INTRODUCTION

Transition to the market has led to a massive shift of capital and labour resources accumulated in the communist economies since the period of central planning. But while changes in labour reallocation are fairly well covered by data and are extensively studied in the transition literature, the fate of communist-era capital and its role in the new post-reform development is much less understood. Everybody agrees that a significant part of Soviet-era capital has been effectively destroyed by the switch from the plan to the market; however, no official data exist to account for that transformation. In addition, unlike data for the labour force, capital stock information for different transition economies is not readily available in a format that allows cross-country comparisons. In a major review article, Campos and Coricelli (2002) noted: '... we do not know much about what happened to communist capital. Did it depreciate very fast? How much of it was reallocated to the emerging private and informal sectors?' (p. 806). Recent data on the levels of capital and capital–labour (K/L) ratios for transition economies are not available in the World Penn Table database or in OECD and EBRD databases.¹

Our paper attempts to answer some of these questions by generating a series of estimates of fixed capital for Commonwealth of Independent States (CIS) countries based on certain assumptions about a survival rate of the old Soviet capital. To compare capital accumulation outcomes across countries, we measure capital on per worker basis in the form of K/L and use the same currency units, purchasing-power-parity (PPP) dollars. Having derived the estimates of 'market-quality' capital stock, we then apply them to a basic growth model with the goal of measuring the role of capital in the GDP recovery. In view of data limitations, we consider our estimates as only tentative indicators of the underlying trends and levels of capital accumulation in the studied countries.

From the standpoint of longer-term economic growth, the Schumpeterian 'creative destruction' of old communist-era capital was a positive development. In some way, it can be compared to the destruction of capital in Europe and Japan during the Second World War, which was followed there by record economic growth in the 1950s and the 1960s. In the CIS, however, old capital was not renewed rapidly as economic and political instability that followed market reforms was not conducive to new investment. By the mid-1990s, investment rates in most of CIS declined to around 20% of GDP or less (World Bank, 1996, 2005, 2007; Blanchard, 1997, Popov, 2000, 2006; Campos and Coricelli, 2002; Svejnar, 2002; EBRD Transition Report, 2006).

Importantly, in contrast to post-war Europe and Japan, the destruction of older capital in the CIS, as well as in other transition economies, was economic rather than physical. Large quantities of older assets inherited from communist times became a major obstacle in the way of new investment – a 'curse of old capital' (Popov, 1999; Aukutsionek, 2004). The situation was further complicated by massive privatisation, which often distributed existing assets to new owners for a nominal price, creating strong incentives for divestment and capital flight (Boycko et al., 1996; Blasi et al., 1997; Goldman, 2003).²

In the late 1990s, in particular after the 1998 Russian rouble default and beginning of the oil export boom, investment in the CIS revived. With the establishment of relative macroeconomic stability, foreign capital, which was insignificant in the first years of reforms, started to flow in on a much larger scale, particularly to countries rich in natural resources (Svejnar, 2002; Heinrich et al., 2002, Pavlinek, 2002, Sohinger, 2005; EBRD Transition Report, 2006; World Development Report, 2005; World Bank, 2006). Were the new additions to capital stock large enough to increase it above the pre-reform levels? What is the role played by the new capital accumulation in the economic recovery in the CIS? In this paper, we try to answer these questions based on our capital stock estimates and other available data.

The paper consists of six sections. Following this introduction, the next section discusses conceptual and statistical issues of accounting for 'old' and 'new' capital in transition economies and reviews the literature. Then the subsequent sections describe the estimation methodology and discusse levels and trends of capital stock accumulation in CIS countries, respectively. The penultimate section contains the analysis of economic growth patterns in the CIS using the newly developed 'market-quality capital' series. The final section concludes.

MEASURING CAPITAL IN TRANSITION ECONOMIES: THE 'DEAD CAPITAL' DILEMMA

Measurement of capital in post-Soviet economies raises major methodological problems. Capital accounting there is largely blind to the depreciation of Soviet-era capital stock that resulted from the market transition. It is clear that a significant part of old capital that remains on the books of enterprises is no longer involved in the production process. Including the 'dead' portion of capital in estimates of capital productivity, growth or K/L ratios inevitably introduces a major bias. In the literature, the existing approaches in dealing with this problem range from ignoring it altogether to relying on official data and using various assumptions aimed to estimate capital lost in transition. Most of these estimates relate to the experience of the largest CIS economy, Russia, and are made in the framework of growth accounting studies.

As could be expected, applying different methodologies to capital accounting leads to different and often conflicting results. Thus Khanin and Suslov (1999), using special methodology to correct for the undercount of capital in official statistics, reported that during 1990–1996 the capital stock in Russia shrank by 18% while labour dropped by 11% implying 1.9% per annum reduction of the K/L ratio (see Table 1).³ De Broeck and Koen (2000), using capital stock series based on official data for the 1991–1997 period, estimated annual growth of capital in Russia at -0.1% and annual growth of employment at -2.2%, with an implied K/L growth of 2.1%. In another study of total factor productivity in CIS countries, Kushnirsky (2001) explicitly estimated capital stock in Russia and nine other CIS countries for 1992–1997. This author used Soviet data for the starting level of capital stocks and World Bank investment statistics for the later years. His data did not account for the transition-induced destruction of capital but indicated that the capital stock in nine out of 10 CIS countries has shrunk between 1992 and 1997 by 20%–23%.⁴ For Russia, capital stock in his estimations remained practically unchanged, implying an increase of the K/L ratio during this period by about 1.8% a year (see Table 1).

Table 1 - Estimates of annual growth of capital and labour in Russia and implied changes in K/L ratio(%).

