2 Debt Accumulation in the CIS-7 Countries: Bad Luck, Bad Policies, or Bad Advice?

Thomas Helbling; Ashoka Mody; Ratna Sahay

Jump to Content Jump to Main Navigation

Author: Mr. Thomas Helbling, Mr. Ashoka Mody¹, and Ms. Ratna Sahay

¹ 0000000404811396https://isni.org/isni/0000000404811396International Monetary Fund

Language:: English

Keywords:: SEM; CIS-7; country; enterprise; private sector; CIS country; long-term debt; debt share; debt problem; growth projection; debt management strategy; Current account deficits; Current account balance; Current account imbalances; Current account; Baltics; Global

Download PDF (3.4 MB)

Abstract

With the exception of Azerbaijan, which is a net energy exporter, the other low-income CIS countries—Armenia, Georgia, the Kyrgyz Republic, Moldova, Tajikistan, and Uzbekistan—face serious external debt problems.1 From a situation of virtually no debt in 1992, a meteoric increase in debt levels occurred thereafter. In particular, multilateral (IMF and World Bank) lending contributed to the high and increasingly unsustainable levels of debt, despite close monitoring by these institutions undertaken through their conditionality. The CIS-7 experience contrasts with that of other transition economies, which have managed the transition without similar debt accumulation, and is more akin to that of the poorer, highly indebted countries that are heavily reliant on official credit.2

Introduction

With the exception of Azerbaijan, which is a net energy exporter, the other low-income CIS countries—Armenia, Georgia, the Kyrgyz Republic, Moldova, Tajikistan, and Uzbekistan—face serious external debt problems.¹ From a situation of virtually no debt in 1992, a meteoric increase in debt levels occurred thereafter. In particular, multilateral (IMF and World Bank) lending contributed to the high and increasingly unsustainable levels of debt, despite close monitoring by these institutions undertaken through their conditionality. The CIS-7 experience contrasts with that of other transition economies, which have managed the transition without similar debt accumulation, and is more akin to that of the poorer, highly indebted countries that are heavily reliant on official credit.²

What caused this rapid accumulation of debt? Was it bad luck—was their transition especially painful because of initial conditions that proved to be more difficult ex post than expected ex ante and adverse exogenous shocks? Was it bad policies—did the countries fail to implement the reforms necessary to navigate the transition process smoothly? Or, was it bad external advice—did the international financial institutions (IFIs) make matters worse by lending too much based on unrealistic expectations of growth under the policies prescribed?

As a first step, this paper analyzes the structure and the dynamics in the CIS-7 countries’ external debt and, where appropriate, provides comparisons with other transition countries in other parts of the world. The analysis demonstrates that debt accumulation was fundamentally the consequence of large recurring current account deficits combined with an unprecedented output decline in the initial years and its slow recovery thereafter. Debt ratios continued to rise despite the pickup in export growth; the significant foreign direct investment (FDI) inflows (albeit uneven across countries); a decline in average interest rates over time; and, surprisingly, a large real exchange rate appreciation during the transition (notwithstanding the depreciations following the Russian crisis). Although the macroeconomic performance in the CIS-7 countries has improved significantly in recent years, the large buildup of debt is proving to be a drag on their economies, requiring painful adjustment to bring debt back to sustainable levels.

The subsequent section reviews the external adjustment experience under IMF programs and analyzes the reasons for large deviations in outcome relative to projections for the two case studies—the Kyrgyz Republic and Moldova. The case studies indicate that severe problems were encountered in accurately projecting the output of the CIS-7 countries in the midst of large changes, and output growth projections were generally over optimistic. Additional complications arose as historical data were revised downward. In essence, this meant that during the early years when debt sustainability exercises were conducted, true debt ratios were much higher than those measured in the data available at the time.

We then consider, in the section on external adjustment, the factors associated with large recurring current account deficits and the constraints on adjustment. Regression results suggest that the extent of external adjustment (measured by the change in the current account deficit as ratio of GDP) was limited as a result of five factors: large initial current account deficits due to the sudden cutoff of fiscal transfers from Moscow to the CIS-7; the limited extent of concurrent fiscal adjustment; poor GDP growth performance in the initial years; easy availability of multilateral loans on a continuous basis; and adverse terms-of-trade shocks on account of the breakup of Council of Mutual Economic Assistance (CMEA) trade, a sharp increase in energy product prices, the Russian crisis, and—in some countries—internal armed conflict.

Thus, while difficult initial conditions, the inability to undertake fiscal adjustment, and exogenous shocks were responsible for the debt buildup, the availability of financing from the multilateral institutions allowed large current account deficits to continue. The hope was that the World Bank- and IMF-supported reform programs would be successfully implemented. In practice, implementation fell short of programs, especially during the initial years, in part because the countries’ implementation capacity was overestimated. Also, the consequences of the initial forecasting errors have proved to be far-reaching because unanticipated and costly long-term debt problems have emerged. This problem was compounded by the fact that initial GDP estimates were typically revised downward ex post.

In hindsight, multilateral agencies (or any other entity) could not have foreseen the full extent of the disruption that the breakup of the Soviet Union would cause. Did the IFIs make a mistake in continuing to lend, however, even though debt was rising consistently over time? Consumption data indicate that foreign aid helped raise domestic consumption levels above those that would have been otherwise possible. It should be noted, however, that this financing was not sufficient to maintain consumption in real terms, with output falling rapidly and inflation surpassing three or even four-digit levels in the initial years. Consequently, poverty and mortality rates rose at disturbingly high rates. At the same time, IFI financing was not free, which raises the question whether more belt-tightening in the early years by the countries was warranted to prevent the large build-up of debt. In view of the trade-off between maintaining decent living standards and ensuring debt sustainability, a conclusion that could reasonably be drawn in hindsight is that more bilateral donor grants should have been given in the initial years, to be replaced later by IFI financing of investments in reform.

Key Statistics and External Debt Dynamics

Table 2.1 shows current levels of external debt in each of the CIS-7 countries, measured in relation to exports, government revenues, and GDP. Apart from Azerbaijan and Uzbekistan, which are rich in natural resources, external debt in the remaining countries ranges from 55 percent of GDP to nearly 120 percent. These levels are much higher than in other transition countries and are comparable to those low-income developing countries the rest of the world that are heavily or severely indebted.³

Table 2.1.

CIS-7 Countries: Key External Debt Ratios

Sources: IMF, World Economic Outlook database; World Bank, Global Development Finance, 2002.

^¹Exports of goods and services.

^²Median.

^³2000.

Table 2.1.

CIS-7 Countries: Key External Debt Ratios

			External Debt Net Present Value 2000		External Debt Face Value end 2001		Debt Service 2001
			In percent of exports¹	In percent of government revenue	In percent of exports¹	In percent of GDP	In percent of exports¹
Armenia			106.0	189.8	185.4	55.1	5.3
Azerbaijan			44.0	90.6	55.5	22.7	3.7
Georgia			103.8	275.1	203.6	63.4	4.1
Kyrgyz Republic			237.3	550.4	323.5	118.5	19.4
Moldova			104.3	280.1	219.9	90.8	12.9
Tajikistan			117.5	697.3	158.4	107.6	5.3
Uzbekistan			125.5	111.2	144.1	39.9	21.9
Memorandum items:
	All other transition
		countries²	…	…	91.9	43.7	17.2
	Other CIS countries		…	…	97.1	62.3	13.5
	Low-income developing
		countries ³	…	…	181.1	53.3	15.7
		Modestly indebted	…	…	187.0	78.8	12.3
		Severely indebted	…	…	237.9	100.3	19.3
		Heavily indebted	…	…	259.4	101.5	12.3

Sources: IMF, World Economic Outlook database; World Bank, Global Development Finance, 2002.

^¹Exports of goods and services.

^²Median.

^³2000.

Table 2.1.

