Abstract

The overall price level increased sharply in transition countries once prices were freed. Disinflation has most fre—uently been gradual, with prices continuing to rise rapidly in subseuent years. This paper identifies the well- and lesser-known features of inflationary processes in central and eastern Europe, the Baltic countries, Russia, and other countries of the former Soviet Union based on a sample of 26 countries and observations spanning the first five to seven years of transition. [JEL: E31, P22, R32]

For several decades, most prices in the centrally planned economies were set administratively, with little regard for cost or demand. Exceptions included farmers’ markets, where various foodstuffs were sold,1 and more or less clandestine parallel markets for other goods and services. Nevertheless, prices were on the whole subject to strict controls. As a result, measured inflation was most of the time low and repressed.2

After these countries embarked in earnest on the transition to a market economy, one of their first steps was to liberalize a large proportion of producer and consumer prices. The liberalization sequence typically started with the freeing of the prices of many goods and some services, but at the consumer level the prices of staples and many other services, particularly housing and utilities, often remained controlled for several years into transition and were adjusted only infrequently.

In most of these countries, price liberalization was followed by lasting, high, open inflation, degenerating in some cases into hyperinflation. Experiences varied tremendously across countries, however. The cumulative increase in the level of consumer prices during the first five years of the new price regime, including the price jump associated with the initial comprehensive liberalization, amounted to 138 percent in the Czech Republic, 1,341 percent in Poland (i.e., a 14.4-fold rise), 80 times in Latvia, close to 2,000 times in Russia, over 18,000 times in Kazakhstan, and over 86,000 times in Georgia (Figure 1).3 In Serbia-Montenegro, the price level skyrocketed, increasing over 78,000,000,000,000,000,000,000 times in 25 months (January 1992 to February 1994), as the country witnessed one of the worst hyperinflations in world history (Bogetic and others, 1996). Nonetheless, by the second half of 1996, the pace of consumer price inflation had come down very significantly in most of the 26 countries under consideration, averaging less than 1 percent a month in 15 of them.4

Figure 1.
Figure 1.

Cumulative Increase in Aggregate Consumer Price Index During the First Five Years of Transition1

(Logarithmic scale; in number of times)

Sources: Tables Al and A2.1 Base dates are as follows: December 1989 for Croatia, Hungary, the former Yugoslav Republic of Macedonia, Poland, the Slovak Republic, and Slovenia; October 1990 for Romania; December 1990 for Bulgaria, the Czech Republic, Estonia, and Latvia; January 1991 for Albania and Lithuania; December 1991 for Russia and other countries of the former Soviet Union.

One lesson from this wide dispersion of outcomes is that liberalization as such cannot be viewed as the cause of the subsequent high inflation rates. The latter are primarily driven by the fiscal travails of the transition (and, in several cases, of war). Of course, they also reflect the process of relative price convergence: tradables prices have tended to increase rapidly toward international levels after price liberalization, even though nontradable prices have remained far below such levels. To some extent, measured inflation after the achievement of convergence (or near convergence) of the price of tradables reflects a catch-up of the price of nontradables.

This study draws on a broad sample of countries to bring out some of the key stylized facts pertaining to absolute and relative consumer price movements in the course of transition. As such, its objective is to serve as a concise reference manual for transition country inflation watchers. The sample size varies somewhat across the different lines of inquiry owing to the inaccessibility of some of the data for some of the countries but is large enough in most cases to allow fairly solid generalizations.5

Ten Stylized Facts

Fact 1. Price Liberalization Translated into a Surge in the Overall Price Level

In the late 1980s to the early 1990s, the fiscal pressures that built up in most of the countries under consideration would have produced high inflation but for the broad-based price controls then in place. Where more than a marginal fraction of prices were free to adjust, those pressures resulted in a marked pickup in inflation (most notably, in Poland). In fact, even in the former Soviet Union, there was a detectable acceleration in prices. Already before the conventional D days of transition (e.g., January 1, 1990 in Poland and January 2, 1992 in Russia), some important decontrol measures were taken in several major countries, causing a first series of large price increases. In most countries, a more comprehensive liberalization followed shortly thereafter, often characterized as the “Big Bang” inaugurating transition. Although such a breakpoint can be identified readily in most countries, the boldness and scope of the approach varied a lot from one to another. Indeed, some countries preferred a more piecemeal liberalization (e.g., Romania versus Poland, or Kazakhstan versus Russia). And in all countries, some prices remained administered well into transition, not least the prices of some key services.