Full table

Dolinskaya (2001), using official data for capital and labour for the 1992–1997 period, estimated capital growth in Russia at 0.1% and labour growth at -1.5%, implying K/L growth of 1.6%. In the same study, the author provided estimates of capital stock adjusted for capacity utilisation and employment growth adjusted for underemployment. With these adjustments, implied annual growth of K/L ratio was just 0.3% a year. Melyantsev (2003) adjusted capital stock for both capacity utilisation and obsolescence of capital as measured by its age and found it shrinking in 1991–1998 at 7.5% a year. With labour force decreasing at 2.5%, this implied a steep reduction of the K/L ratio at 5% a year.⁵ However, application of a similar adjustment methodology to the 1999–2001 period indicated an increase of the implied K/L ratio at the rate of 1.8% a year (see Table 1).

The recent World Bank (2006) publication, Measuring Capital for the 21st Century, provides capital stock data for several transition economies for the year 2000. It includes capital estimates for three countries of the CIS group: Georgia, Moldova and Russia. The estimates are in dollars but use market rather than the PPP exchange rates. While the utilisation of market exchange rates is justified for the purposes of that study, to investigate the dynamics of various components of broadly defined national wealth in individual countries, it makes the data unsuitable for cross-country comparisons. See Appendix A for a more detailed review of this issue.

In several studies, the problem of transition-induced destruction of capital is addressed by adjusting the official capital stock based on capacity utilisation data. This approach has at least two problems. One of them is related to the origin of capacity utilisation measures. Based on surveys of managers, these estimates depend crucially upon respondents' and survey-takers' definitions of fully utilised capacity. In Russia, for example, four main organisations that conduct surveys on capacity utilisation in industry report rates of utilisation for the same periods that differ by 25–30percentage points (Oomes and Dynnikova, 2006).⁶ One of the leading experts in this field admits: '...our idea of existing production capacities is at present quite vague and does not permit reliable analysis of this subject' (Aukutsionek, 2004). No less important, even the most accurately measured capacity utilisation rate may not be a good measure of effective capital stock as its dynamics can be dominated by demand-side factors not related to longer-term usability of existing assets. This is particularly true for transition economies of the 1990s. It is well-known that during the transformational recession, all CIS countries experienced a deep drop in capacity utilisation levels. However after 1998–1999, utilisation rates recovered with many previously unused parts of capital stock reemployed in the new market environment (Aukutsionek, 2004; Oomes and Dynnikova, 2006). Thus relying on capacity utilisation can lead to a major underestimate of the market-worthy part of the Soviet capital stock.

Instead of using capacity utilisation, some authors suggest estimating the degree of transition-induced destruction of capital by applying to it a higher rate of depreciation. Using GDP and employment data, production function assumptions and the Kalman filter statistical technique, Hall and Basdevant (2002) estimate a hypothetical depreciation rate for the Russian capital stock in 1994–1998. According to their results, annual depreciation of fixed capital fluctuated between 4.5% and 10%, with the fastest depreciation occurring in the earlier part of the period. This approach provides an interesting insight into possible depreciation patterns of the Soviet-era capital. However, it implicitly assumes a direct linkage between the fall in output and the reduction of capital stock, which may not hold if one allows for changes in the capacity utilisation rate.⁷

An alternative to the above approaches is to make an assumption about the overall share of Soviet-era capital that was effectively destroyed by the switch to the market. Estimates of the portion of the overall capital stock 'killed off' by the market transition vary. Some economists consider it to be no more than 15%–25% (McKinsey, 1999); others believe that it may be as high as 50% (Kushnirsky, 2001). Deliktas and Balcilar (2005) estimated that between 1989 and 1995, 25 transition economies lost a half of their capital stock.

The most specific estimate of the rate of transition-induced destruction of the communist-era capital stock is probably provided in a study of capital formation in Hungary by Darvas and András (2000). The authors of this study estimated that the transition to the market in Hungary effectively eliminated 33.3% of the stock of existing equipment and 12.5% of the stock of structures. Applying their approach to the Soviet economy and taking into account that, before the transition, the ratio between the equipment and structures in the Soviet capital stock was approximately 30:70, implies that the switch to the market could have made about 20% of the Soviet capital stock redundant. However, this estimate looks too low if one considers that before the transition, the Soviet economy was less market-oriented than was Hungary's. Thus a larger share of Soviet capital was doomed to become redundant. For example, if one assumes that 50% of equipment and 15% of structures in the USSR were of non-market quality, the rate of market adjustment for all of the Soviet capital stock would have to be about 25%.

Based on existing data, there seems to be no way to directly determine the break-down of Soviet capital into market and non-market segments. However, there are several ways to estimate these shares indirectly. One of them is to consider the GDP-based estimates of overall distortion of the CIS economies on the eve on transition. According to estimates by Popov (2000), by the beginning of transition, compared to the market economies of similar level of development, about 40% of the GDP in CIS countries was allocated to what Popov termed the wrong sectors and trade destinations, such as the vast defense industry and unprofitable trade with Cuba. It is reasonable to assume, at least initially, that the degree of misallocation of capital stock was of the same magnitude. However, over time, some of this misallocated capital stock, including most business structures, could be redirected to new uses. This implies that an overall share of Soviet capital permanently unusable under the market should be less than 40%.

Another way to arrive at an estimate of the share of non-market-worthy capital is to examine enterprise-level surveys and survey-based studies. Thus, according to an industry-wide survey by the Russian Government Center for Economic Analysis in 2002, enterprises reported that only 4.4% of their spare capacities could be used to produce competitive products for domestic or foreign markets (World Bank, 2003). Taking into account that the overall level of spare capacities reported in this survey was 44%, the implication is that about 40% of nominally existing capacities of these enterprises were not 'market-worthy'. An earlier survey (Borisov, 1998) found that in Russian machine-building, just 19%–54% of existing equipment, depending on industry, was capable of manufacturing market-quality products. Both of these studies point to the 30%–40% range for a share of non-market-worthy assets in the capital stock. However, these estimates relate only to the industrial sector. Adding to this calculation all of residential capital, which at the end of the Soviet period comprised about one-third of the total, should reduce this estimate to 20%–25%.⁸

Thus, while existing studies offer a wide range of estimates for the market-induced destruction of Soviet capital from15% to 50%, a more careful consideration points to a narrower range in the area of 20%–40%. In the next section, we will match these estimates with different assumptions for depreciation rates and other data with the aim of narrowing them down to one baseline assumption.

