Analysis of Recent Growth in Low-Income CIS Countries

Contributor Notes

Author’s E-Mail Address:;

This paper analyzes factors that determine recent economic growth in the low-income countries of the Commonwealth of Independent States.2 The main findings are as follows: (1) productivity gains in export-oriented sectors and expansion of exports may have become the main sources of growth in five of the seven CIS-7 countries, while in the early years of transition the output recovery was mainly driven by consumption; (2) economic growth has concentrated in agriculture and the raw material sectors, and, thus, is vulnerable to changes in external conditions; and (3) structural reforms matter for growth, which is consistent with previous research on growth in transition countries.


This paper analyzes factors that determine recent economic growth in the low-income countries of the Commonwealth of Independent States.2 The main findings are as follows: (1) productivity gains in export-oriented sectors and expansion of exports may have become the main sources of growth in five of the seven CIS-7 countries, while in the early years of transition the output recovery was mainly driven by consumption; (2) economic growth has concentrated in agriculture and the raw material sectors, and, thus, is vulnerable to changes in external conditions; and (3) structural reforms matter for growth, which is consistent with previous research on growth in transition countries.

I. Introduction

It has been six years since the resumption of economic growth in the CIS-7 countries after more than five years of decline following the breakup of the former Soviet Union. As noted previously, low-income countries of the Commonwealth of Independent States (or CIS-7) include Armenia, Azerbaijan, Georgia, the Kyrgyz Republic, Moldova, Tajikistan and Uzbekistan. Despite the recent global economic slowdown, growth in the CIS-7 countries has continued.3 In 1998–2002, average annual real GDP growth in the CIS-7 countries was 5.4 percent.

This paper focuses on the sources of recent economic growth in the CIS-7 countries and whether this growth can be sustainable. It focuses on demand-side decomposition of GDP, growth accounting, and a sectoral breakdown of GDP. Econometric analysis is used to identify factors driving economic growth.

The breakdown of GDP by expenditure components points to a shift in the sources of growth during the 1990s from consumption to exports and investment. In the early 1990s, the output recovery and growth in the CIS-7 countries was mostly driven by consumption, financed by external borrowing and income from the underground economy. Since the late 1990s, however, economic growth in most CIS-7 countries resulted from the expansion of exports and an increase in investment. High energy endowments in some countries, artificially low energy prices, and low wages may have contributed to the resilience of exports to the recent global economic slowdown.

A growth accounting framework is employed to analyze whether efficiency gains contributed to the recent growth. The results of the growth accounting exercise demonstrate that, starting from 1998, overall total factor productivity (TFP) growth resumed for the first time since the breakup of the Soviet Union and averaged over 5 percent in three of the CIS-7 countries. This points to the possibility that transition efforts have begun to show results on the production side. The results, however, need to be interpreted with caution because of data weaknesses and difficulties in the measurement of changes in the informal sector activity, labor hoarding, capacity utilization, and depreciation rates.

A sectoral analysis reveals that recent GDP growth in the CIS-7 countries was concentrated in agriculture, extraction industries, production of goods with low degree of processing, trade, and services. The analysis also indicates that with the stagnant share of manufacturing and an increasing dependence on exports of raw materials and agricultural products, the CIS-7 countries are becoming increasingly vulnerable to changes in external conditions.

Finally, the paper presents the results of an econometric analysis of the main factors explaining economic growth in the CIS-7 countries. The results, based on data through 2002, are consistent with previous findings (e.g., Fischer, Sahay, and Vegh, 1998 and Havrylyshyn and Van Rooden, 2000) that lower fiscal deficits, exchange rate stability, as well as implementing structural reforms lead to a higher growth.

The rest of the paper is organized as follows. Section II covers growth performance in the CIS-7 countries in comparison with the rest of the CIS and Central and Eastern Europe and the Baltics (CEEB).4 Section III discusses the sources of growth based on a demand-side decomposition of GDP. Section IV focuses on a growth accounting framework and analysis of TFP growth rates. Section V turns to the sectoral breakdown of GDP and employment. Section VI presents the results of the econometric analysis of the main factors that determine growth in transition. Section VII concludes and discusses policy implications.