The surge in the overall price level, as measured by the consumer price index (CPI), associated with comprehensive price liberalization was typically several orders of magnitude larger than prior and subsequent inflation rates, reflecting an abrupt, one-time adjustment. In 24 of the 26 countries, the largest single monthly increase in the CPI (marked in bold in Tables Al and A2) was indeed recorded at that time. Armenia and Serbia-Montenegro subsequently experienced larger increases, but in the context of hyperinflation.

Fact 2. The Initial Burst Was Followed by a Prolonged Period of Relatively High Inflation

Even in countries with an established tradition of financial prudence (e.g., the Czech and Slovak Republics) or in countries pursuing consistently tight financial policies (e.g., Albania), inflation did not instantly settle at low levels following price liberalization. In those in which a looser stance prevailed, be it as a result of economic policy failure or military conflict, inflation remained high and in some cases degenerated into bouts of hyperinflation (Ukraine in 1993, Armenia in 1994, Georgia in 1993, Tajikistan in 1995, and Bulgaria in late 1996).6 The sources of inflation persistence have included money creation and wage pressures (Coorey and others, 1996); indexation of nominal incomes (Pujol and Griffiths, 1996), (Czyzewski and others, 1996); and relative price adjustments (De Masi and Koen, 1996), (De Broeck and others, 1997), (International Monetary Fund, 1996a, 1996b, and 1996c), and (Coorey and others, 1996).

Fact 3. At Higher Frequencies, Seasonality and Administrative Intervention Have Caused Potentially Misleading Fluctuations Around “Core” (Dis)inflation (Pseudo-Turning Points)

Inflation in transition countries has typically been rather volatile, complicating the identification of inflation trends. In part, volatility at high frequencies (monthly observations) reflects changes in the stance of financial policy or other fundamentals. As such, it corresponds to an unstable underlying, or “core,” rate of inflation. However, volatility also results from two types of seasonal variations, which could respectively be described as “natural” (weather-induced) and “artificial” (administered price-related).

The first type of seasonal variations is more important in transition economies than in advanced economies because seasonal food prices (particularly for fruits and vegetables) represent a larger share of the CPI basket, exceeding 40 percent in 1993–94 in 16 of the 21 countries for which information was available, against less than 20 percent in advanced economies. Natural seasonality proved strong enough (and core inflation low enough) in a number of countries for the aggregate price level to have declined in the summer months one or sometimes several years in a row.7 In 1996, this phenomenon was observed in half of the 26 countries.

Although administrative price adjustments are not always spaced at regular intervals, first-of-month, first-of-quarter, and first-of-year increases in those prices are systematic enough to be described as a second source of seasonality. Hence, in many countries, an inflationary spike materializes at the beginning of each year, which should not ipso facto be viewed as a turning point in inflation. These administrative price adjustments are typically very large. For instance, in April 1995, consumer prices in the Czech Republic rose by 1 percent, but rail fares by 66 percent, postal charges by 40 percent, and telecommunications charges by 20 percent. To the extent that the increases in administered prices represent permanent increases in relative prices rather than a catch-up on continuously evolving market-determined prices, they are destined to become less prominent over the years, as cost-recovery ratios rise (see Fact 5). It could even be argued that the only genuine seasonal component of those price changes is the one corresponding to the short-run catch-up effect.