ESTIMATING METHODOLOGY: COMBINING THE OLD AND THE NEW CAPITAL

The task of estimating fixed capital stock in a CIS country during the transition period will be approached in three steps: (i) estimation of the market-worthy part of the old Soviet capital; (ii) estimation of the new capital resulting from the post-reform investment; (iii) combining the old and the new components of capital:

where K_t is the total capital stock in reform year t, K_t(old) is the value of old fixed capital remaining in service in year t, K_t(new) is value of new fixed capital stock accumulated by the year t since the start of market transition, all measured in constant PPP dollars in prices of 2000.

Knowing the amount of pre-reform capital stock K₀, the share of this capital that became unusable under the market conditions, , and the rate of depreciation of the remaining market-worthy part of the old capital, r_old, we can find the amount of old capital surviving in reform year t as:

The amount of new capital in the year t is found by summing new fixed investment made between the first year of reforms and the year t with investment depreciated at the rate r_new:

K_t(new) is the stock of new capital accumulated between 1992 and year t (t=0 for the year 1992), I_t-n is fixed investment in year t-n measured in constant PPP dollars, and r_new is the depreciation rate for new capital.

Based on the above and assuming the last year of pre-reform period for the CIS countries being 1991, the total amount of fixed capital in year t is found as:

K₁₉₉₁ is the amount of the Soviet capital stock at the end of 1991. In calculating the stock of new investment, we assume geometric pattern of depreciation and 5% annual depreciation rate, the value of r in equation 4.⁹

Annual fixed investment valued in constant PPP dollars in prices of 2000 is computed based on shares of fixed investment in the GDP provided in the World Bank World Development Indicators (2007) database. These shares are adjusted to reflect the overstatement of real value of investment in the CIS data. A number of studies, including the International Comparison Project, have established that, after the market transition, prices of investment goods increased much faster than those of other goods and services. As a result, the reported shares of investment spending in the GDP of the CIS countries based on nominal prices exceed similar shares computed in real terms by 1.5–2.5 times (Tabata, 1997; Mezdunarodnye sopostavlenia, 2005; Khanin, 2006). To arrive at the real value of investment, expressed in PPP dollars, one should use special PPP exchange rates for investment spending. However, such rates for CIS countries are not available other than for a few countries and for select years. To overcome this problem, we used PPP exchange rates for investment spending assuming that on average one-third of investment was spent on internationally tradable components (such as equipment), priced at market-exchange rate and two-thirds – on domestic components (such as construction) – priced at levels close to the general price level for GDP. The resulting price ratios used to adjust the World Bank data based on nominal prices are presented in Appendix B.

Estimating 'Old' capital

For estimates of the old Soviet capital inherited by the CIS countries from the pre-reform period, we start with the official Soviet data on gross capital stock in rubles for the year 1989, K₁₉₈₉. For Russia, based on Goskomstat data, capital stock at the end of 1989 was RR1,834 billion (Goskomstat, 2006, p. 323). To convert these data into PPP dollars of 2000, we used the PPP dollar/ruble ratio based on World Development Indicators figure for the Russia's 1989 GDP in dollars and in rubles.¹⁰ Applying the resulting conversion ratio of 2.373 to the capital stock, we arrive at the $PPP2000 valued estimate for the Russian capital stock at the end of 1989: $4,352.4 billion. For other CIS countries, which in 1989 were republics of the USSR, capital stocks are computed based on proportion of their production capital measured in rubles to that of Russia, as reported in Narodnoye Khozyaistvo (1989).¹¹ To bring the estimates of CIS capital stocks forward to the end of 1991, we need to add to it the investment made in 1990–1991, expressed in $PPP2000, and subtract the 1990–1991 depreciation. Assuming a 5% geometric depreciation rate, we estimate the pre-reform capital stock at the end of 1991 as ¹²

Here K₁₉₈₉, K₁₉₉₁, I₁₉₉₀, I₁₉₉₁ stand for capital stock and fixed investment in respective years, all measured in constant PPP dollars. The results of these calculations are presented in Table 2.

Table 2 - Capital stock of CIS countries in 1989–1991.

Full table

Selecting the market adjustment ratio

To extend the Soviet capital series to the post-reform years using Equation (2), we need to select the market-adjustment ratio and the depreciation rate r_old. The review of existing estimates of market-induced destruction of Soviet capital presented in the previous section suggests a range of 0.20–0.40 for . Choosing a depreciation rate involves a trade-off between and r_old. Selecting a lower level of market adjustment, such as 0.20, implies that a significant part of the Soviet-era equipment and structures remains in the capital stock. To account for that, the average rate of depreciation for such capital stock should be high. If, on the contrary, one assumes a high level of market adjustment, such as 0.40, the implication is that most of the old equipment and significant part of older structures are gone. This will require applying a relatively lower rate of depreciation to the remaining capital stock.

Figure 1 presents the outcomes of nine different scenarios resulting from matching three different market adjustment ratios, 0.20, 0.30 and 0.40, with depreciation rates ranging from 3% to 7% applied to the Soviet-era capital stock. As expected, different combinations of higher and lower r_old over time lead to approximate convergence with alternative match-ups of lower and higher r_old.

Figure 1.

Capital stock using different market-adjustment ratios and depreciation rates, Russia, 1991–2005.

Full figure and legend (31K)

In principle, we could use multiple combinations of market adjustment and depreciation rates and even extend the range of these variables to include smaller iterations. However, in practical terms it would make the cross-country comparisons unmanageable. Based on the above arguments, we settle instead on a baseline assumption of =0.3 and the depreciation rate r_old=0.05. The assumption that on average the market transition has destroyed 30% of the Soviet-era capital stock in all of the CIS countries is admittedly crude. However, it corresponds to the mid-range of existing estimates and it is supported by the indirect indicators coming from other studies and surveys. Thus it is a reasonable starting point for an exploratory investigation.