CIS-7 Countries: Key External Debt Ratios

			External Debt Net Present Value 2000		External Debt Face Value end 2001		Debt Service 2001
			In percent of exports¹	In percent of government revenue	In percent of exports¹	In percent of GDP	In percent of exports¹
Armenia			106.0	189.8	185.4	55.1	5.3
Azerbaijan			44.0	90.6	55.5	22.7	3.7
Georgia			103.8	275.1	203.6	63.4	4.1
Kyrgyz Republic			237.3	550.4	323.5	118.5	19.4
Moldova			104.3	280.1	219.9	90.8	12.9
Tajikistan			117.5	697.3	158.4	107.6	5.3
Uzbekistan			125.5	111.2	144.1	39.9	21.9
Memorandum items:
	All other transition
		countries²	…	…	91.9	43.7	17.2
	Other CIS countries		…	…	97.1	62.3	13.5
	Low-income developing
		countries ³	…	…	181.1	53.3	15.7
		Modestly indebted	…	…	187.0	78.8	12.3
		Severely indebted	…	…	237.9	100.3	19.3
		Heavily indebted	…	…	259.4	101.5	12.3

Sources: IMF, World Economic Outlook database; World Bank, Global Development Finance, 2002.

^¹Exports of goods and services.

^²Median.

^³2000.

Structure, Maturity, and Creditors

To facilitate comparison with other transition countries, comparisons are made in “transition time.”⁴ This expositional device allows us to account for the fact that transition started at different times in the various countries. In figures below, the time t₀ denotes the year in which the transition from planning to market began; for the CIS countries, t₀ was 1992—somewhat later than for Central and Eastern Europe.

Figure 2.1 shows that the CIS-7 external debt is largely debt issued or guaranteed by the public sector. The share of public and publicly guaranteed external debt in total external debt has been higher than the average in other transition economies or in low-income developing countries. While recently declining, the share has, remained on average above 80 percent (in 2000), which is comparable to the share in low-income developing countries that are heavily indebted. Among the CIS-7 countries, Azerbaijan, the Kyrgyz Republic, and Tajikistan stand out with shares below average, which largely reflects higher shares of privately owned debt related to export credits and foreign direct investment.

Second, over time the share of official concessional external financing (based on new commitments) in CIS-7 countries has reached levels observed in low-income developing countries (above 40 percent).⁵ Initially, however, the share of this type of financing was small, because the CIS-7 countries’ external debt reflected primarily energy import-related debt to Russia and Turkmenistan on commercial terms. Despite the recent increase, the share of official concessional debt remains below those in low-income developing countries that are heavily indebted, even though the external debt ratios in the two groups of countries are comparable.

Third, the CIS-7 external debt is mostly long term in nature. Over time, the maturities of new external financing commitments have exceeded 30 years on average, comparable to those in low-income developing countries that are heavily indebted and above those in other low-income developing and transition countries.

Fourth, the average interest rate of new debt commitments in the CIS-7 countries has fallen over time to about 1 percent, comparable to rates found in low-income developing countries that are heavily indebted. Average interest rates in low-income developing and other transition countries are substantially higher.

Fifth, the CIS-7 countries owe more to the IMF and the World Bank than other transition countries or low-income developing countries (Figure 2.2). Relative to the CIS average, Tajikistan and Uzbekistan, where reform efforts have been lagging, have relatively smaller shares of debt owed to the IMF and the World Bank, whereas faster reformers—especially Armenia and the Kyrgyz Republic—have higher shares

The CIS-7 debt shares owed to other IFIs also total more than those of other transition countries, but are comparable to those of low-income developing countries. Similarly, CIS-7 countries owe a larger share to bilateral official creditors than do other transition countries but less than low-income developing countries. In contrast, the share of debt owed to private creditors in the CIS-7 is comparable to that of low-income developing countries but falls short of that of other transition countries. The most notable change over time in the structure of the CIS-7 countries’ external debt is the shift from debt owed to bilateral official creditors to that owed multilateral official creditors and, to a smaller degree, to that owed private creditors. This pattern reflects the declining dependence on other CIS countries, notably for energy imports.

In sum, the debt structure of the CIS-7 economies has become similar over time to that of other low-income countries. While initially the debt profile of CIS-7 countries was similar to that of other transition economies, a notable difference has emerged: namely, a much higher share of public and publicly guaranteed debt, indicating that private sector entities in the CIS-7 economies have substantially lower access to international capital markets than those in other transition economies.

Decomposing External Debt Dynamics

We decompose the changes in the CIS-7 countries’ external debt into the main contributing factors (see Appendix for technical details). Balance of payments identities imply that the change in the stock of external debt between any two periods must equal the sum of the trade of goods and services; the transfer balance (these two items together add up to the primary external current account balance); interest payments on existing external debt and the change in foreign exchange reserves (and other assets held by residents abroad), minus the non-debt creating net capital. Since, FDI flows are the most important non-debt creating flows, they are shown separately in the figures below.

It is customary to use the ratio of external debt to GDP rather than the debt stock in nominal terms to assess the burden that the external debt can impose on the economy as a whole because some sectors have significant revenues in domestic currency (e.g., the government). When considering ratios, factors contributing to the changes in the denominator (e.g., the domestic GDP in U.S. dollar terms) also need to be taken into account. In addition to real GDP growth in domestic currency terms, changes in the real exchange rate can also potentially alleviate or aggravate the debt burden.⁶ If the real exchange rate appreciates, the external debt burden declines, and vice versa.

Figure 2.3 illustrates the debt-to-GDP decompositions. They are based on annual data on transition time. All contributing flows are shown as cumulative flows from transition time t₁ onward, showing the total contribution from the beginning to that point in time. The primary current account balance in time t₅, for example, would be the sum of primary current account balances from t₁ to t₅, The obvious exception to this timing convention concerns the changes in the debt-to-export ratios, which are based on the end-of-period debt ratios in t₀.

Large primary current account deficits are the single most important factor contributing to the rise in CIS-7 external debt. The cumulative sum of the primary current account is, on average, about seven times as large as in other transition countries. Also, as expected, interest payments have been less significant for the CIS-7 countries than for other transition countries because a substantially larger share of CIS-7 debt is on concessional terms, while other transition countries have borrowed more from the private sector on commercial terms. FDI inflows and high export growth have mitigated the debt problem in the CIS-7 countries as well as in other transition economies. Interestingly, the relative significance of FDI flows has been greater in the CIS-7 countries than in the other transition countries, although export growth has been much higher in the latter.⁷ In both country groups, the average contribution of real GDP growth to reducing the debt burden is similar.

Perhaps, the most surprising result of this exercise is the positive contribution of the real appreciation of the CIS-7 currencies against the U.S. dollar in reducing the debt burden. On average, this effect turned out to be half as large as the absolute value of the cumulative current account or about 25 percentage points of GDP. Even the depreciations of the CIS-7 currencies against the ruble and other currencies in late-1998 and 1999 do not appear to have made much of a difference.

External Adjustment: Programs and Outcomes

Having established the primary importance of current account deficits in contributing to the external debt burden, we now explore whether such large deficits either were or could have been predicted. Since virtually all countries had IMF-supported programs, projections and actual performance of macroeconomic indicators provide potentially useful insights into what went wrong. Given that CIS-7 countries began with large imbalances in their fiscal and external accounts, medium-term projections in IMF-supported programs typically reflected large adjustments to ensure sustainable debt paths.

Despite large initial current account deficits in the CIS-7, there seemed to be good reasons ex ante to believe that a combination of financing and adjustment policies would help achieve medium-term debt sustainability, given the two mutually reinforcing factors. First, permanent productivity increases could be expected once many of the pre-existing distortions of the communist system were removed and market-oriented reforms were adopted.⁸ Positive productivity shocks, especially to the external sector, initially can generate current account deficits.⁹ Over time, however, productivity growth allows the deficits to gradually turn into surpluses, which can then service the external debt. Second, external aid was planned not only to finance investments in reforms but also to allow for gradual reduction of the external imbalances so as to avoid disruptive adjustment and large social costs. As transition proceeded, however, events did not quite shape up as planned.