Recognizing the magnitude of seasonal price fluctuations and the resulting potential for confusion, the statistical authorities have started to publish seasonally adjusted CPI series in several countries (Hungary and the Czech Republic, for instance), or are considering doing so (Russia). In the absence of a seasonally adjusted index, the emphasis is often placed on the 12-month rate of change in the CPI. Seasonality is indeed eliminated in that way (except to the extent that seasonality itself is not invariant over time), but the 12-month rate of inflation is a lagged measure that only gradually captures inflections in trend. When the latter are rather abrupt or frequent, this is a major drawback. Alternatively, therefore, it may be useful to monitor movements in the price of nonfood goods, which is generally the least volatile component of the aggregate CPI, bearing in mind, however, that the medium-run increase in the relative price of services implies that this component tends to understate overall inflation.

Fact 4. Prices of Goods Moved Rapidly Toward International Levels, with the Possible Exception of Certain Staples

The phasing out of price setting and the introduction of a market-determined exchange rate allowed the forces of international arbitrage to drive the prices of tradable goods toward those prevailing on world markets. Full convergence could, of course, not be expected in the short run since the prices of tradable goods embody a nontradable domestic distribution component, and are further influenced by taxes and subsidies (including protectionist regulations). Also, the forces of competition may not be sufficiently strong to enforce price equalizationquickly. The extent of convergence is also difficult to measure precisely, not least because ofquality differences that cannot be fully taken into account (Franz, 1996).

Even so, the extensive evidence gathered in the context of the 1993 round of the European Comparison Program (ECP) and the follow-up study on member countries of the Commonwealth of Independent States (Organization for Economic Cooperation and Development, 1996) points to a fair degree of convergence of goods prices (Tables 1 and 2).8

Table 1.

Price Level Comparison with Austria, 1993

(In percent of price in Austria)

article image
Source: European Comparison Program database, 1993.

Including beverages and tobacco.

Table 2.

Price Level Comparison with Turkey, 19951

(In percent of price in Turkey)

article image
Sources: Organization for Economic Cooperation and Development (1996), and Main Economic Indicators, various issues.

In 1995, the overall price level in Turkey was 34 percent of that in Austria.

Including beverages and tobacco.

For some highly tradable goods, prices in a number of transition countries were almost as high as in Austria (used in this paper as a benchmark advanced economy) by 1993 (e.g., gasoline in Hungary and the Czech Republic) or even higher (color television sets in most countries shown in Table 1). In Russia, the prices of many consumer durables had risen above U.S. levels by 1995 (Goskomstat, 1996a), reflecting a combination of factors, including inefficiencies in distribution enduring partly because of the ability of incumbent networks to discourage the entry of potential competitors, transportation costs, and taxation. In contrast, the prices of some staples remained much lower in transition countries, for example, bread in Albania (McNeilly and Schiesser-Gachnang, 1996), or many basic food items in Russia, the prices of which are not controlled at the federal level but are subject to regulations locally (see also Fact 9).9 Prices also typically stood far below advanced economy levels for services (rents, electricity, railway fares, and thequintessential nontradable, haircuts).10

Fact 5. Service Prices First Lagged but Then Started to Catch Up with the Prices of Goods, as Services Became Increasingly Commercialized

Many services were provided under central planning for a nominal fee, or none at all. When price liberalization was undertaken in earnest, in most countries a number of services were on the list of exceptions, notably housing, utilities, health care, and transportation. Their prices were adjusted significantly as goods prices were set free but often by less than the increases in the latter. The discrete nature of administrative price changes and their political costs caused the relative prices of services in relation to goods to decline for some time. However, this trend was typically reversed within a year or two as the relative price of services started to rise with a vengeance. The resulting J-pattern was observed in 17 of the 23 countries for which data were gathered (Figure 2).11 The most conspicuous outliers were countries where market-oriented reforms were extremely gradual (Tajikistan, Turkmenistan) or partly reversed (Bulgaria). A more intriguing outlier is the Slovak Republic, possibly reflecting a reluctance to adjust rents, or a less distorted price structure at the onset of transition.12

Figure 2.
Figure 2.

Price of Services Relative to Overall Price Level1

(Year averages unless otherwise noted)

Sources: National authorities; and Statistical Committee of the Commonwealth of Independent States.1 End-year observations for Belarus, Bulgaria, Kazakhstan, Lithuania, Russia, and the Slovak Republic. Base (100 and year 0) = 1988 for Poland; 1989 for Croatia, the Czech Republic, Hungary, the Slovak Republic, and Slovenia; 1990 for the Baltic countries, Russia, and other countries of the former Soviet Union; May 1990 for Bulgaria; and October 1990 for Romania.