Figure 2 presents the results of applying the baseline assumptions to Russia. The upper curve reflects the official unadjusted data on capital stock. The lower curve indicates the amount of capital with a one-time market adjustment of 30%. The space between the two curves represents the 'walking dead' part of the capital stock, assets that remained on the books of enterprises but were unusable under the new market conditions. In the next section, we will present and analyse similarly derived capital stock time series for other CIS countries.

Figure 2.

Official Russian capital stock and estimated Russian capital stock using 30% market-transition adjustment and 5% geometric depreciation in $PPP 2000, 1989–2005.

Full figure and legend (21K)

LEVELS AND TRENDS IN CAPITAL ACCUMULATION

To compute K/L ratios, CIS countries' capital stocks estimated as described above were divided by the labour force, as reported in World Bank (2007). The choice of labour force rather than employment for the denominator of K/L ratios is justified in view of long-term nature of the fixed capital stock and short-term nature of employment fluctuations.¹³ Table 3 and Figures 3 and 4 present data for old, new and total capital stock for 11 of 12 CIS countries on per worker's K/L basis, as well as rates of change of related variables.¹⁴

Figure 3.

K/L ratio in CIS economies in 2005: new versus old capital.

Full figure and legend (57K)

Figure 4.

K/L ratio in CIS economies, 1991–2005.

Full figure and legend (33K)

Table 3 - Capital, labour and K/L ratio in CIS economies in 2005: summary statistics.

Full table

As expected, the overall K/L ratio at the end of the period, 2005, is found to be highest in Russia ($40,326). The other three larger CIS economies – Kazakhstan ($32,626), Belarus ($33,384) and Ukraine ($32,657) – also have relatively high K/L ratios. They are followed by a group of three countries with K/L in $20–25,000 range consisting of Armenia ($24,132), Azerbaijan ($21,657) and Georgia ($20,402). The low K/L group includes the rest of the CIS: Moldova ($15,395), Uzbekistan ($12,860), Kyrgyz Republic ($11,101) and Tajikistan ($10,032).

The data in Table 3 and Figure 3 show the relative role of new investment in total capital accumulation. By the end of 2005, the average unweighted share of new capital in capital stock of CIS countries has reached about two-fifths (39%). The absolute leader in new capital accumulation is oil- and gas-exporter Azerbaijan with 58%. It is followed by the larger CIS economies: Belarus, 49%; Ukraine, 46%; Russia, 46%; Kazakhstan, 42%; and Uzbekistan, 44%. Other lower K/L countries show shares of new capital between 25% and 37%.

In view of a significant drop in investment activity during most of the 1990s, the proportion of new capital in total capital stock may appear high, for example in excess of 40% for the larger economies such as Russia, Ukraine, Belarus and Kazakhstan. This seems to contradict the reality of these economies, where older pre-1992 stocks very much dominate through their physical presence, as anybody who travelled in the former USSR in the last several years can readily testify. However, one should keep in mind that our estimation focuses only on the market-quality part of capital stocks. This estimate ignores the economically non-viable part of existing assets regardless of their physical presence. In essence, it represents the hypothetical market value of fixed capital stocks under the condition of a perfect market for used assets (Bohm et al., 2002).

One way to verify the credibility of these estimates is to compare them to K/L ratios of market economies with similar levels of GDP per capita but no history of central planning. In such countries, all of the capital stock is supposed to be of market-quality. Assuming comparable levels of capital efficiency, the K/L levels in these countries should be similar to those of transition economies. Two recent studies provide estimates of capital-stock accumulation in countries of South America to which transition economies are often compared (Hofman, 2000; Souza, 2005). Based on estimates from these studies converted to $PPP 2000, the K/L ranged there in 1994 from $21,000 per worker in Columbia to $35,500 in Mexico and $38,500 in Argentina. In Brazil, a country with population and GDP similar to that of Russia, the capital stock per worker was $29,000 in 1994 and $37,000 in 2000 (Hofman, 2000, p. 51; Souza, 2005, Table 1).¹⁵ These numbers compare reasonably well to our estimates.

Some of the trends and patterns of capital accumulation in CIS can be observed in the data presented in Table 3 and in Figure 4 (also see Appendices C and D). Over the period of observation, out of 11 CIS countries only one, Azerbaijan, was able to increase its market-quality capital stock above the level that existed at the end of 1991. In Belarus and Russia, the losses of the 1990s were largely restored, while in other countries declines were significant, ranging between 10% in Ukraine to 33% in Georgia (see column 4 of Table 3).

In terms of K/L ratios, the trends were somewhat different. Due to the shrinkage of the labour force in such CIS countries as Armenia (-31%), Belarus (-9%) and Ukraine (-15%), the K/L ratio increased by 2005 in spite of the fall of capital stock (see Figure 4 and Appendices C and D). In Azerbaijan, on the contrary, fast growth of labour force (+31%) exceeded the growth of capital (+16%) leading to the overall decline of K/L. In Russia, a major decline in the labour force occurred between 1991 and 1994–1995 leading to an initial increase in K/L. After that, the K/L in Russia declined and started to recover only in 2002. The lower capital stock economies, Kyrgyz Republic, Moldova, Tajikistan and Uzbekistan, show a steady downward trend in the K/L that reflects both the decline of capital stock and an increase in labour force.¹⁶ Armenia, Georgia and Kazakhstan demonstrate the fastest recovery in the K/L ratio after 2002 (see Figure 4).