II. Recent Growth Developments in the CIS-7 Countries

The CIS-7 countries exhibited strong, although uneven, growth in the five-year period 1998–2003 (see Table 1). During this period, growth in the CIS-7 countries was higher than in the CEEB and other developing countries, and was broadly equal to that in the other CIS countries.

Table 1.

Real GDP Growth in the CIS-7 Countries, 1998-2003

(In percent)

article image
Source: World Economic Outlook (IMF, 2003)


Hereafter “Other CIS” group excludes Turkmenistan

Using the concept of stabilization time, the comparison among the three groups of countries shows that the relative magnitude of the recovery in the CIS-7 countries surpassed that in the other CIS and the CEEB.5 In stabilization time, time is measured from a base year, T(0), which is the year of the start of a stabilization program. Stabilization programs in the CIS-7 and the other CIS countries started between September 1993 and February 1995, while stabilization programs in the CEEB countries began during the period January 1990–October 1993.

Although the CIS-7 countries demonstrated strong performance compared to the other CIS countries and even the CEEB countries, as Figure 1 shows, it is unclear to what extent this growth was merely a stronger rebound from a deep fall in real GDP (see Table 2), a result of reforms, or a response to external developments.6 However, transition indicators of the European Bank for Reconstruction and Development (EBRD) demonstrate that reforms in the CIS-7 countries did not progress as much as in the other transition countries.

Figure 1.
Figure 1.

Real GDP Recovery

Citation: IMF Working Papers 2004, 151; 10.5089/9781451857061.001.A001

Sources: Authors’ calculations; Fisher and Sahay, 1996, IMF, World Economic Outlook.
Table 2.

Contraction Depth of Real GDP


article image
Source: World Economic Outlook, IMF, 2003.

According to official data, the initial output decline in Uzbekistan was much lower than in other CIS countries. It is possible that Uzbekistan’s GDP data are less accurate than those in the rest of the CIS-7 countries. Alternatively, according to Zettelmeyer (1998), Uzbekistan’s output drop was cushioned by low initial industrialization, its cotton production that could readily be sold on international market, and its self-sufficiency in energy.

The Russian crisis in 1998 and the recent global economic slowdown could have had an impact on growth in the CIS-7 countries. While initially the crisis led to a slowdown in most of the CIS-7 countries, as Russia recovered, stronger import-demand from Russia provided an impetus for recovery.7 The real depreciation of CIS-7 currencies against the ruble and other currencies could have supported an increase of the market share of the CIS-7 countries.8

All in all, the recent growth spurt in the CIS-7 countries could be partially explained by their low initial real GDP levels, but also by increased demand from Russia and other countries, and by improved productivity. In the following sections we turn to an analysis of the forces driving growth in the CIS-7 countries and address the question of the sustainability of growth.

III. Demand Decomposition: Has Growth Been Driven by Consumption?

In this section we turn to the analysis of decomposition of GDP by expenditure components to examine what were the main sources of growth in the CIS-7 countries on the demand side. Unfortunately, insufficient data for Tajikistan and Uzbekistan do not allow us to calculate contributions of expenditure components to GDP growth.9

The analysis of the expenditure components of the GDP shows that after 1998 the recovery of the output and growth in five CIS-7 countries started to be driven by net export and investment, rather than by consumption as it was in earlier years (see Table 3).10 High consumption prior to 1998 had most likely driven the recovery by stimulating an increase in capacity utilization and possibly investment.11

Table 3.

Contribution of Expenditure Components to Real GDP Growth, 1998–2002

article image
Sources: National committees on statistics; IMF and authors’ estimates.1/For explanations of methodology used in deriving data see Box 1 in the Appendix.

While consumption continued to be the main driving force for growth in Moldova, in Armenia, Azerbaijan, and Georgia growth in 1998–2002 appeared to be driven by the expansion of exports and investment in export-oriented sectors. In the Kyrgyz Republic, real GDP growth appears to have been export-driven after 1998. However, the data do not fully reflect this because of the supply shock (landslide in the gold mine Kumtor) in 2002.