Services largely coincide with nontradables, and one interpretation of the J-pattern runs along the lines of the Balassa-Samuelson model and has been put forward for the Baltic countries by Richards and Tersman (1996). Slower trend productivity growth in the nontradables sector translates into a gradual increase in the relative price of nontradables. This relative price shift will be all the stronger as many services, which were traditionally undersupplied, see their demand surge under market conditions (e.g., distribution, finance and insurance, and tourism). Some have speculated, however, that productivity may not rise faster in the production of tradables than in that of services (see Halpern and Wyplosz, 1995).

An alternative interpretation, dubbed the costrecovery hypothesis, distinguishes between capital-intensive and other services (Zavoico, 1995). The former prominently include housing, utilities, and transportation, the generation of which is based on a large capital stock inherited free of debt from the pre transition era. In the early stages of transition, the prices of those services were typically set even below maintenance costs, partly as an element of the social safety net. Over time, however, these prices are being raised significantly, with a view to financing the upkeep and replacement of the associated capital stock.

Fact 6. As the Structure of Relative Prices Has Become More Fully Market Determined, It Has Moved Closer to That Prevailing in Advanced Market Economies

The structure of relative prices within each transition country has clearly changed over time as is shown by the evolution of cross-correlations with the baseyear structure of prices for foodstuffs, nonfood goods, or services (Table 3).13 In most cases, these movements have been gradual and monotonic in the sense that relative prices have diverged more and more from their initial structures.14 An outlier in this respect is Georgia, where the 1993 hyperinflation may have randomized the relative price structure. The magnitude of the changes in relative price structures as measured by the evolution of correlation coefficients over time is apparent when the latter are compared with their analog in the United States (see the memorandum item in Table 3). The gradual nature of this process is evident from the persistence of significant dispersion in itemspecific inflation rates half a decade into transition, as documented, among other countries, for Russia (Organization for Economic Cooperation and Development, 1997). Nevertheless, relative price variability has on the whole tended to decline over time.15

Table 3.

Evolution of Domestic Price Structures1

(Cross-correlation with base year)

article image
Sources: National authorities; and authors’ calculations.

July observations unless noted.

January observations.

October for 1990, July for subsequent years.

December observations.

The shift away from base-year patterns has brought relative prices closer into line with those prevailing in advanced economies. At an aggregate level, that is clear from the increase in the relative price of services. At a more disaggregated level, this has been confirmed for the Baltic countries, for instance, by the finding of a significant positive correlation between the changes in the 20 or so main components of the CPI between 1993 and 1996 and the initial 1993 price gap vis-a-vis Austria for each of those components (International Monetary Fund, 1996a, 1996b, and 1996c). Even in the most advanced transition countries, however, the structure of relative prices remains quite different from that in neighboring market economies: in Poland for example, relative prices had come much closer to German levels in 1990 than in 1987 (Berg, 1994) but they still were far from having converged by 1996 (Figure 3). This observation is consistent with the fact that price structures worldwide are correlated with purchasing-power-parityadjusted incomes per capita (Nuxoll, 1996).

Figure 3.
Figure 3.

Price of Selected Goods and Services in Poland, 1996

(In percent of the price in Germany)

Source: National Bank of Poland (1996).

Looking at relative prices within broad categories, it appears that the structure of food prices had by 1993 come much closer to market economy levels than that of services (Table 4). More extensive realignment was therefore to be expected in subsequent years among service prices than among food prices, as Table 3 confirms.

Table 4.

Cross-Correlation of Price Structures Against Austria’s, 1993

article image
Sources: European Comparison Program database, 1993; and authors’ calculations.

Fact 7. Relative Prices Have Been Gradually Converging Across Transition Countries

As a corollary to Fact 6, the structure of relative prices has tended to become more similar across transition countries than in the early 1990s, when progress with liberalization varied more across countries.16

The evidence supporting this contention is indirect in the case of central and eastern Europe, as it is based on the common movement toward a market economy price structure rather than on bilateral comparisons between transition countries over time.

