Model One-sided Estimation and Forecasting,mimeo.

  • Nadal De Simone, F., 2005Recent French Inflation Behavior: Is It Any Different from the Euro Area’s?in France: Selected Issues, Country Report No. 05/397 (Washington: International Monetary Fund), pp. 421.

    • Search Google Scholar
    • Export Citation
  • Rossi, M., 2005, “Inflation Persistence: Is There a Role for Relative Prices?in Ukraine: Selected Issues, Country Report No. 05/416 (Washington: International Monetary Fund pp. 5568.

    • Search Google Scholar
    • Export Citation
  • van Elkan, R., and others, 2006, “A Factor Analysis of EU Inflation: Implications for New Members’ Euro Adoption Prospects,Washington: International Monetary Fund, mimeo.

    • Search Google Scholar
    • Export Citation
1

Prepared by Marco Rossi.

2

Applications of Forni and others (2000 and 2001) can be found in Nadal de Simone (2005) and van Elkan and others (2006).

3

The countries are: Belarus, Bulgaria, Czech Republic, Estonia, Hungary, Latvia, Lithuania, Poland, Romania, Russia, Slovak Republic, and Ukraine. Seasonal adjustment is done with X-11, which may not be the best seasonal adjustment for Belarus data.

4

See Forni and others (2003) for a definition of the relationship.

5

Nadal de Simone kindly shared the Matlab code used in Nadal de Simone (2005).

6

See Coorey and others (1996) for a discussion of the role of relative price adjustments.

7

The ARIMA specification includes a constant and the dependent variable (idiosyncratic inflation) with 1 and 12 lags.

Appendix I.

Table A1.

Countries in the Sample

article image
Table A2.

Data Sources and Definitions of the Variables Used in the Econometric Analysis

article image

References

  • Barro, R., 1997, Determinants of Economic Growth: A Cross-Country Empirical Study, (Cambridge, Massachusetts: MIT Press).

  • Basu P. and D. McLeod, 1992, “Terms of Trade Fluctuations and Economic Growth in Developing Economies,Journal of Development Economics, Vol. 37, pp. 89110.

    • Search Google Scholar
    • Export Citation
  • Blattman C., J. Hwang and J. Williamson, 2006, “Winners and Losers in the Commodity Lottery: The impact of Terms of Trade growth and Volatility in the Periphery 1870–1939,forthcoming in Journal of Development Economics.

    • Search Google Scholar
    • Export Citation
  • Broda, C., 2004, “Terms of Trade and Exchange Rate Regimes in Developing Countries,Journal of International Economics, Vol. 63, pp. 3158.

    • Search Google Scholar
    • Export Citation
  • Easterly W., M. Kremer, L. Pritchett and L. Summers, 1993, “Good Policy or Good Luck? Country Growth Performance and Temporary Shocks,Journal of Monetary Economics, Vol. 32, pp. 45983.

    • Search Google Scholar
    • Export Citation
  • Fischer S., 1993, “The Role of Macroeconomic Factors in Growth,Journal of Monetary Economics, Vol. 32, pp. 485512.

  • Kaufmann, D., A. Kraay and M. Mastruzzi, 2005, “Governance Matters IV: Governance Indicators for 1996–2004, World Bank Working Paper (Washington: World Bank).

    • Search Google Scholar
    • Export Citation
  • Mendoza, E., 1997, “Terms-of-Trade Uncertainty and Economic Growth,Journal of Development Economics, Vol. 54, pp. 32356.

  • Newey, W. and K. West, 1987, “A Simple, Positive Semi-Definite, Heteroskedasticity and Autocorrelation Consistent Covariance Matrix,Econometrica, Vol. 55, 3, pp. 70308.

    • Search Google Scholar
    • Export Citation
  • Sala-i-Martin, X., G. Doppelhofer, R. Miller, 2004, “Determinants of Long-Term Growth: A Bayesian Averaging of Classical Estimates (BACE) Approach,American Economic Review, Vol. 94, 4, pp. 81335.

    • Search Google Scholar
    • Export Citation
  • Schadler, S., A. Mody, A. Abiad and D. Leigh, 2006, “Growth in the Central and Eastern European Countries of the European Union: A Regional Overview”, forthcoming IMF Occasional Paper (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation
  • Turnovsky S. and P. Chattopadhyay, 2003, “Volatility and Growth in Developing Economies: Some Numerical Results and Empirical Evidence,Journal of International Economics, Vol. 59, pp. 26795.

    • Search Google Scholar
    • Export Citation
  • World Bank, 2005, “Window of Opportunity to Enhance Competitiveness and Sustain Economic Growth”, Country Economic Memorandum (Washington: World Bank).

    • Search Google Scholar
    • Export Citation
8

Prepared by Kristian Hartelius

9

A Hausman specification test, however, favors fixed effects.

10

The coefficients on the interaction terms in this paper show how the growth impact of changes in the terms of trade depends on other variables. The coefficient reported for “TOT change” is conditional on the variables included in the interaction terms being equal to zero, and therefore varies with the specification.

11

In estimating the standard errors, the lag length is set to a maximum of five years.

12

The negative coefficient on “TOT change” is for a notional country that is totally closed and outside of the ECA region (see footnote 3). Let TOT change = X1. The country-specific terms of trade elasticity can be calculated as -0.067*X1 + 0.08*X1*ECA + 0.0013*X1*openness.

13

A regression (not reported) including both the ECA and the CIS dummies further indicates that the results for the ECA region to a large extent are driven by the CIS countries excluding Russia.

14

As a proxy in the regressions, we use the “Regulatory Quality” indicator developed by Kaufmann and others (2005), which measures the incidence of market-unfriendly policies, such as price controls or inadequate bank supervision, as well as perceptions of the burdens imposed by excessive regulation in areas such as foreign trade and business development.

Republic of Belarus: Selected Issues
Author: International Monetary Fund