This Selected Issues paper and Statistical Appendix analyzes the reasons behind the relatively low rates of savings in Bulgaria and prospects for their evolution over the medium term. The paper argues that low saving rates largely reflect the current stage of transition—characterized by still low income levels, incomplete structural reforms, the memories of the financial and banking crises of 1996–97, and an adverse demographic structure. An analysis of prospective saving rates indicates that as the transition process advances, saving rates may increase by 5 percentage points over the medium term.

Abstract

This Selected Issues paper and Statistical Appendix analyzes the reasons behind the relatively low rates of savings in Bulgaria and prospects for their evolution over the medium term. The paper argues that low saving rates largely reflect the current stage of transition—characterized by still low income levels, incomplete structural reforms, the memories of the financial and banking crises of 1996–97, and an adverse demographic structure. An analysis of prospective saving rates indicates that as the transition process advances, saving rates may increase by 5 percentage points over the medium term.

I. Savings in Bulgaria: Recent Developments and Prospects1

A. Introduction

1. After plunging at the start of the transition process, saving rates in Bulgaria have remained lower than in more advanced transition countries. As a result, although investment rates have been modest—in part reflecting the same set of factors that led to low savings—the external current account deficit has averaged more than 5½ percent of GDP in 1999–2001. While a large proportion of this deficit has been covered by FDI inflows, gross external debt remained high at about 78 percent of GDP at end-2001. Looking ahead, sustaining high growth rates of GDP and spending requirements related to EU accession will require a sizable increase in investment. In this context, without a big rise in savings, continued large current account deficits could jeopardize the goal of reducing gross external debt over the medium term.

2. This chapter analyzes the reasons behind the relatively low rates of savings in Bulgaria and prospects for their evolution over the medium term. It argues that low saving rates largely reflect the current stage of transition—characterized by still low income levels, incomplete structural reforms, and the memories of the financial and banking crises of 1996–97—and an adverse demographic structure. An analysis of prospective saving rates indicates that as the transition process advances, saving rates may increase by 5 percentage points over the medium term. Under this scenario, national savings would increasingly finance domestic investment, the current account deficit would decline to about 4½ percent of GDP—a level likely to be covered by FDI inflows—and the ratio of gross external debt to GDP would reach 50 percent.

3. The chapter is organized as follows. Section B compares the evolution of saving rates in Bulgaria and other transition countries. Section C reviews the factors that may explain, in general, differences in saving rates across countries. Section D presents a simple econometric analysis to estimate the quantitative importance of some of those factors and assess what may account for the differences between Bulgaria and other transition countries. It also analyzes the prospects for the evolution of savings in Bulgaria over the medium term. Section E summarizes the conclusions and policy implications.

B. The Facts

4. Bulgaria experienced a savings collapse at the outset of the transition process. While this phenomenon occurred in other transition countries, the decline was larger and more protracted in Bulgaria, where (national) savings declined from more than 30 percent of GDP in the late 1980s, to about 15 percent in 1990–92, and to less than 10 percent on average during 1993–96 (Figure 1).

Figure 1.
Figure 1.

National Saving Rates in Selected Countries, 1989-2001

Citation: IMF Staff Country Reports 2002, 173; 10.5089/9781451804454.002.A001

Sources: World Development Indicators; IFS; and Bulgarian authorities.

5. In contrast, the saving rates in other countries generally bottomed out about three years after the start of transition reforms. In Poland and Hungary, for instance, savings declined from 40 percent and 26 percent of GDP, respectively, to about 10 percent of GDP in 1992–93. In Estonia and Latvia, where the transition started somewhat later, savings fell from almost 30 percent of GDP in 1993 to about 18 percent and 13 percent of GDP respectively in 1996. These are relatively low levels of savings in comparison with several groups of countries (Table 1). Subsequently, saving rates in all these countries stabilized (Estonia), recovered somewhat (to 18 percent of GDP in Latvia) or increased dramatically (to more than 20 percent of GDP in Poland and Hungary).

Table 1.

Average National Saving Rates in Selected Groups of Countries

(In percent of GDP)

article image
Sources: World Bank World Development Indicators, and staff estimates.

Countries with average per capita GDP in 1995-99 below US$2,000, at 1995 prices. GDP per capita in Bulgaria was slightly below US$1,500 during the same period.