The most recent observations demonstrate that the pace of investment started to pick up in some CIS-7 countries, mostly because of foreign direct investment coming to export-oriented industries (with the exception of Uzbekistan).12 The share of investment in the GDP, however, still remains low in the CIS-7 countries compared to that in the other country groups (see Table 4).13

Table 4.

Investment Ratios, 1999–2002

(Average; in percent of GDP)

article image
Sources: IMF, OECD.

The increased contribution of net exports to growth in the CIS-7 countries reflects the greater importance of exports for the CIS-7 economies, as the share of exports in GDP remained high or increased (see Table 5). As is noted in Section I, the growth of exports may be partially attributed to the effects of the Russian crisis in 1998 and the recent global economic slowdown.14 The latter also could have had a positive impact on CIS-7 exports, as other countries searched for cheaper suppliers. Indeed, the competitiveness of the CIS-7 countries remained high throughout the period 1997–2002, as average monthly wages stayed below 40 dollars.

Table 5.

Export of Goods and Services, 1997–2002

(In percent of GDP)

article image
Source: World Economic Outlook (IMF, 2003)

IV. Growth Accounting

This section explores to what extent improvements in total factor productivity (TFP) contributed to real GDP growth in 1998–2002. A commonly used approach to link output to inputs and to measure economic performance over time is to set up a growth accounting framework. The growth accounting exercise in this paper follows the methodology of De Broeck and Koen (2000) and makes use of their dataset, extended until 2002.15 The period of extension differs from country to country due to the availability of data on employment. However, we recognize that because of data uncertainties any point estimate is likely to be misleading.

The exercise is based on the assumption that output is produced according to a neoclassical Cobb-Douglas production function in the form,


where Yt is total output at time t; At is the total factor productivity index at time t; Kt is the capital stock; Lt is the labor stock; and 0 < α < 1 is the elasticity of the output with respect to capital. Output growth can be decomposed into capital growth, labor growth, and the residual, which is referred as total factor productivity growth. In the absence of information on factor prices that would allow approximating the elasticities of output with respect to capital and labor, we assume that these elasticities are constant over time and across countries and sum up to 1. The elasticities of output with respect to capital and labor are set equal to 0.3 and 0.7, respectively, which is consistent with previous studies on growth accounting (e.g. De Broeck and Koen, 2000). From equation (1), the change in TFP in logarithmic terms is calculated as


where gi is the growth rate of variable i

The standard formula for the capital stock accumulation is


where It is investment at time t, and δ is the rate of depreciation.16 We make a crude assumption on the depreciation rate for the total capital stock in the economy, taking it as a constant, at the 3 percent level.17 It seems to be reasonable overall, however, the depreciation rate of 3 percent could be perceived as too low for some sectors which require investment in plant and equipment.

The model deserves some further explanation and justification. In particular, some problems may arise from interpretation of TFP growth, abstracting from changes in capacity utilization, the use of a constant depreciation rate, as well as from general data issues. Because measurements of TFP growth have certain biases, they should not be interpreted as simply an estimate of the rate of exogenous technological progress. Changes in TFP in this basic growth accounting framework should be interpreted as residuals that reflect any factor affecting the efficiency with which inputs were used, including changes in working hours, hidden employment, labor hoarding, capacity utilization, changes in resource allocation, and the accuracy with which output and inputs are measured. More generally, this implies that, except for changes in reported employment and investment, the impact of the transition is reflected entirely in changes in estimated TFP.

The basic growth accounting framework could in theory be extended to account explicitly for factor utilization.18 We are not aware, however, of any estimation of capacity utilization rates for labor and capital for the CIS-7 countries. Therefore, this extension for the CIS-7 countries is virtually impossible without making arbitrary assumptions.

To test the robustness of the assumption about the choice of the depreciation rate, alternative calculations were conducted, including a scenario with a very high (up to 70 percent) depreciation rates (obsolescence) in 1992 and a low depreciation rate (1 percent) afterwards. All the findings show very similar qualitative patterns of the changes of TFP and capital stock to the scenario using the constant 3 percent depreciation rate.