To maximize the period covered under IMF-supported programs, case studies are presented on the first two countries that entered such a program, the Kyrgyz Republic and Moldova, to illustrate what went wrong. Since our main variable of interest—external debt-to-GDP ratio—depends as much on the evolution of GDP as on external debt in absolute terms, we looked at IMF-supported program projections on both variables. Errors in GDP growth projections in these countries turned out to be the key to understanding the evolution of the debt-GDP ratios. The extent of output collapse in the early years of transition was simply not anticipated.¹⁰ A forthcoming working paper includes a case study on Georgia and provides a more extensive analysis of the GDP growth projections.¹¹

Case Studies

The Kyrgyz Republic

The Kyrgyz Republic has perhaps the most acute external debt problem, at nearly 120 percent of GDP. Figures 2.4 and 2.5 clearly illustrate the policy planning problems of the early years. The projected and actual medium-term paths are presented for the external current account and the external debt (in U.S. dollars and in percent of GDP) under various annual programs approved during 1993–98. As is typical in the analyses if IMF-supported programs, all projected paths start in the year in which the program was approved and are reported for five- to six-year periods ahead.

Consistent with the expectation that productivity would rise rapidly, the first program, the 1993 Stand-by Arrangement, predicted a sharp initial increase in the current account deficit and a rapid decline thereafter. The external debt was expected to stabilize quickly at around 45 percent of GDP in this scenario. Interestingly, during the first two years of the program (t₁ and t₂), the actual current account deficits turned out to be below projections. The difference was especially large in U.S. dollar terms while that in percentage points of GDP was smaller, an indication that growth fared worse than expected.

Given the better-than-expected outcomes in the first two years of the first program, the paths for the external current account deficits envisaged under subsequent programs were revised from the V-like shapes to “stretched U-shapes.” Thus, the projected external current account deficits projected in later programs were not anticipated to increase to the maximum levels envisaged under the earlier Stand-by Arrangement but were expected to remain large for a longer time span and to decrease only very gradually. Despite large deficits for a longer period of time, the external debt was expected to increase only gradually and converge to about 45 percent of GDP, reflecting among other things sustained anticipation of rapid GDP growth in U.S. dollar terms.

The deviations between program projections and actuals occurred under subsequent programs and reflected the unexpected deterioration in the external current accounts in 1996 (t₄) and in 1998 (t₆), the year of the Russian crisis. These large shocks do not seem to have led to a call for more ex ante adjustment, and projected external current account deficits remained in the range of 5 percent of GDP. In program documents, the deterioration in external debt ratios was noted but for a long time was not considered a problem, and references were made to the large share of concessional funds in the overall external financing received.

As new programs were initiated and new statistical systems were set up, large discrepancies in the measurement of GDP began to emerge. Comparing early program documents with later ones suggests that, at some point, the U.S. dollar GDP in 1992—the initial year of the transition—was revised downward belatedly by about two-thirds of the value used in the early projections. This revision alone implied ex post increases of about 40 percentage points in the debt-to-GDP ratio at the end of the projection period under the first Stand-by Arrangement.

Overall, the graphical analysis illustrates how the combination of programmed large external current account deficits, initial growth optimism, and subsequent revelations of statistical overstatement of GDP turned out to be a lethal mix for misjudging the rapid increase in external debt ratios. When large external current account deficits were planned, the debt dynamics became more vulnerable to unexpected deviations from the projections. External debt, manageable at about 40 percent of GDP at the end of t₄ basically worsened during a period of three years (from t₅ to t₇).

Moldova

The first IMF-supported program with Moldova began in late 1993, only a few months later than that of the Kyrgyz Republic. Figures 2.6 and 2.7 show the comparisons between projected and actual medium-term paths for the external current account and the external debt under various annual programs approved during 1993 to 1998. The projected adjustment in the external current account deficits (as a percent of GDP) over the medium term appears somewhat more ambitious than in the case of the Kyrgyz Republic. This is reflected in declining debt-to-GDP ratio projections. Initially, during t₀ to t₃, the actual external current account balance in U.S. dollar terms performed noticeably better than anticipated, which is also reflected in the external debt path. In percent of GDP, the performance is only better in some of these early years because as GDP in U.S. dollar terms was overpredicted. The latter also explains the overshooting of the actual external debt as percent of GDP in t₃.

Starting in t₄, the actual current account began to deteriorate rapidly compared to program targets, both in U.S. dollar terms and as percent of GDP—a process that culminated in a forced, abrupt adjustment after the Russian crisis in t₆. Correspondingly, the external debt began to increase rapidly and to exceed program projections by rising margins, although the deterioration in the actual debt dynamics as compared to the projected one, also reflects the unanticipated sharp real depreciation after the Russian crisis. While program projections suggest that the deterioration in the external current account was to be reversed quickly, policy measures were insufficient, and the sequence of adverse shocks had larger effects than anticipated.

As in the case of the Kyrgyz Republic, the Moldovan experience shows how the combination of slow planned adjustment in large initial current account deficits (as percent of GDP), large shocks to the external current account, overly optimistic growth projections, and mismeasurement of data led to a surge in the external debt to unsustainable levels within a period of four years. This problem was compounded, particularly in the Moldovan case, by the sharp real exchange rate depreciation that followed the 1998 Russian crisis.

Decomposing the Forecast Errors in External Debt Ratios

A more systematic analysis of the forecast errors in the external debt-to-GDP ratios leads to similar conclusions (Table 2.2). A time span of four years for the calculation of errors seems to be an acceptable compromise between a medium-term forecast horizon and a reasonable number of forecast errors, given the short sample sizes.¹² The errors reported in the table are the differences between the debt-to-GDP ratio four years ahead, predicted at the time of program approval and the actual debt-to-GDP ratio four years later. Given the differences in reporting across time and countries, a few simplifying assumptions had to be made to allow for a unified methodology (see Appendix).

Table 2.2.

External Debt Ratios: Four-year Forecast Error Decomposition for Two CIS-7 Countries

Sources: Staff calculations based on Staff reports and WEO database.

^¹P¹, P², and P³denote the respective annual programs in ESAF/PRGF arrangements while R1, R2, and R3 denote respective reviews under Extended Fund Facilities (EFF).

^²Date of Executive Board Approval of arrangement or annual program, respectively.

^³The fourth year of the projection horizon.

^⁴ Number of years covered by the projections

^⁵First-order approximation of forecast error in debt-to-GDP ratio. There is an unexplained residual error which is due to second-order errors, which can be large in case of large errors to components. A positive sign means an overprediction for variables that are positive (underprediction for negative variables such, as deficits).

Table 2.2.

External Debt Ratios: Four-year Forecast Error Decomposition for Two CIS-7 Countries

			Kyrgyz Republic					Moldova
							ESAF/
Arrangement			SBA	ESAF			PRGF	SBA	EFF
Annual
Program				P1¹	P2¹	P3¹	P1¹			R2	R3
Approval (first year of projection)²			Apr-93	Jun-94	Nov-95	Mar-97	Jun-98	Dec-93	May-96	Jun-97	Dec-98
Last year of projection³			1996	1997	1998	2000	2001	1996	1999	2000	2001
Projection span (years)⁴			4	4	4	4	4	4	4	4	4
External Debt
	(percent of GDP; end of period)
		Projection	47	36	40	92	85	37	24	40	81
		Outcome	63	76	95	126	119	59	111	108	91
		Forecast error (percentage points)	16.5	40.1	55.2	33.8	33.8	21.8	87	68	10
Attributable to forecast errors in
	percentage points of GDP⁵
	Cumulative flows in U.S. dollars		-1.2	9.0	17.3	-6.0	-3.2	19.2	9.4	0.6	-2.1
		External current account	-13.9	-1.4	18.6	2.0	14.2	-13.4	0.6	6.6	-13.2
		FDI	-4.4	-4.3	-1.7	-5.3	-4.4	3.9	-2.4	0.4	-1.3
		Other (inc. residual)	17.1	14.8	0.4	-2.6	-14.9	28.8	11.2	-6.5	12.4
	Cum. flows due to GDP forecast error		13.5	10.8	9.8	11.2	8.0	2.1	4.6	6.9	0.4
	Revisions to initial debt in U.S. dollars		-1.0	0.4	16.2	—	—	-0.1	10.2	15.9	6.1
	Stock effects due to GDP forecast error		0.4	10.4	16.2	21.0	27.5	0.0	23.6	22.9	2.7

Sources: Staff calculations based on Staff reports and WEO database.