CAPITAL ACCUMULATION'S CONTRIBUTION TO ECONOMIC GROWTH

To investigate the impact of our measure of market-quality capital on economic growth, we used our capital stock data in the framework of Solow's neoclassical growth model:

where Y is real GDP; K the capital; L the labour and A is the total factor productivity (TFP). Assuming a Cobb–Douglas production function, we have:

In (7), and (1-) measure capital and labour share of income, respectively. In our study, we assume =0.4 and (1-)=0.6 and apply that breakdown to all CIS countries for the two sample periods: 1995–2005 and 1998–2005.¹⁷ We then estimate the contributions of capital and labour to output growth and determine TFP as a residual. The simplified growth accounting formula used for actual computation was

Y/Y, K/K, L/L, A/A are measured by average annual growth rates of respective variables.

The GDP growth data used in the computations are taken from the World Bank (2007) and are based on the official national account statistics. These data are often criticised as being flawed. The alternative approach would be to use the GDP growth estimates coming from individual researchers. However, such data do not cover all CIS countries and/or are not available for all years. In addition, much like for the Soviet period, there is no consensus approach for such growth estimates and the methodology of GDP growth computation used in some of these studies is controversial.¹⁸ On this issue, we are generally more in agreement with De Broeck and Koen (2000) who after reviewing various adjustment to official CIS–GDP growth data proposed by individual authors concluded that 'notwithstanding the numerous caveats ... official national accounts appear to broadly reflect the main characteristics of the underlying output movements during the transition' (p. 7).

Results for growth accounting for CIS countries are presented in Table 4 for 1995–2005 and Table 5 for 1998–2005 and also in Figures 5 and 6. These results largely confirm findings of previous studies, according to which the key factor of GDP decline and recovery in the CIS countries in the period of transition was the dynamics of TFP (De Broeck and Koen, 2000; Kushnirsky, 2001; Dolinskaya, 2001; Melyantsev, 2003; Ahrend, 2006). However, using capital stock data adjusted for market transition provides some additional insights.

Table 4 - Contributions to economic growth in CIS countries of capital, labour and total factor productivity: 1995–2005 (=0.4).

Full table

Table 5 - Contributions to economic growth in CIS countries of capital, labour and total factor productivity: 1998–2005 (=0.4).

Full table

Data in Tables 4 and 5 confirm a larger contribution of capital to growth in countries, where capital accumulation was relatively fast: Azerbaijan, Belarus and Kazakhstan. Here, for 1995–2005 and 1998–2005, capital contributed between 10% and 17% of GDP growth. In Armenia, Kazakhstan, and Uzbekistan for 1995–2005 and in Ukraine for 1998–2005, the contribution of capital was positive but small, between 1% and 7%, while in all other countries it was negative for both periods.

In terms of TFP growth, the best results are demonstrated by Georgia and Ukraine, where in both periods it contributed close to 120% of the GDP growth. The TFP contribution of 100% or more for both of the considered periods was also observed in Armenia, Moldova and Russia. TFP contribution was the smallest in Azerbaijan and Uzbekistan, but even there it was more significant than that of labour and capital (see Tables 4 and 5).

The relatively large contribution of TFP compared to capital should not come as a surprise since the basic growth accounting framework used here only reflects the quantity of capital, while improvements in its quality are captured by TFP. Acceleration of investment activity at the end of the 1990s brought in new equipment and improvements to remaining Soviet-era capital thus increasing the overall productivity of capital stock. This process was the opposite to what happened in the early 1990s, when lack of investment worsened the obsolescence of capital and contributed to the drop of TFP (De Broeck and Koen, 2000). The increasing efficiency of capital in the CIS countries in the recent years is reflected by the dynamics of K/Y ratio presented in Figure 7. Here one can see the significant increase in capital productivity in particular in 1998–2005 period. By 2003–2005, the K/Y ratio in most of the CIS economies has converged around 3–3.5, which matches the K/Y range for market economies of a comparable level of GDP per capita (Hofman, 2000; Souza, 2005).

Figure 7.

Capital–output ratio in CIS economies: 1995–2005.

Full figure and legend (31K)

SUMMARY AND CONCLUSIONS

The goal of this paper was to generate a series of estimates of fixed capital for the CIS countries, based on the assumption that a major part of the old Soviet capital inherited by these countries from the communist past was effectively destroyed with the advent of the market. In contrast to official capital stock data, which are blind to reform-induced depreciation, our estimates are focused on the market-quality part of capital and should thus provide a more accurate picture of capital formation in post-communist economies. To compare capital accumulation outcomes across CIS countries, we measured capital stock both in absolute terms and on per worker basis in the form of the K/L ratio and used uniform currency units (PPP dollars).

The results of our estimation indicate that, over the period of observation between 1992 and 2005, out of 11 CIS countries only one, Azerbaijan, was able to increase its market-quality capital stock above the level that existed at the start of transition. This country also leads in the proportion of new, post-reform investment in the capital stock. In Russia, thanks to the revival of investment since 1999, the losses of the 1990s were largely restored. However, in most of the other countries the decline was significant ranging between 10% in Ukraine and 33% in Georgia. Smaller, non-resource-rich economies of the CIS have been falling behind in capital accumulation. Having started from lower levels of market-quality capital at the beginning of reforms, these countries also experienced larger declines in their capital stock. In Central Asian countries, the reduction of capital stock was complemented by the rapid growth of the labour force leading to even sharper declines of K/L.

In absolute terms, Russia remains the highest capitalised CIS country with (K/L) ratio of about $40,000 per worker. The lowest capitalised countries, all Central Asian states, have K/L ratios of $10–$13,000. Since the start of market reforms, the gap between the highest and lowest capitalised countries, Russia and Tajikistan, has increased more than 1.5-fold: from 2.5:1 to 4:1. These results seem to confirm divergent trends in capital accumulation among transition economies coming from enterprise-level surveys, such as the recent World Bank/EBRD studies (EBRD Transition Report, 2005; World Bank, 2005).