We recognize that lack of data is somewhat of a problem, and 13 years is a very short period to uncover patterns in total factor productivity, especially considering the fact that the 13-year time span includes a catastrophic systemic collapse (combined negative supply and demand shocks), only one or two business cycles at most and, perhaps, the resumption of economic activities conducive to TFP growth. However, general TFP trends still provide useful insights.

Average annual growth rates of real GDP, capital, labor, and TFP for different periods are summarized in Table 6. TFP changes were largely negative in the early 1990s, but as stabilization policies took hold, TFP changes turned positive in the CIS-7 countries after 1993. As growth of factor inputs cannot account for the growth performance in the CIS-7 countries during the five-year time span 1993–2002, total factor productivity seems to be important.

Table 6.

Output and Total Factor Productivity Growth 1971–2002

(In percent; period average)

article image
Sources: De Broeck and Koen (2000); and authors’ estimates.

Based on assumption of a 3 percent depreciation rate.

The evolution of real GDP, capital, and labor productivities19 suggests a sharp fall in input productivities associated with the drop in output in the early years of the transition (see Figure 2). By 2002, average capital productivity in the CIS-7 countries almost surpassed its 1993 level. Average labor productivity as of 1999, however, was far below the level in 1993, possibly reflecting labor mobility across different sectors and labor hoarding.

Figure 2.
Figure 2.

GDP, Labor Productivity, and Capital Productivity, 1993–2002

(Average Index; 1993=100)

Citation: IMF Working Papers 2004, 151; 10.5089/9781451857061.001.A001

After 1998, for the first time since the start of the transition period, the overall total factor productivity appears to be increasing in all CIS-7 countries pointing to a possibility that reform efforts have begun showing results on the production side. The latter conclusion, however, must be interpreted with caution because the increase in total factor productivity captures not only the technological progress, but also changes in capacity utilization, the rebound after the Russian crisis, the increase in GDP due to the legalization of the informal sector, and changes in hidden employment.20

The same conclusion can be supported by a regression analysis with the changes in TFP as the dependent variable. The set of independent variables include unemployment rate, initial output per capita, and various transition indicators from the EBRD. The dataset includes the CIS-7 countries and covers the period 1991–2002. The results of the regression analysis are presented in Table 7.

Table 7.

Regression Results: Dependent Variable: Changes in Total Factor Productivity (TFP)

(Fixed-effect estimation)

article image
Source: authors’ estimates.

The regression results show a positive correlation between changes in TFP and infrastructural reform and large-scale privatization, and a negative correlation with unemployment rate. Country fixed-effect coefficients, which reflect initial conditions, are negative and highly significant.

V. GDP Sectoral Decomposition

In this section we turn to the analysis of the sectoral breakdown of GDP from 1995 to 2001 and examine to what extent growth was concentrated in individual sectors, in particular, the raw material sector and the production of goods with a low degree of processing.

Different sectors of GDP are characterized by their share in total GDP, sector share of employment in total employment, and sector share of investment in total investment (see Tables 8 to 10). In addition, we use a growth accounting framework in each sector separately to illustrate recent developments in total factor productivity in key sectors of the economy: industry, agriculture, construction, transport, and trade.21 While we acknowledge data limitations, the results of this section shed light on the evolution of sectoral composition of GDP and update the previous findings of Easterley and Fischer (1994) and De Broeck and Koen (2000). These sectors do not capture the total economy, because the services sector (which is generally viewed as the fastest growing in transition economies) is left aside.

Table 8.

The CIS-7 Countries: Structure of GDP by Sector, 1995–2001

(In percent)

article image
Source: CIS Goskomstat.
Table 9.

Investment by Sector, 1995–2002

(In percent)

article image
Source: CIS Goskomstat.
Table 10.

CIS-7: Employment by Sector, 1995–2000

(In percent)

article image
Source: CIS Goskomstat.

As is commonly accepted, the growth of services during this timeframe was remarkable, especially in the early years of transition because many services were created for the first time. Unfortunately, data are insufficient to analyze growth and TFP in services. The reasons are twofold. First, most countries still use an outdated statistical classification for the sectoral breakdown of GDP, which partially includes services in other sectors (mainly industry and agriculture). Therefore, for this sector it is difficult to identify value added of the output and total employment. Second, it is virtually impossible to calculate capital growth in this sector because of lack of data on investment.