^¹P¹, P², and P³denote the respective annual programs in ESAF/PRGF arrangements while R1, R2, and R3 denote respective reviews under Extended Fund Facilities (EFF).

^²Date of Executive Board Approval of arrangement or annual program, respectively.

^³The fourth year of the projection horizon.

^⁴ Number of years covered by the projections

Table 2.2.

External Debt Ratios: Four-year Forecast Error Decomposition for Two CIS-7 Countries

			Kyrgyz Republic					Moldova
							ESAF/
Arrangement			SBA	ESAF			PRGF	SBA	EFF
Annual
Program				P1¹	P2¹	P3¹	P1¹			R2	R3
Approval (first year of projection)²			Apr-93	Jun-94	Nov-95	Mar-97	Jun-98	Dec-93	May-96	Jun-97	Dec-98
Last year of projection³			1996	1997	1998	2000	2001	1996	1999	2000	2001
Projection span (years)⁴			4	4	4	4	4	4	4	4	4
External Debt
	(percent of GDP; end of period)
		Projection	47	36	40	92	85	37	24	40	81
		Outcome	63	76	95	126	119	59	111	108	91
		Forecast error (percentage points)	16.5	40.1	55.2	33.8	33.8	21.8	87	68	10
Attributable to forecast errors in
	percentage points of GDP⁵
	Cumulative flows in U.S. dollars		-1.2	9.0	17.3	-6.0	-3.2	19.2	9.4	0.6	-2.1
		External current account	-13.9	-1.4	18.6	2.0	14.2	-13.4	0.6	6.6	-13.2
		FDI	-4.4	-4.3	-1.7	-5.3	-4.4	3.9	-2.4	0.4	-1.3
		Other (inc. residual)	17.1	14.8	0.4	-2.6	-14.9	28.8	11.2	-6.5	12.4
	Cum. flows due to GDP forecast error		13.5	10.8	9.8	11.2	8.0	2.1	4.6	6.9	0.4
	Revisions to initial debt in U.S. dollars		-1.0	0.4	16.2	—	—	-0.1	10.2	15.9	6.1
	Stock effects due to GDP forecast error		0.4	10.4	16.2	21.0	27.5	0.0	23.6	22.9	2.7

Sources: Staff calculations based on Staff reports and WEO database.

^¹P¹, P², and P³denote the respective annual programs in ESAF/PRGF arrangements while R1, R2, and R3 denote respective reviews under Extended Fund Facilities (EFF).

^²Date of Executive Board Approval of arrangement or annual program, respectively.

^³The fourth year of the projection horizon.

^⁴ Number of years covered by the projections

The results of the forecast error decomposition suggest that the external debt-to-GDP ratio—or, alternatively, the cumulative current account deficit to GDP ratio—was typically under-predicted. It is clear from the tables that with GDP in U.S. dollar terms being over-predicted, economic growth was not large enough to reduce the burden of already accumulated debt in percent of GDP as planned and, at the same time, allow for continued large (planned) deficits with little or no effect on debt ratios. This effect of the GDP overprediction accounts for a substantial portion of today’s debt problems. In addition, revisions to the initial debt levels, on which program projections were built, contributed positively to the forecast errors ex post. The forecast errors in GDP growth in U.S. dollar terms were not only due to errors in predicting real GDP growth. Errors in predicting the real exchange rate and statistical uncertainty about initial GDP levels appear to have been at least as important if not more important.¹³ In sum, the CIS-7 debt management strategy which allowed for initially large external imbalances that were expected to turn around as growth picked up rapidly, was undermined due to large forecasting errors.¹⁴

External Adjustment: A Comparison with Other Transition Economies

A comparison of CIS-7 countries with other transition countries indicates that in more recent years the difference in the level of external current account deficits between the two groups has narrowed markedly. In particular, in the tenth year of the transition (t₉ in Figure 2.8), the external imbalances in both groups were virtually the same. The dynamic path of the current accounts differed markedly, however. The CIS-7 countries started the transition with large current account deficits that persisted for about six years (up to the Russian crisis) and began to adjust rather quickly thereafter. On average, other transition countries started with current account surpluses that slowly eroded and turned into deficits. Given these differences in the adjustment dynamics between the CIS-7 and other transition countries, this section examines the adjustment record over the period t₀ to t₉, which corresponds to the years 1992–2001 for the CIS-7 countries, and finds a further breakdown into transition phases to be an important part of the story. Specifically, we refer to the years t₀ to t₁ as the initial phase (or initial conditions), the years t₂ to t₅ as the early phase, and the years t₆ to t₉ as the later transition phase.

What Explains the Difference in the Adjustment Record? Five Hypotheses

Initial Conditions and Distortions

All the transition economies (including the CIS-7) inherited a system of distorted relative prices, state ownership of productive capital, and a large dependency on CMEA trade; however, the degree of distortions, economic structures, and patterns of specialization varied widely across the countries. The differences in initial conditions were important determinants of varying inflation and growth performances during the transition.¹⁵ The unraveling of heavy economic dependence on the Soviet system, both via CMEA trade as well as via a complex system of tax and transfers with the center, appears to have contributed to the deeper and more prolonged output decline in the CIS-7 (Figure 2.9). In addition, these countries, most of which have been net energy importers, had to cope with a very large initial terms-of-trade shock, as energy prices rose to commercial terms overnight. The higher initial distortions in the CIS-7 countries as compared to other transition economies were reflected in relatively larger external and fiscal imbalances in the initial years, as shown in Figure 2.8. Moreover, these initial deficits were also associated with smaller adjustments later, as shown by the negative correlations with changes in the deficits in subsequent years (Table 2.3).

Table 2.3.

Transition Countries: Sample Correlations in Transition Time

(Marginal significance levels in parenthesis)

Source: Staff calculations.

^¹In percent of GDP.

^²Change in average during four-year period compared to average in t₀-t₁ or the previous four-year period.

^³Change in external debt at end of four-year period compared with value al end of t₆ or the end of the previous four-year period.

Table 2.3.

Transition Countries: Sample Correlations in Transition Time

(Marginal significance levels in parenthesis)

		External Current Account Balance¹
		Average			Adjustment²
		t₂-t₉	t₂-t₅	t₆-t₉	t₂-t₅	t₆-t₉
External current account balance¹
	Average t₀-t₁	0.52	0.75	—	-0.95	—
		(0.01)	(0.00)		(0.00)
	Average t₂-t₉	—	—	—	—	—
	Average t₂-t₅	—	—	0.43	—	-0.60
				(0.04)		(0.00)
	Average t₆-t₉	—	—	—	—	—
	Adjustment t₂-t₅	—	—	—	—	—
	Adjustment t₆-t₉	—	—	—	—	—
Real GPD growth
	Average t₂-t₉	-0.28	—	—	—	—
		(0.18)
	Average t₂-t₅	—	-0.10	—	0.20	—
			(0.63)		(0.36)
	Average t₆-t₉	—	—	-0.28	—	-0.40
				(0.19)		(0.05)
Multilateral Disbursements
	Average t₀-t₁	-0.39	-0.34	—	0.13	—
		(0.07)	(0.12)		(0.54)
	Average t₂-t₉	-0.57	—	—	—	—
		(0.00)
	Average t₂-t₅	—	-0.61	—	0.42	—
			(0.0)		(0.04)
	Average t₆-t₉	—	—	-0.25	—	0.30
				(0.24)		(0.16)
	Adjustment t₂-t₅	—	—	—	0.35	—
					(0.11)
	Adjustment t₆-t₉	—	—	—	—	0.01
						(0.93)

General Government Balance¹					External debt²
Average			Adjustment²		Change³
t₂-t₉	t₂-t₅	t₆-t₉	t₂-t₅	t₆-t₉	t₂-t₅	t₆-t₉
0.58	0.57	—	-0.43	—	0.07	—
(0.00)	(0.00)		(0.04)		(0.74)
0.46	—	—	—	—	—	—
(0.03)
—	0.49	0.34	—	-0.41	—	0.16
	(0.02)	(0.11)		(0.05)		(0.48)
—	—	0.29	—	—
		(0.16)
—	—	—	0.54	—
			(0.01)
—	—	—	—	0.30
-0.19	—	—	—	—	—	—
(0.36)
-0.12	—	-0.14	—	-0.12	—
(0.57)		(0.53)		(0.57)
—	—	0.20	—	0.17	—	-0.34
		(0.36)		(0.43)		(0.10)
—	—	—
—	—	—

Source: Staff calculations.