For the Baltic countries, Russia, and other countries of the former Soviet Union, however, more direct, albeit fragmentary, evidence is available. The dispersion of food prices across capital cities of these countries fell sharply between mid-1994 and mid-1996: the average of the coefficients of dispersion of the prices of 16 food items across 10 of these cities dropped by almost half between those two dates.17 A similar trend is observed for nonfood goods prices, but the monitored sample is too small for this finding to be statistically significant.

Fact 8. Wide Price Level Disparities (in Common Currency Terms) Remain and May Be Expected to Persist Across Transition Countries

Paralleling the process of alignment of relative prices, overall consumer price levels have tended to come closer together as well across transition countries (compare the dispersion in Tables 1 and 2 with that in Figure 4).18 The ranking of transition countries by price levels was somewhat unstable in the first phases of transition owing to asynchronous price liberalization and exchange rate unification measures. More recently, however, it may have started to settle closer to some equilibrium level, as market forces are becoming more generally operative. Nevertheless, substantial differences in price levels remain (Figure 4), and are likely to persist, reflecting disparate development levels (see also Fact 10). By 1996, Slovenia and Croatia clearly stood out with price levels reaching almost twice the median. Prices were well below the median in some of the countries of the former Soviet Union with the exception of Russia and Georgia. Poland, Hungary, and the Baltic countries stood about midway between the top two and the Czech and Slovak Republics, the latter representing the median.

Figure 4.
Figure 4.

Overall Price Level Gaps, 1996

(In percent)

Sources: Tables 1 and 2; IMF, International Financial Statistics, and World Economic Outlook database; and authors’ computations.

Fact 9. Over Time, Prices of Similar Goods and Services and Inflation Rates Have Converged Across Regions Within Countries

Just as prices have come closer together across countries, they have also converged across regions within countries. Market integration has a domestic as well as an international dimension. This phenomenon has been documented in some of the larger countries, where physical and institutional obstacles impeded market integration more conspicuously in the aftermath of initial price liberalizations.

In Russia, for example, although most consumer prices were decontrolled at the federal level in early 1992, local restrictions remained for some months or even years in a number of regions (Koen and Phillips, 1993). They long persisted in some of the regions endowed with the natural or other resources making the associated subsidies fiscally sustainable, most notoriously Ulyanovsk (Lenin’s birthplace) until 1996.19 Nevertheless, some signs of convergence of price levels have been detected, with geographical price dispersion coming closer to the level observed in large advanced economies such as Canada (De Masi and Koen, 1996). Other evidence pointing in the same direction includes time-series analysis of the prices of food products in 25 cities of the Volga and Central regions, which shows that despite the obstacles posed by recalcitrant local governments, mafia activity, and poor infrastructure, prices and inflation rates have tended to converge across and within cities (Berkowitz and others, 1996). However, significant regional price level disparities are bound to persist even in the absence of any regulatory constraints, owing to transportation costs, which given Russia’s climate and the size of the country represent a substantial portion of the price of many products.

In Kazakhstan, inflation rates have also converged over time across its 20 regions, first for (more easily tradable) goods and later for (less tradable) services (De Broeck and others, 1997).

In Poland, some convergence of food prices across major urban areas can be observed. Looking at the behavior of the prices of 12 major items in nine large cities during the first half of the 1990s, dispersion measures drop significantly, although not always monotonically, over time.20 For 8 products, the coefficient of variation exhibits a declining trend, and for 9 products, the same holds for the ratio between the extreme values.21

Fact 10. Full Convergence of the Overall Price Levels to Those Prevailing in Advanced Market Economies Can Be Expected Only in the Very Long Run

By 1996, consumer price levels in transition countries had generally started to move closer to market economy comparators. The unweighted average consumer price level ratio vis-a-vis Austria rose from one-fourth in 1993 to over one-third in 1996 (for the countries appearing in Table 1).22 The increase was particularly rapid in Russia, Ukraine, Belarus, the Baltic countries, and Moldova. The residual gap nevertheless remained large, as worldwide cross-country regressions of prices on per capita incomes would predict (Richards and Tersman (1996): on the whole, there is indeed a significant positive correlation between price and per capita income levels across transition countries (Figure 5).23 The observed pattern is also broadly in line with the comparison with Turkey (Figure 4), and with the position of Portugal, shown here as the European Union member with the lowest price level.