Agriculture has significant potential as a source of economic growth, export diversification, and gainful employment. However, it is likely that increased agricultural efficiency could be accompanied by labor shedding. Agriculture provides direct employment to more than 40 percent of the labor force and accounts for more than 25 percent of GDP in the CIS-7 countries.

On average, TFP growth in agriculture was positive but unstable in part because of varying weather conditions and declining investment (Table 8). Among the CIS-7 countries, Moldova has shown the highest TFP growth in agriculture between 1998 and 2002, which partly reflects the recent breakup of collective farms.

Agriculture may be becoming more labor intensive. It has the highest and increasing share of employment among most of the CIS-7 countries (Table 9), the second highest (but slightly declining) share of GDP (Table 8), a diminishing share in total investment, and unstable declining TFP growth after 1998 (Table 11). This could be an indication that these economies are turning more to subsistence farming. The results of the latest EBRD study (Raiser, Schaffer, and Schuchhardt, 2003) also support the conclusion that in the poorest CIS countries, where a social safety net was not available, many people have been forced back into subsistence farming. However, this could be just a transitory phenomenon because of employment declines in industry and a saturated services sector, pushing people to opt out of more productive sectors to self-employment opportunities in agriculture.22

Table 11.

CIS-7: TFP by Sector, 1995–2002

(Percentage change)

article image
Sources: De Broeck and Koen (2000); and authors’ estimates.
Table 12.

Share of Commodities with a Low Degree of Processing in Total Exports of the CIS-7 Countries, 1995 and the Latest Available Year

article image
Source: IMF, Recent Economic Development Reports.

For Azerbaijan, Moldova, Tajikistan, and Uzbekistan, data refer to 1994.

In industry, real output growth, on average, was higher than TFP growth, which demonstrates increasing capacity utilization in this sector. Industry accounts for the second largest share of GDP in the CIS-7 countries, a segment that has been growing since 1998.23 The share of employment in industry in percent of total employment in the CIS-7 countries either stabilized or declined insignificantly in 1998–2000. TFP growth in industry turned positive for all CIS-7 countries after 1998 (with the exception of Moldova). Although investment in industry as a share of total investment remained high, it has been declining in Georgia and Tajikistan since 1998. The highest share of investment in industry was in Azerbaijan, where more than 70 percent of total investment went to the petroleum sector.

Industrial growth in the CIS-7 countries still depends on production of raw materials and manufacturing of goods with a low degree of processing, such as oil in Azerbaijan, precious stones and metals in Armenia, nonferrous metallurgy in the Kyrgyz Republic, aluminum and cotton in Tajikistan, and gold and cotton in Uzbekistan. New investment is mainly directed into industries associated with production and manufacturing of raw materials and goods with a low degree of processing mainly for exports.

The share of exports of goods with a low degree of processing in total exports has recently increased in all the CIS-7 countries except Armenia (Table 13).

Table 13.

CIS-7: Average TFP Growth by Sector, 1971–2002

(In percent)

article image
Sources: De Broeck and Koen (2000); and authors’ estimates.

Transport includes communications.

Trade includes procurement.

Average TFP growth turned out to be positive in almost all key sectors in the CIS-7 countries in 1998–2002, indicating possibly the first positive results of transition efforts on the production side (Table 8).24 Sectoral capital and labor reallocation significantly affected overall productivity during the early years of the transition. Although in recent years national statistical agencies have begun to improve their estimates of private sector employment and investment, the productivity effect from changes in sectoral output over the recent period merits further examination.

Recent TFP growth varies for different sectors. In agriculture, average TFP growth was positive in all CIS-7 countries, except Azerbaijan in 1995–2002. Industrial TFP growth resumed in the CIS-7 countries (with the exception of Moldova) in 1998–2002. The highest recent TFP growth was observed in the trade sector, where it varied from 5 percent in the Kyrgyz Republic to 33 percent in Armenia. The only exception is Tajikistan, where TFP in the trade sector declined by 16 percent in 1998–2001. In the transport sector TFP continued to decline on average in all countries but the Kyrgyz Republic in 1998–2002. In construction, average TFP growth resumed over the same period of time.