^¹In percent of GDP.

^²Change in average during four-year period compared to average in t₀-t₁ or the previous four-year period.

^³Change in external debt at end of four-year period compared with value al end of t₆ or the end of the previous four-year period.

Table 2.3.

Transition Countries: Sample Correlations in Transition Time

(Marginal significance levels in parenthesis)

		External Current Account Balance¹
		Average			Adjustment²
		t₂-t₉	t₂-t₅	t₆-t₉	t₂-t₅	t₆-t₉
External current account balance¹
	Average t₀-t₁	0.52	0.75	—	-0.95	—
		(0.01)	(0.00)		(0.00)
	Average t₂-t₉	—	—	—	—	—
	Average t₂-t₅	—	—	0.43	—	-0.60
				(0.04)		(0.00)
	Average t₆-t₉	—	—	—	—	—
	Adjustment t₂-t₅	—	—	—	—	—
	Adjustment t₆-t₉	—	—	—	—	—
Real GPD growth
	Average t₂-t₉	-0.28	—	—	—	—
		(0.18)
	Average t₂-t₅	—	-0.10	—	0.20	—
			(0.63)		(0.36)
	Average t₆-t₉	—	—	-0.28	—	-0.40
				(0.19)		(0.05)
Multilateral Disbursements
	Average t₀-t₁	-0.39	-0.34	—	0.13	—
		(0.07)	(0.12)		(0.54)
	Average t₂-t₉	-0.57	—	—	—	—
		(0.00)
	Average t₂-t₅	—	-0.61	—	0.42	—
			(0.0)		(0.04)
	Average t₆-t₉	—	—	-0.25	—	0.30
				(0.24)		(0.16)
	Adjustment t₂-t₅	—	—	—	0.35	—
					(0.11)
	Adjustment t₆-t₉	—	—	—	—	0.01
						(0.93)

General Government Balance¹					External debt²
Average			Adjustment²		Change³
t₂-t₉	t₂-t₅	t₆-t₉	t₂-t₅	t₆-t₉	t₂-t₅	t₆-t₉
0.58	0.57	—	-0.43	—	0.07	—
(0.00)	(0.00)		(0.04)		(0.74)
0.46	—	—	—	—	—	—
(0.03)
—	0.49	0.34	—	-0.41	—	0.16
	(0.02)	(0.11)		(0.05)		(0.48)
—	—	0.29	—	—
		(0.16)
—	—	—	0.54	—
			(0.01)
—	—	—	—	0.30
-0.19	—	—	—	—	—	—
(0.36)
-0.12	—	-0.14	—	-0.12	—
(0.57)		(0.53)		(0.57)
—	—	0.20	—	0.17	—	-0.34
		(0.36)		(0.43)		(0.10)
—	—	—
—	—	—

Source: Staff calculations.

^¹In percent of GDP.

^²Change in average during four-year period compared to average in t₀-t₁ or the previous four-year period.

^³Change in external debt at end of four-year period compared with value al end of t₆ or the end of the previous four-year period.

The Transition

All transition countries faced bursts of inflation; volatility in relative prices; large-scale changes in economic structure; and losses of subsidies and transfers, especially those related to energy consumption nonetheless, the CIS-7 countries experienced the most difficult transition challenges. Debt began to accumulate rapidly as subsidies and transfers from the center came to a halt, and near market prices for energy-related products began to be paid. The downward spiral in output performance bad important implications for adjustment—it reduced the present value of the stream of future taxes and raised the net present value of programmed expenditure.¹⁶ Accordingly, fiscal policy was more expansionary, which, in turn, explained the larger-than-projected external current account imbalances.

Large Exogenous Shocks During the Transition

The Russian crisis had a significant effect on external current account balances and external debt profiles.¹⁷ External demand and the terms of trade worsened in the CIS-7 (Figure 2.9). Some countries like Georgia and Armenia experienced armed internal conflicts.

Policy Performance

For policymakers, policy planning involved significant uncertainties, as both transition paths and steady states were largely unknown. Notwithstanding these problems, it is clear in hindsight that there was insufficient adjustment of policies, particularly at the initial stages of transition. Both stabilization and structural reform policies were not sufficiently ambitious not only in achievements but also in targets. It took a long time for credit to tighten and for the economies to stabilize from high inflation levels. Fiscal policy was also expansionary in the initial years. Interestingly, fiscal consolidation in the later years did not translate into concurrent external adjustment (Figure 2.8). As regards structural reforms, the pace was much slower than other transition countries (Table 2.4). In particular, the slow liberalization of the external regime is also likely to have hindered export growth initially (Figure 2.9).

Table 2.4.

Transition Countries: Ranking of Structural Reform Outcomes

(Ranking based on end-of-period levels in structural reform index)

Sources: EBRD, Transition Report, various issues.

^¹The overall EBRD structural reform index is the unweighted average sectoral reform indices (data availability for some of the indices varies over time and country).

^²Data for T₁ rather titan T₀.

Table 2.4.

Transition Countries: Ranking of Structural Reform Outcomes

(Ranking based on end-of-period levels in structural reform index)

		Overall EBRD Structural Reform Index¹
	Initial Reform Level at T0₀	End of Early Transition Period (T₅)	End of Second Transition Period (T₉
Eastern European Countries and the Baltics
Albania		21	14	17
Bulgaria		10	19	8
Croatia²		7	11	8
Czech Republic		3	2	3
Estonia		7	3	2
Hungary²		2	1	1
Latvia		5	6	7
Lithuania		11	7	5
Macedonia,
	Former Yugoslav
	Rep. of²	7	15	16
Poland²		1	4	3
Romania		18	18	10
Slovak Republic		3	5	6
Commonwealth of Independent States
CIS-7
Armenia		13	16	12
Azerbaijan		21	21	20
Georgia		20	12	11
Kyrgyz Republic		12	9	15
Moldova		14	13	12
Tajikistan		15	22	21
Uzbekistan		21	20	22
Others
Belarus		15	23	23
Kazakhstan		15	10	12
Russia		6	8	17
Turkmenistan		24	24	24
Ukraine		18	17	19

Sources: EBRD, Transition Report, various issues.

^¹The overall EBRD structural reform index is the unweighted average sectoral reform indices (data availability for some of the indices varies over time and country).

^²Data for T₁ rather titan T₀.

Table 2.4.

Transition Countries: Ranking of Structural Reform Outcomes

(Ranking based on end-of-period levels in structural reform index)

		Overall EBRD Structural Reform Index¹
	Initial Reform Level at T0₀	End of Early Transition Period (T₅)	End of Second Transition Period (T₉
Eastern European Countries and the Baltics
Albania		21	14	17
Bulgaria		10	19	8
Croatia²		7	11	8
Czech Republic		3	2	3
Estonia		7	3	2
Hungary²		2	1	1
Latvia		5	6	7
Lithuania		11	7	5
Macedonia,
	Former Yugoslav
	Rep. of²	7	15	16
Poland²		1	4	3
Romania		18	18	10
Slovak Republic		3	5	6
Commonwealth of Independent States
CIS-7
Armenia		13	16	12
Azerbaijan		21	21	20
Georgia		20	12	11
Kyrgyz Republic		12	9	15
Moldova		14	13	12
Tajikistan		15	22	21
Uzbekistan		21	20	22
Others
Belarus		15	23	23
Kazakhstan		15	10	12
Russia		6	8	17
Turkmenistan		24	24	24
Ukraine		18	17	19

Sources: EBRD, Transition Report, various issues.

^¹The overall EBRD structural reform index is the unweighted average sectoral reform indices (data availability for some of the indices varies over time and country).

^²Data for T₁ rather titan T₀.