Having derived the estimates of market-quality capital stock, we applied them to a basic growth model, with the goal of measuring the role of capital in the recovery of GDP growth in the CIS region. The results of this growth accounting exercise indicate that in both of the studied periods, 1995–2005 and 1998–2005, GDP growth was dominated by the dynamics of TFP, while the contribution of capital growth was modest. Even in the countries where the capital stock has grown fast, such as Azerbaijan, capital's contribution ranged from 10% to 20%. The reason behind this seems to be the fact that capital accumulation in the CIS in the 1990s impacted GDP growth primarily through improvements in the quality and age of the capital stock rather than through its physical expansion. In the framework of our capital stock data, this explanation is supported by the relatively high proportion of new capital in the total capital stock and by the rapid decline of the K/L ratio, which is the direct evidence of improvements in capital productivity.

To conclude, after 15 years of development under the market, in most of CIS economies the creative destruction stage of capital formation seems to be coming to the end with the modernisation of capital increasingly contributing to the revival of productivity and growth. However, the smaller resource-poor CIS countries are still suffering from the consequences of the investment drought of the 1990s and can hardly overcome these consequences without massive investment from abroad.

Notes

¹ Individual studies are reviewed in the following section.

² In the early 1990s in many TE, in particular in the CIS, asset stripping had reached, in words of some observers, the level of 'national sport' (Brown and Earle, 2002).

³ In this and later works Khanin and his colleagues maintain that official Russian data for fixed capital stock, even after periodical revisions, do not correctly reflect its current replacement value (Khanin, 2006; Khanin and Fomin, 2006). However, in their own estimates they use a very broad definition of fixed capital including in it all assets that are kept on the books of enterprises, regardless of their age, condition and usability (Khanin and Fomin, 2006, p. 17). Such an approach leads to a very high estimate of Russian capital stock, surpassing the official one by about 10 times, in 2001, for example, 206.5 trillion versus 20.2 trillion rubles. Taken at face value, this estimate implies a capital-output ratio of 23, and annual depreciation, even computed by the authors at a low 2% rate, reaching almost one-half of GDP. While trying to account for all assets recorded on balance sheets of enterprises might be a legitimate approach, in our study the goal is different as we aim to estimate only the 'market-worthy' part of capital that remained usable after the transition. On Soviet capital stock estimation, see also Kontorovich (1989).

⁴ Not included in the study were Tajikistan and Turkmenistan.

⁵ During the Soviet period, by some estimates, the K/L ratio in the economy was increasing rapidly: 7.4% a year in the 1950s, 5.4% in the 1960s, 5.0% in the 1970s and 4% in the 1980s (Easterly and Fischer (1995)). However, these estimates probably overstate the actual growth of the K/L as they use official data for capital, not fully adjusted for the impact of inflation and depreciation (Nove, 1981; Kontorovich, 1989, 2001; Khanin, 2003).

⁶ Measurement of capacity utilisation in post-Soviet economies is further complicated by the issue of the so-called 'mobilisation capacities.' Since Soviet times many industrial enterprises in the CIS are obligated to keep on reserve but not use a part of their capacities in case they are needed in a war situation or other national emergency. In some surveys, these are excluded from the totals while in others they are included.

⁷ The allowed depreciation rates in this period were even higher but had little meaning in an environment of severe inflation. In a detailed study of depreciation in Russian industry, Linz (1999) reported that between 1992 and 1995 just 5% of enterprises used annual depreciation rate of 10% or less, while at least a quarter of enterprises used rates above 50% and 3% used rates above 75%. However in the same period, market prices for capital goods increased much faster than the officially allowed revaluation, making these depreciation rates irrelevant to actual replacement of the capital stock, with or without market adjustment. For example, in 1992 the official revaluation of capital stock increased its value by nearly five times, while 1992 prices of construction and equipment increased by 15–18 times (Goskomstat, 2006, p. 644).

⁸ The lower estimate is based on assumption that residential capital was not affected by the market transition. This, of course, is a simplification, as some of the Soviet housing stock was tied to industries and locations that would become non-viable under the market. This includes one-company-towns, military bases, settlements in the Far North where residential housing had to be abandoned.

⁹ Geometric pattern of depreciation and depreciation rate of 5% are chosen based on review of recent studies of the investment accounting literature (see: Fraumeni, 1997; Katz and Herman, 1997; Larson et al., 2000; Bohm et al., 2002; World Bank, 2006).

¹⁰ The data for the Soviet-era GDP remain a subject of a debate. Numerous estimates continue to circulate in the literature, especially since the discussion of the CIA worked on this topic in the early 1990s. The World Bank numbers on the Soviet GDP in PPP dollars for 1989–1991 that we use in this study match one of the more authoritative of these estimates, coming from the International Comparison Project and cited favourably in a review article by Bergson (1997). The official ruble-denominated data for the same period from Goskomstat are closely correlated with estimates by at least some of the individual researchers (see, eg, Rosefielde and Kuboniwa, 2003). However, we acknowledge the existence of alternative estimates (see, eg, Kontorovich, 2001; Khanin, 2006).

¹¹ Following Kushnirsky (2001), capital stock for individual republics is assumed to have the same allocation between residential and non-residential assets.

¹² Data for the 1990–1991 investment by republic are taken from World Bank (2007) and Kushnirsky (2001).

¹³ Another reason for using labour force rather than employment data is that some of the CIS countries provide unrealistically low number of unemployed counting among them only those officially registered as such with employment agencies. However, even labour force data can be imprecise as it may not adequately account for some categories of workers, such as illegals. It is well known that illegal migration is becoming an increasingly important part of labour force in some of the CIS countries, in particularly in Russia.

¹⁴ Turkmenistan is not included in the computations as data for new capital formation for Turkmenistan is not available for all years.

¹⁵ In the original sources, the capital stock is presented on per capita basis measured in constant PPP dollars of 1980 and 1996. Conversions were made using US deflators for fixed investment.