The sectoral reallocation has not necessarily formed a base from which recovery could proceed into sustainable economic growth. Until recently, overall GDP growth in the CIS-7 countries was concentrated in agriculture and industries related to production of raw materials and goods with a low degree of processing. Thus, growth in the CIS-7 countries remains dependent on changes in world prices of raw materials and vulnerable to changes in external conditions. Altogether, this may indicate that for growth to become sustainable, the CIS-7 countries would require further diversification of commodities and export markets. This, in turn, would necessitate new investment, for which continuation of structural reforms is a necessary condition.

Growth prospects could be enhanced by better realizing the trade potential. According to the latest studies,25 a number of factors have had a negative impact on trade—such as restrictive trade policies in Central Asia, weaknesses in the physical infrastructure, and corruption and governance problems in customs and transport services. The poor trade environment has also been affected by insufficient foreign investment, which is critical to overcome deficiencies in the diversification of production.

VI. Econometric Analysis of Major Growth Factors

In this section we turn to the analysis of effects of policy and structural reforms on growth in the CIS-7 countries. We ran regressions, similar to those used in previous studies on growth in transition, but on extended datasets which include the latest observations through the year 2002.26 We examine two datasets: the first one consists of 24 transition countries;27 the second one includes the CIS-7 countries only to test for specific features of growth performance in those countries.

We use the average annual GDP growth rate as the dependent variable. As explanatory variables, we include macroeconomic policy variables to account for the effects of the exchange rate regime and the government budget deficit on growth.28 To measure the progress of reforms, we use the EBRD transition indicators.29

The beginning of transition for a country is defined as a year when most prices were liberalized or when a new national currency was introduced, whichever was earlier. This is a logical way to define the start of the transition period, since prices determined by the market are the most important feature of the market economy. Also, in most countries prices were liberalized before the start of the stabilization programs.

The regression specification for a country i is as follows.


where αi is a country specific intercept,30 GDPgrowth is real GDP growth, ER is a dummy for the exchange rate regime, Fiscal is a variable to control for fiscal policy (general government budget deficit in percent of GDP), InitialShock is a variable determining the impact of initial conditions, warCIS7 is a dummy to control for an impact of civil conflicts in the CIS-7, and TranInd is a set of EBRD transition indicators.

The dummy ER for the exchange rate regime has a value of one in case of a fixed de facto exchange rate or a 2 to 5 percent band for the exchange rate, and a value of zero otherwise.31 The exchange rate regimes are defined using the classification of Reinhart and Rogoff (2000), which distinguishes between de facto pegs or bands and announced pegs or bands.

To control for the impact of initial conditions prior to the transition period we introduced an additional variable, a one-year decline in real GDP (in percent) relative to the pre-transition year (IS).32 It allows us to construct a homogeneous variable for impacts of the initial shocks and to make it comparable across all transition countries. This variable is calculated as follows:


where t is the year when most consumer prices were liberalized.

For a test of the effect of civil conflicts on growth, we use a war dummy for the CIS-7 countries (WarCIS7), which take the value of one in period t, if a CIS-7 country had a military conflict in year t, and zero otherwise.

The EBRD indices (TranInd) cover small-scale privatization, large-scale privatization, governance and enterprise reforms, price liberalization, trade liberalization, competition policy, banking reform and interest rate liberalization, securities markets and nonbank financial institutional reform, and reform of infrastructure. Each of the transition indices takes values from 1 to 4, where 1 represents little or no reform and 4 stands for full achievement of reform in a particular area, according to the EBRD classification system.33

The results of the regressions for all transition countries are summarized in Table 14.34 We used a general-to-specific approach in the regression analysis: the first regression contains all explanatory variables, including insignificant variables. In the other regressions all insignificant explanatory variables were subsequently eliminated.35

Table 14.

Regression Results: Dependent Variable: Real GDP Growth

(Fixed Effect Estimation)

article image
Source: Authors’ estimates.