Figure 5.
Figure 5.

Price and Per Capita GNP Level1

Sources: World Bank, World Bank Atlas; European Comparison Program database, 1993; Organization for Economic Cooperation and Development (1996); Tables 1 and 2; and authors’ computations.11996 price level relative to Austria in percent, and 1995 per capita GNP level.

Two additional indicators support the idea that price levels will rise as transition countries grow, namely dollar wages and the structure of consumption (Figure 6). Dollar wages are positively related to productivity and have been rising vigorously, often from low levels (Table A3).24 They are virtually bound to rise over time, as part of the regionwide catch-up process. Likewise, the structure of consumption will shift as those countries become richer, with the share of food declining and that of services increasing. However, the speed of convergence toward advanced economy levels of productivity will naturally tend to slow down as the gap diminishes.25

Figure 6.
Figure 6.

Price Level, Dollar Wage and Share of Food in Consumption

Sources: European Comparison Program database, 1993; Table A3; national authorities; and authors’ computations.1 Dollar wage in July 1996.2 Percent of food in total consumption in 1993.3 Price level in 1996 as a percent of Austria’s.

Scope for Further Research

Some of the stylized facts outlined above are by now part of conventional wisdom, but others less so. Some have been firmly established, while others have only been tentatively identified, with sample length and width too small for comfort. Moreover, the chaotic nature of the reform process in certain countries tends to overshadow some of the trends highlighted as stylized facts. There is therefore considerable room left for more thorough investigations allowing for more refined characterizations of absolute and relative price trends. Finally, over and beyond the phenomenology of inflation undertaken in this paper, much remains to be done on its underlying causes. As time series lengthen and more disaggregated price information becomes available, more robust conclusions will be within reach.

Annex

Table A1.

Consumer Price Inflation1

(Monthly percent change)

article image
article image
Sources: National authorities; Statistical Committee of the Commonwealth of Independent States; and CESMECON.

For each country, the largest single monthly price increaseqwhich in most (but not all) cases coincides with the month of most comprehensive price liberalizationqis marked in bold.

Albania: open inflation started around mid-1990 but a monthly CPI only began to be monitored half a year later.

Hungary: 1990 data are approximations, derived by decumulating a series based on 12-month rates of change.

Romania: the 1990 data prior to November may reflect monthly imputations based on a quarterly series.

Table A2.

Consumer Price Inflation1

(Monthly percent change)

article image
article image
Sources: National authorities; Statistical Committee of the Commonwealth of Independent States; and CESMECON.

For each country, the largest single monthly price increase—which in most (but not all) cases coincides with the month of most comprehensive price liberalization—is marked in bold.

Belarus: hybrid CPI for 1991; CPI thereafter.

Georgia: Retail price index (RPI) through 1993; CPI for Tbilisi in 1994–95, CPI in five main cities in 1996.

Kyrgyz Republic: RPI through end-1994; CPI thereafter.

Lithuania: RPI until May 1992; CPI thereafter.

Moldova: RPI through late 1993; CPI since.

Russia: hybrid CPI for 1991, urban CPI for 1992; expanded average monthly CPI for 1993–94; within-month CPI for 1995–96. The estimates for September to December 1994 were subsequently revised by Goskomstat, when the within-month index was retropolated to the last four months of 1994, to 8 percent for September, 15 percent for October, 15 percent for November, and 16 percent for December (no decimals published).

Tajikistan: RPI for 1991; hybrid CPI for 1992–93; CPI thereafter.

Turkmenistan: RPI through 1992, Laspeyres CPI thereafter.

Ukraine: hybrid CPI for 1991–92; CPI thereafter.

Uzbekistan: RPI through 1993; CPI thereafter.