¹⁶ A note of caution: the existing labour force data may not fully reflect the dynamics of illegal labour flows and its impact on the size labour force in other CIS countries including the labour abundant countries in Central Asian region.

¹⁷ The choice of =0.4 is based on computation of average share of profit-type incomes in the national income of Russia in 1995–2005. The alternative value of =0.3 was also used. The differences were not important, and the results are available from the authors.

¹⁸ Among the better-known alternative estimates of GDP growth in the post-transition period are those provided for Russia by Gregory Khanin and his colleagues (Khanin and Suslov, 1999; Khanin, 2006; Khanin and Fomin, 2006). However, in revising the official GDP figures, Khanin and his co-authors rely heavily on coefficients of elasticity between GDP growth and electricity consumption and rail freight volume derived from the US experience in 1970–1975 (see eg Khanin, 2006, pp. 153–159). Assuming that such coefficients, especially coming from the relatively short and atypical period in the US economic history, the beginning of the energy crisis, would hold for Russia in the 1990s may not be very realistic. On the quality of macroeconomic data in the CIS, see also Koen (1996), Kontorovich (2001) and Rosefielde and Kuboniwa (2003).

References

1. Ahrend, R. 2006: Russia's post-crisis growth: Its sources and prospects for continuation. Europe–Asia Studies 58(1): 1–24.
2. Aukutsionek, S. 2004: Are Russian enterprises capacity constrained? Problems of Economic Transition 46(10): 80–96.
3. Bergson, A. 1997: How big was the Soviet GDP? Comparative Economic Studies 39(1): 1–14.
4. Blanchard, O. 1997: The Economics of post-communist transition. Oxford University Press: Oxford and New York.
5. Blasi, JR, Maya, K and Douglas, K. 1997: Kremlin capitalism: privatizing the Russian economy. Cornell University Press: Ithaca.
6. Bohm, B, Gleib, M and Zeigler, D. 2002: Disaggregated capital stock estimation for Austria – methods, concepts and results. Applied Economics 34: 23–37. | Article |
7. Borisov, V. 1998: Perspectivy mashinostroennia (Prospects of manufacturing). Ekonomist 1: 52–61.
8. Boycko, M, Shleifer, A and Vishny, R. 1996: Privatizing Russia. The MIT Press: Cambridge.
9. Brown, JD and Earle, JS. 2002: Gross job flows in Russian industry before and after reforms: Has destruction become more creative? Journal of Comparative Economics 30(1): 96–133. | Article |
10. Campos, NF and Coricelli, F. 2002: Growth in transition: What we know, what we don't, and what we should. Journal of Economic Literature XL: 793–836.
11. Darvas, ZM and András, S. 2000: Capital stock and economic development in Hungary. Economics of Transition 8(1): 197–223. | Article |
12. De Broeck, M and Koen, V. 2000: The great contractions in Russia, the Baltics and the other countries of the former Soviet Union: a view from the supply side. IMF Working Paper WP/00/32, 1–37.
13. Deliktas, E and Balcilar, M. 2005: A comparative analysis of productivity growth, catch-up, and convergence in transition economics. Emerging Markets Finance and Trade 41(1): 6–28.
14. Dolinskaya, I. 2001: Explaining Russia's output collapse: aggregate sources and regional evidence. IMF Working Paper WP/01/06, 1–32.
15. Easterly, W and Fischer, S. 1995: The soviet economic decline. World Bank Economic Review 9(3): 341–372. | Article |
16. EBRD Transition Report, annual, various years. European Bank for Reconstruction and Development: London.
17. Fraumeni, BM. 1997: The measurement of depreciation in the US national income and product accounts. Survey of Current Business 77: 7–23.
18. Goldman, M. 2003: The privatization of Russia: Russian reform goes awry. Routledge: New York.
19. Goskomstat. 2006: Goskomstat (Rosstat), annual: Rossiiskii Statisticheskii Ezhegodnik [(Russian Statistical Annual)]. State Committee for Statistics/Federal Statistical Service: Moscow.
20. Hall, SG and Basdevant, O. 2002: Measuring the capital stock in Russia: An unobserved component model. Economics of Planning 35: 365–370. | Article |
21. Heinrich, A, Kusznir, J and Pleines, H. 2002: Foreign investment and national interests in the Russian oil and gas industry. Post-Communist Economies 14(4): 495–507. | Article |
22. Hofman, A. 2000: Standardised capital stock estimates in Latin America: A 1950–94 update. Cambridge Journal of Economics 24: 45–86. | Article |
23. Katz, AJ and Herman, SW. 1997: Improved estimates of fixed reproducible tangible wealth. Survey of Current Business 77: 69–92.
24. Khanin, G. 2003: Ekonomicheskaia istoriia Rossii v noveishee vremia [(Economic history of Russia in modern times)], Vol. 1, Novosibirsk State Technical University: Novosibirsk.
25. Khanin, G. 2006: Economic growth and the mobilization model. In: Ellman, M (ed). Russia's Oil and Natural Gas: Bonanza or Curse? Anthem Press: London pp. 151–172.
26. Khanin, G and Fomin, D. 2006: Otsenka vosproizvodstva osnovnogo kapitala ekonomiki Rossii. Voprosy statistiki (10): 6–19.
27. Khanin, G and Suslov, N. 1999: The real sector of the Russian economy. Europe-Asia Studies 51(8): 1433–1454. | Article |
28. Kontorovich, V. 1989: Inflation in the soviet investment and capital stock data. Soviet Studies 41(2): 318–329.
29. Kontorovich, V. 2001: Economists, soviet growth slowdown and the collapse. Europe-Asia Studies 53(5): 675–695. | Article |
30. Kushnirsky, F. 2001: A modification of the production function for transition economies reflecting the role of institutional factors. Comparative Economic Studies XLII(1): 1–30.
31. Larson, FD, Butzer, R, Mundlak, Y and Crego, A. 2000: A cross-country database for sector investment and capital. The World Bank Economic Review 14(2): 371–391.
32. Linz, S. 1999: Who is shouldering the burden of transition? An analysis of depreciation rates in Russian industry. Comparative Economic Studies XLI(2–3): 1–44.
33. Melyantsev, V. 2003: Russia's 20th century attempts to embark on modern economic growth. Working Paper, Moscow State University: Moscow.
34. Mezdunarodnye sopostavlenia. 2005: Mezdunarodnye sopostavlenia valovogo vnutrennego produkta za 1999–2000 i 2002 god (International comparisons of gross domestic product in 1999–2000 and 2002). Goskomstat, Moscow.
35. Narodnoye Khozyaistvo SSSR (The Economy of the USSR). 1989: and annual, State committee for Statistics: Moscow.
36. Nove, A. 1981: A note on growth, investment, and price indices. Soviet Studies 33(1): 142–145.
37. OECD. various years: Organization for economic development: OECD databases. OECD: Paris.
38. Oomes, N and Dynnikova, O. 2006: The utilization – adjusted output gap: Is the Russian economy overheating? IMF Working Paper (International Monetary Fund), WP/06/68.
39. Pavlinek, P. 2002: The role of foreign direct investment in the privatisation and the restructuring of the Czech motor industry. Post-Communist Economies 14(3): 359–379. | Article |
40. Popov, V. 1999: Investment, restructuring and performance in transitional economies. Post-Communist Economies 11(4): 445–462. | Article |
41. Popov, V. 2000: Shock therapy versus gradualism: The end of the debate (explaining the magnitude of transformational recession). Comparative Economic Studies XLII(1): 1–58.
42. Popov, V. 2006: Shock therapy versus gradualism reconsidered: Lessons from transition economies after 15 years reforms. Comparative Economic Studies 49(1): 1–31. | Article |
43. Rosefielde, S and Kuboniwa, M. 2003: Russian growth retardation then and now. Eurasian Geography and Economics 44(2): 87–101. | Article |
44. Sohinger, J. 2005: Growth and convergence in European transition economics the impact of foreign direct investment. Eastern European Economics 43(2): 73–94.
45. Souza, M. 2005: Capital stock in Latin America. Economy and Energy IX(50): 18–32.
46. Svejnar, J. 2002: Transition economies: Performance and challenges. Journal of Economic Perspectives 16(1): 3–28. | Article |
47. Tabata, S. 1997: The investment crisis in Russia: A research report. Post-Soviet Geography and Economics 38(9): 558–566.
48. McKinsey Global Institute. 1999: Unlocking economic growth in Russia. McKinsey Global Institute: Moscow.
49. World Bank. 1996: From plan to the market. World Bank: Washington, DC.
50. World Bank. 2003: Russian Economic Report, 2003, Russia Country Department, Economics Unit. World Bank: Washington, DC.
51. World Bank. 2005: World Development Report 2005. World Bank: Washington, DC.
52. World Bank. 2006: Measuring capital for the 21st century: Where is the wealth of nations? World Bank: Washington, DC.
53. World Bank. 2007: World development indicators. World Bank: Washington, DC.