Donor Financing and Overoptimism

Donor financing, though gradually declining over time, has been relatively high as a share of GDP Apart from initial energy-related loans by Russia and Turkmenistan that were closer to commercial terms, financing during the transition has generally been on concessional terms. It can be argued that official financing may have contributed to the present debt problems because it allowed for delaying of the needed adjustment. Figure 2.10 illustrates how official financing, especially by multilateral financial institutions, accommodated the external imbalances of the CIS-7 countries. Sample correlation coefficients support the following interpretation: higher multilateral disbursements are associated with larger external imbalances. The seemingly generous multilateral loans were, to some extent, the result of lower-than-expected growth, which made the related capital inflows more expansionary than originally envisaged. Simple correlation coefficients also suggest that, while official external financing accommodated large external imbalances, it also supported relatively stronger adjustment in the early stages of the transition (Table 2.3). Adjustment (the change in the external current account balance) was larger in countries that had higher official external financing both in levels and in terms of change.

External Adjustment: Econometric Evidence

Multivariate cross-section regressions for 25 transition economies were carried out to examine the relative importance of the factors listed above and, thereby, explain the differences in adjustment patterns among transition economies. External adjustment is measured by the change in the primary external current account (in percent of GDP) in the estimated equation below:

\begin{matrix} Δ {\bar{c a}}_{t, t + 3} = α + β_{1} {\bar{c a}}_{i n i} + β_{2} Δ {\bar{g b}}_{t, t + 3} + β_{3} Δ Y_{t, t + 3} + β_{4} {\bar{o f}}_{t, t + 3} + β_{5} Δ {X^{*}}_{t, t + 3} \\ + β_{6} Δ {\bar{T T}}_{t, t + 3} + β_{7} Δ {\bar{P L}}_{t, t + 4} + β_{8} Δ {\bar{F T L}}_{t, t + 3} + ɛ_{t, t + 3} . & (1) \end{matrix}

Variables in small letters are in percent of GDP, a Δ in front of a variable indicates a change against the previous period, and a bar over a variable denotes an average over a transition phase. The combination of a Δ and a capital letter variable indicates a percentage change. To illustrate the notation: Δ¯ca_{t, t+3} denotes the change in the average external current account balance as a percent of GDP in the phase starting in t and ending in t+3 compared to the average during the previous phase. Thus, the goal is to explain the change in the primary external current account balance (ca) by relating that change to the initial primary external current account (ca_ini) the general government balance (gb), the level of GDP (Y), disbursements by multilateral financial institutions (of), foreign demand (exports, X*), the terms of trade, the EBRD index of domestic price liberalization (PL), and the EBRD index of foreign exchange and trade liberalization (FTL).

The rationale behind the specification closely follows the hypotheses in the previous section: most variables are measured as changes because we are interested in adjustment. The initial current account imbalance is measured in levels to determine if adjustment was more difficult as the initial imbalances increased. Similarly, the disbursements by multilateral financial institutions is measured in levels as well, given the conjecture that higher donor financing may have discouraged adjustment.

Table 2.5 reports the result for the early transition phase, which is most relevant for this discussion. We report the results only for the early transition years because our main interest is in understanding how debt built up so quickly at the start of the transition. The first column shows the full equation as specified above. The second column is a parsimonious reduction and only includes variables that turned out to be significant at the 10 percent level. To be precise, we tested for the joint exclusion of all variables that did not meet the 10 percent benchmark significance levels before excluding them. Standard errors are heteroskedasticity consistent. The final column uses a weighted absolute distance estimator to check the robustness of the results with regard to outliers in our relatively small sample. The results suggest that, while the magnitudes of some coefficients vary with the estimator, the qualitative implications are generally robust, except for the foreign demand variable, which turns insignificant with the robust estimator.

Table 2.5.

Determinants of External Adjustment in Transition Countries¹

(Standard errors in parentheses)

Source: Staff calculations.

^¹See text for details. Constant is not reported.

Table 2.5.

Determinants of External Adjustment in Transition Countries¹

(Standard errors in parentheses)

Dependent Variable
Estimator
Explanatory
Variables	OLS	OLS	Robust
	-0.66	-0.64	-0.79
	(0.087)	(0.060)	(0.089)
	0.37	0.40	0.34
	(0.132)	(0.100)	(0.129)
	-0.91	-1.04	-0.77
	(0.524)	(0.547)	(0.423)
	0.16	0.12	0.11
	(0.051)	(0.051)	(0.052)
	5.32	7.87
	(2.955)	(3.461)
	175.74	156.17	106.65
	(63.363)	(59.008)	(48.935)
	-3.791
	(4.403)
	-1.021
	(0.901)
	0.939	0.936
	2.699	2.872	2.766

Source: Staff calculations.

^¹See text for details. Constant is not reported.

Table 2.5.

Determinants of External Adjustment in Transition Countries¹

(Standard errors in parentheses)

Dependent Variable
Estimator
Explanatory
Variables	OLS	OLS	Robust
	-0.66	-0.64	-0.79
	(0.087)	(0.060)	(0.089)
	0.37	0.40	0.34
	(0.132)	(0.100)	(0.129)
	-0.91	-1.04	-0.77
	(0.524)	(0.547)	(0.423)
	0.16	0.12	0.11
	(0.051)	(0.051)	(0.052)
	5.32	7.87
	(2.955)	(3.461)
	175.74	156.17	106.65
	(63.363)	(59.008)	(48.935)
	-3.791
	(4.403)
	-1.021
	(0.901)
	0.939	0.936
	2.699	2.872	2.766

Source: Staff calculations.

^¹See text for details. Constant is not reported.

The results suggest that initial conditions (captured by the initial primary current account imbalances), fiscal imbalances, official financing, terms of trade, and foreign demand (exports) were associated with external adjustment during the early transition phase, albeit to varying degrees. All the coefficients are significant, and their signs are consistent with our priors. While fiscal imbalances are reflected in external imbalances, the relationship is not strictly proportional. The level of disbursements by multilateral financial institutions had a negative effect on the external adjustment during the early transition phase, as conjectured, although this effect is only significant at the 10 percent level. The coefficient on the GDP is positive and significant, suggesting that output declines lowered the external adjustment above and beyond the fiscal channel. Interestingly, the structural reform indices (domestic and external liberalization) turned out to be insignificant, indicating that although structural reforms proceeded slower than anticipated, a slow pace was not the main constraint on external adjustment. Overall, the regressors explain more than 90 percent of the variation in the change of the average external current account balance.¹⁸

What Has the Debt Accumulation Financed?

From a policy perspective, it is critical to know whether the accumulation of external debt financed consumption or investment, particularly investments in reforms. If it is the latter, the outlook for the debt burden and sustainability in the medium term would not be quite as worrisome.

Given that the CIS-7 economies started out with suppressed consumption levels, an initial jump in consumption at the early stages of transition was to be expected—this occurred also in the better performers of Eastern Europe at the start of their transition.¹⁹ Such a jump would also be consistent with the permanent productivity increase hypothesis discussed above because permanent income increases ahead of actual income. Indeed, consumption in the CIS-7 countries in the initial phase increased sharply (Figure 2.11).²⁰ The fact that real exchange rate appreciation was much faster in the CIS-7 economies than in other transition countries is consistent with the pattern of debt-financed consumption booms. The investment dynamics, however, have differed between the two groups of countries. In the CIS-7 countries, gross fixed-capital formation as a percent of GDP has declined, on average, over time. In other transition countries, investment ratios have fluctuated around 23 percent of GDP during the entire transition.

In sum, Figure 2.11 suggests that the accumulation of external debt in the CIS-7 countries has largely financed higher consumption as a share of GDP. On closer examination, a large part of this consumption appears to be related to energy products When consumption is measured in real terms (rather than as a share of GDP), the picture is somewhat more sobering. The across-the-board collapse of output combined with inflationary bursts at the start of transition hurt real consumption substantially. This was reflected in rising poverty levels and mortality rates.

Conclusion and Policy Implications

Starting from virtually no debt in 1992, external debt ratios in the CIS-7 countries worsened rapidly. This debt buildup occurred even as their currencies appreciated and the average interest rate on the debt declined during the years 1992–2001. A large and increasing share of the debt was to multilateral institutions, especially the IMF and the World Bank.

Much of the CIS-7 debt problems today can be traced to a combination of adverse initial conditions—the cutoff of subsidies from Moscow, the breakdown of CMEA trade, the dismantling of the planned system, and large terms-of-trade shocks as energy prices rose to near commercial terms—coupled with external shocks during the transition (the Russian crisis in 1998 and internal armed conflict in some countries), and delayed macroeconomic policy response, availability of multilateral loans, biased growth projections, overoptimism on the part of official lenders regarding macroeconomic policy performance, and considerable statistical uncertainty.

While it is always hard to assess the right combination of adjustment and financing for countries facing large macroeconomic imbalances, three points are worth noting in the context of the CIS countries. First, financing the large current account and fiscal deficits with loans required the clear expectation of rapid productivity increases, which did not happen fast enough. Output continued to decline for a considerable period, and the growth pickup was slow. Debt ratios, therefore, rose rapidly. Second, because financing was not accompanied by faster adjustment, it can be argued that multilateral aid abetted the delayed adjustment in the CIS-7. In the absence of any form of financing, the countries would have been forced to sharply reduce their current account deficits, mainly by cutting back on imports. This, however, would likely have entailed huge social costs such as unemployment and a rise in poverty and mortality rates. Consequently, if rapid adjustment were neither feasible, because the initial conditions were too harsh or the institutional capacity to implement reforms were rudimentary, nor desirable, because it would have entailed even larger social costs, the financing gap in the initial years should have been closed by external grants. Would such grant aid have been forthcoming had the transition path been predicted accurately? We do not know.

The good news is that most CIS countries are pursuing and achieving a relatively ambitious reform agenda. Inflation has fallen steadily and has reached single-digit levels in five out of the seven countries: average inflation rate declined to 16 percent in 2001 from 1,872 percent during 1992–95. The turnaround in the fiscal balance has also been impressive: on average, fiscal deficits declined to about 6 percent of GDP in 1999–2001 from about 15 percent in 1992–95. The progress in structural reforms has been even more impressive in some ways than in other transition countries considering the fact that the CIS countries started at a lower base.

A more tentative conclusion also follows from the analysis. The diagnosis that heavily indebted countries are prone to remain that way because of their short time horizons, biasing decisions toward debt-financing consumption and forgoing investment opportunities,²¹ does not seem to apply to the CIS countries. By undertaking significant structural reforms and preserving macroeconomic balance, these countries are creating the right conditions for a brighter future. To help sustain their progress and to prevent the large debt from being a drag, some external help in the form of debt forgiveness appears to be warranted in these specific cases.

Appendix. Decomposing the External Debt Dynamics—Summary of the Methodology

The decomposition of the external debt dynamics is based on balance of payments identities, which imply that the following equation holds:

\begin{matrix} F_{t + 1} - F_{1} = - C_{1} + r_{1} * F_{1} + Δ R_{t + 1} - K_{t} + Z_{t} . & (2) \end{matrix}

The equation states that the change in the stock of external debt in U.S. dollars between the beginning of period t and the beginning of period t+1, denoted as _Ft+1-F_t, equal the sum of interest payments on external debt, r*F, and the accumulation of foreign exchange reserves (and other assets held by residents abroad), ΔR; minus the sum of the external current account balance excluding interest payments, C, and non-debt creating net capital flows, K. In practice, the above equation will not hold exactly because valuation changes and other factors that are unrelated to current or financial account flows can affect the change in the face value of the external debt. For this reason, Figure 2.3 includes an entry other factors denoted with Z_t.

In the case of the CIS-7 countries, the contribution of the change in foreign exchange reserves (or other assets held by residents abroad) to the change in total debt is small, so that they are subsumed in the residual Z_t. Also, for practical purposes, non-debt creating FDI inflows are the most important item in the category non-debt creating net capital flows, so that the figure refers to them directly. Hence, any other flows in the general category are also part of the residual.

To assess the debt burden and the debt dynamics, it is customary to use the ratio of external debt to GDP (in U.S. dollars, the standard currency denomination in external debt statistics) rather than the debt stock in nominal terms.²² For this purpose, the identity can be reformulated with all terms expressed as ratios to GDP Y:

\begin{matrix} f_{t + 1} - F_{1} = - c_{1} + (\frac{r_{t} - {\hat{y}}_{t} - {\hat{q}}_{t} - π_{t} *}{(1 + {\hat{y}}_{t}) (1 + {\hat{q}}_{t}) (1 + π_{t})}) f_{1} - k_{1} + z_{t}, & (3) \end{matrix}

where small letter variables denote variables as ratios of U.S. dollar GDP except for those with a hat on top, which denote rates of change (as fractions). Specifically, ŷ_t, ${\hat{q}}_{t}$ , and π,* denote the rates of change of real GDP, the real exchange rate against the U.S. dollar, and U.S. dollar inflation, respectively.²³

The second term on the right hand side of equation 2 shows how GDP growth can reduce the burden of existing debt. This term contains only the first-order terms of what actually amounts to an approximation of the change in external debt as a fraction of GDP. Hence, the residual term z_t now also encompasses higher-order approximation terms, which are typically very small If growth in nominal U.S. dollar GDP exceeds, on average, the implied nominal interest on external debt, current account deficits need not add to the debt burden. See Cohen (1988) on external debt sustainability. While growth in excess of market interest rates is unlikely in a steady state, it can exceed interest rates during transition, especially with large shares of official financing. To assess the extent to which GDP growth has alleviated the external debt burden, the interest and growth components are shown separately in Figure 2.3. For the growth component, we distinguish between real GDP growth and real exchange rate changes, since the two factors can be driven by distinct forces.

With these identities, one can also derive formulas to decompose the debt ratio over several years, as is done in the figures in the main text. All components in those figures are shown as cumulative annual flows (factors) or changes in transition time. For example, the change in debt in period j refers to

\begin{matrix} f_{t_{0} + j} - f_{t_{0}} . & (5) \end{matrix}

and the external current account in period j to the sum

\begin{matrix} - \underset{k = 1}{Σ} j_{c_{t_{0} + k}}, & (6) \end{matrix}

where t₀ refers to the beginning of the transition.

For the decomposition of the forecast errors for the external debt-to-GDP ratios, we had to make a few simplifying assumptions, given differences in reporting across time and countries. In particular, we used the external current account balance, including interest payments on external debt, rather than the primary external current account balance. This seems an acceptable simplification, given that interest rates on the CIS-7 countries’ external debt were low and stable. The decomposition is based on a first-order approximation of the contribution of forecast errors in the constituent elements of the debt-to-GDP ratio to the forecast error in that ratio. Specifically, the forecast error in the external debt ratio in period t+h is decomposed as follows:

\begin{matrix} f_{t + h} - f_{t + h \ t} \approx - n_{t + h \ t} (\frac{N_{t + h} - N_{t + h \ t}}{N_{t + h \ t}}) + \\ (f_{t \ t} - n_{t + h \ t}) (\frac{Y_{t + h} - Y_{t + h \ t}}{Y_{t + h \ t}}) + f_{t \ t} (\frac{F_{t} - F_{t \ t}}{F_{t \ t}}) . & (7) \end{matrix}

where N denotes the sum of the external current account balance, non-debt creating FDI inflows, and the residual Z, where t+h |t denotes the forecast of a variable in t+h prepared in t, and where Y is now the GDP in U.S. dollars (as above, a small letter variable stands for a ratio). This formula allows for subsequent revisions in initial values, as the notation t | t suggests. With the sign ≈, we draw attention to the fact that this approach only provides a rough approximation, as it does not take into account higher-order terms in the forecast errors of the constitutive elements of the external debt ratio.