Appendices

APPENDIX A

A recent World Bank (2006) publication, Measuring Capital for the 21st Century, provides capital stock data for several transition economies for the year 2000 compiled using the perpetual inventory method. It includes capital estimates for three countries of the CIS group: Georgia, Moldova and Russia. However, the focus of that study is on the dynamics of various components of broadly defined national wealth in individual countries. To convert national currency-based data on capital accumulation into dollars, it uses market rather than PPP exchange rates (see World Bank, 2006, p. 17). The resulting estimates of fixed capital stock are not, strictly speaking, suitable for international comparisons of K/L ratios. This is illustrated by the data in Table A1, compiled from the World Bank study. From the table, the reported per capita level of capital stock in Georgia is just $595 compared to $4,338 for Moldova and $12,593 for Russia. Taken at face value, these numbers indicate that in Georgia, the capital stock per capita and by implication per worker, was seven times below than that in Moldova. As estimated, it is also lower than that in such countries as Bangladesh ($817), Burkina Fasso ($821) and Ghana ($686). For reference, at the end of the Soviet period, the K/L ratio in Georgia was higher than that in Moldova and only 1.5 times below than that in Russia (Narkhoz SSSR, 1989, pp. 277–278).

Table A1 - Per capita stock of fixed capital and capital–output ratios in three CIS countries in 2000 (capital and GDP valued in US$ at market exchange rates).

Full table

The unsuitability of estimates of capital stock based on market exchange rates for international comparisons can also be illustrated by the computation of capital–output ratios. Using market exchange rates based capital stock and GDP estimates for transition economies generates unrealistically wide disparities in K/Y ratios, from 1.0 in Georgia to 9.0 for Russia and 14.4 for Moldova (see Table A1).

APPENDIX B

Investment price-adjustment coefficients are used to correct the shares of fixed investment provided in the World Development Indicators (2007) database, where these shares are based on nominal prices. Compared to similar shares valued in real terms, nominal shares are overestimated due to relatively faster growth of prices on investment goods and services compared to other components of the GDP. To estimate the degree of overstatement, one has to know by how much prices of investment goods surpass these of all GDP components. The adjustment coefficients are calculated assuming that on average one-third of investment is spent on internationally tradable components (such as equipment), priced at market exchange rate and two-thirds – on domestic components (such as construction), priced at levels close to the general price level for the GDP (Table B1).

Table B1 - Investment price adjustment coefficients for CIS countries, 1992–2005.

Full table

APPENDIX C

Table C1

Table C1 - Fixed capital stock of CIS countries adjusted for market transition (Bil $PPP 2000): 1991–2005.

Full table

APPENDIX D

Table D1

Table D1 - Labour force in CIS countries (million people): 1991–2005.

Full table

APPENDIX E

Table E1

Table E1 - Capital–output (K/Y) ratios in CIS countries, 1992–2005.

Full table