References

Berg, Andrew, and others, 1999, “The Evolution of Output in Transition Economies: Explaining the Differences,” IMF Working Paper 99/73 (Washington: International Monetary Fund).
- Search Google Scholar
- Export Citation
Blanchard, Olivier J., 1997, The Economics of Post-Communist Transition (Oxford: Clarendon Press).
- Search Google Scholar
- Export Citation
Calvo, Guillermo A., Ratna Sahay, and Carlos A. Végh, 1995, “Capital Flows in Central and Eastern Europe: Evidence and Policy Options,” IMF Working Paper 95/57 (Washington: International Monetary Fund).
- Search Google Scholar
- Export Citation
Campos, Nauro F., and Fabrizio Coricelli, 2002, “Growth in Transition: What We Know, What We Don’t, and What We Should,” Journal of Economic Literature, Vol. 40, (September), pp. 793–836.
- Search Google Scholar
- Export Citation
Cohen, Daniel, 1988, “Which LDCs Are Solvent?” European Economic Review, Vol. 32 (March), pp. 687–93.
- Search Google Scholar
- Export Citation
De Melo, Martha, Cevdet Denizer, and Alan Gelb, 1996, “Patterns of Transition from Plan to Market,” World Bank Economic Review, Vol. 10 (September), pp. 397–424.
- Search Google Scholar
- Export Citation
Easterly, William R., 2001, “Growth Implosions and Debt Explosions: Do Growth Slowdowns Cause Public Debt Crises?” Contributions to Macroeconomics, Vol. 1.
- Search Google Scholar
- Export Citation
Easterly, William R., 2002, “How Did Heavily Indebted Poor Countries Become Heavily Indebted? Reviewing Two Decades of Debt Relief,” World Development, Vol. 30 (October), pp. 1677–96.
- Search Google Scholar
- Export Citation
Fischer, Stanley, and Ratna Sahay, 2000, “The Transition Economies After Ten Years,” IME Working Paper 00/30 (Washington: International Monetary Fund).
- Search Google Scholar
- Export Citation
Fischer, Stanley, and Ratna Sahay, and Carlos A. Végh, 1996a, “Stabilization and Growth in Transition Economies: The Early Experience,” Journal of Economic, Perspectives, Vol. 10 (Spring), pp. 45–66.
- Search Google Scholar
- Export Citation
Fischer, Stanley, and Ratna Sahay, and Carlos A. Végh, 1996b, “Economies in Transition: The Beginnings of Growth,” American Economic Review, Vol. 86 (May), pp. 229–33.
- Search Google Scholar
- Export Citation
Gelos, R. Gaston, and Ratna Sahay, 2000, “Financial Market Spillovers in Transition Economies,” IMF Working Paper 00/71 (Washington: International Monetary Fund).
- Search Google Scholar
- Export Citation
Helbling, Thomas, Ashoka Mody, and Ratna Sahay, forthcoming, “Debt Accumulation in the CIS-7 Countries: Bad Luck, Bad Policies, or Bad Advice,” IMF Working Paper (Washington: International Monetary Fund).
- Search Google Scholar
Obstfeld, Maurice, and Kenneth Rogoff, 1996, Foundations of International Macroeconomics (Cambridge, Massachusetts: IIT Press).
- Search Google Scholar
- Export Citation
Reinhart, Carmen M., Kenneth S. Rogoff, and Miguel A. Savastano, 2003, “Debt Intolerance,” Brookings Papers on Economic Activity, No.l (Washington: Brookings).
- Search Google Scholar
- Export Citation

The authors are in the IMF Research Department. They would like to thank Philip Lane, John Odling Smee, and several colleagues in the European II department of the IMF for helpful comments, and Gohar Abajyan and Emily Conover for research assistance, The views expressed here are those of the authors and should not be attributed to the International Monetary Fund.

The Commonwealth of Independent Stales (CIS) comprise Armenia, Azerbaijan, Belarus, Georgia, Kazakhstan, the Kyrgyz Republic, Moldova, Russia, Tajikistan, Turkmenistan, Ukraine, and Uzbekistan. of these, the low-income countries, known as the CIS-7, are Armenia, Azerbaijan, Georgia, the Kyrgyz Republic, Moldova, Tajikistan, and Turkmenistan.

See Easterly (2001) on how developing countries became highly indebted in the early 1980s or Reinhart, Rogoff. and Savastano (2003) on the experience of the newly independent Latin American countries in the 1820s.

We follow the country classification used in the World Bank’s Global Development Finance, our main data source in this section. Specifically, low-income developing countries are defined on the basis of a gross national income (GNI) per capita of less than US$745 while low-income developing countries that are heavily indebted either have an external debt-to-exports ratio of 220 percent or more or an external debt-to-GNI ratio of over 80 percent.

See Fischer, Sahay, and Végh (1996a) and Fischer and Sahay (2000).

The per capita income in CIS-7 countries is comparable to that of low-income developing countries. This qualifies them to receive external financing on concessional terms from various sources, including from the IFIs.

Strictly speaking, U.S. dollar inflation also needs to be considered as discussed in the Appendix. This factor, however, is outside the control of the transition countries, and its contribution to the debt dynamics has been small.

It should he noted that among the CIS-7 countries, the distribution of FIM inflows is very uneven. Azerbaijan, the oil-producing country, has received three times the CIS-7 average, while the slow reformers, Tajikistan and Uzbekistan, have received very little FDI.

Blanchard (1997).

Obstfeld and Rogoff (1996).

Campos and Coricelli (2002).

Helbling, Mody, and Sahay (forthcoming).

The attribution of the overall error to the components had to be approximated because as the latter enter the debt-to-GDP ratio nonlinearly. We used a first-order approximation (see Appendix). Even though in the case of large errors in components (e.g., unanticipated large real depreciation) the second-order terms become significant, we do not report them here because the first-order approximation is sufficient to illustrate our main arguments.

As shown in Helbling, Modys, and Sahay (Forthcoming).

In the case of Moldova, the reporting in pragram documents docs not allow for the use of identical time spans for the calculation of forecast errors, which explains the varying number of years in the forecast errors for GDP

See Fischer, Sahay, and Végh (1996a and 1996b) and De Melo, Denizer, and Gelb (1996). Berg and others (1999) confirm this finding but show that the effect of the initial conditions declines over time, while policy performance becomes increasingly important.

See, for example, Easterly (2001).

See Gelos and Sahay (2000).

We experimented with other specifications as well. For example, we estimated equations that used the change in official financing (as a percent of GDP) rather than the level. The main conclusions remain similar. We also re-specified our original equation using two-year rather than four-year averages for the regressors and the dependent variable. Interestingly, disbursements by multilateral institutions now appear more important in explaining the adjustment dynamics during the first two two-year periods, while the general government balance becomes insignificant.

See Calvo, Sahay, and Vegh (1995).

As a caveat, we note that the averages for the early years of the transition exclude the data for some CIS-7 countries because they did not report national accounts data by expenditure in the early phases of the transition.

See, For example, Easterly, (2002).

Exports or government revenues are other frequently used denominators.

The real exchange rate against the U.S. dollar is defined as

\begin{matrix} Q_{t} = \frac{P_{t}}{S_{t} P_{t} *}, & (4) \end{matrix}

where P_t denotes the GDP deflator, S_t the price of US$1 in national currency, and P_t* a U.S. dollar-based “world” price index for traded goods (taken from the IMF’s World Economic Outlook database).

Topics

Countries

Series

About

Resources

Abstract

Introduction

Key Statistics and External Debt Dynamics

Structure, Maturity, and Creditors

Decomposing External Debt Dynamics

External Adjustment: Programs and Outcomes

Case Studies

The Kyrgyz Republic

Moldova

Decomposing the Forecast Errors in External Debt Ratios

External Adjustment: A Comparison with Other Transition Economies

What Explains the Difference in the Adjustment Record? Five Hypotheses

Initial Conditions and Distortions

The Transition

Large Exogenous Shocks During the Transition

Policy Performance

Donor Financing and Overoptimism

External Adjustment: Econometric Evidence

What Has the Debt Accumulation Financed?

Conclusion and Policy Implications

Appendix. Decomposing the External Debt Dynamics—Summary of the Methodology

References

Within Same Series

Other IMF Content

Other Publishers Content

Asian Development Bank

Inter-American Development Bank

Nordic Council of Ministers

The World Bank

Share Link

Table of Contents

Sections

Figures

References

Metrics