6. The initial plunge in saving rates in transition economies has been partly explained by the existence of involuntary savings under central planning. Involuntary savings at the household level would have been the result of official prices being at below-market-clearing levels, and a lack of generalized access to black markets. Perhaps more importantly, very high savings were necessary to finance the large investments targeted by the planners in pursuance of growth. Greater enterprise autonomy and price liberalization early in the transition did away with these involuntary savings. Denizer and Wolf (1998) estimate that the saving rate of a market economy with the same characteristics of Bulgaria in 1989 would have been about 22 percent of GDP—about 10 percentage points less than the actual saving rate.

7. The large declines in output at the start of the transition process—resulting from deep transformational recessions—were also responsible in part for the drop in savings. The output declines varied from about 15 percent in Poland or Hungary, to more than 40 percent in the Estonia and Latvia, and output started to grow only about three years after the start of reforms—a timing coincident with the bottoming out of saving rates and the start of their recovery (Figure 2). A desire to smooth consumption during these recessions—which were perceived as temporary—resulted in a lower saving rate.

Figure 2.
Figure 2.

Index of Real GDP in Selected Countries, 1989-2001

Citation: IMF Staff Country Reports 2002, 173; 10.5089/9781451804454.002.A001

Source: World Economic Outlook database.

8. In Bulgaria, savings developments were affected by a transformational recession that was prolonged and deepened by the lack of necessary structural reforms and prudent macroeconomic policies. After falling by more than 30 percent in 1990–1993, real GDP experienced a short-lived and very small recovery in 1994–95. The severe financial crisis of 1996–97 caused another deep output decline of about 17 percent. It is not surprising that in this context savings remained low—if only to smooth consumption while waiting for an eventual turnaround in output. Government policies also affected savings during these years, to a much larger extent than in the most advanced transition countries, for several reasons. Capital expenditure was extremely low. The large fiscal deficits were only partially offset by higher private savings, and these clearly unsustainable deficits led to high inflation that depleted most savings. Finally, in the absence of structural reforms and any efforts at enterprise restructuring, the conditions for a turnaround in growth performance were not in place.

9. The banking crisis of 1996 resulted in a significant reduction in financial intermediation and a loss of confidence in the banking system. Negative capital and insufficient liquidity to withstand a run on deposits led to the closure of 18 banks (a third of all banks, accounting for 30 percent of total assets). The loss of confidence in the banking system was such that the ratio of deposits to GDP fell from 54 percent at end-1995 to only 16 percent in 1997. This episode undermined the trust of the public in the financial system and the government’s macroecomic policies.

10. It was not until 1997 that Bulgaria adopted a set of sound macroeconomic policies—in the context of a currency board arrangement—and a truly ambitious program for structural reform. These policies set the basis for a resumption of still-continuing positive growth despite unfavorable external developments, such as the Russian crisis, the conflicts in neighboring Balkan countries, and the recent economic slowdown in the EU. With Bulgaria’s economic recovery, the saving rate has increased to around 13 percent of GDP in the last few years but still remains below that in the most advanced transition countries. While tight fiscal policy and a rise in capital expenditure have increased public savings to an average of 3 percent of GDP, which compares favorably to other transition countries, private savings in Bulgaria, at about 10 percent of GDP, are well below the rates observed in other countries. Table 1 shows that, compared with other groups of countries—not only with other transition economies—overall saving rates in Bulgaria are also relatively low.

C. Explanations

11. Theoretical and empirical studies have identified a variety of factors that influence the level of savings in an economy, the most important being the level of income and growth, demographic factors, and the development of the financial system. Other variables, such as government savings and social security systems, have also been found to play a role. This section briefly reviews the links between these factors and savings.

The level of income and growth

12. In a poor economy, income is well below long-run or permanent income—at least, if there are prospects of catching up or beginning to close the gap vis-à-vis richer countries. With a preference for consumption smoothing, people would like to consume a great deal in relation to income when they are poor, and the saving rate would be low. This effect may be particularly important in the context of transition and EU accession, where the prospects for medium-term growth may be quite strong. In addition, in poor economies a large proportion of income must be spent on necessities that guarantee subsistence.2 For both of these reasons, the “income effect” tends to raise the saving rate as an economy becomes richer. A “substitution effect” works in the opposite direction: as an economy develops, the rate of return on saving may drop—for instance, under decreasing marginal returns to the accumulation of capital—and this would lower the saving rate as income increases. The empirical evidence—see, for example, Edwards (1996); and Masson, Bayoumi, and Samiei (1995)—is that saving rates are positively correlated with income per capita, suggesting that the income effect dominates. In theoretical models with a life-cycle setting, income growth will also have a positive impact on savings, because the savings of workers increase faster relative to the retirees’ dissavings.

Demographics

13. The age structure of the population appears to play an important role in the determination of savings. In life-cycle models, individuals have low or negative saving rates when they are young, positive savings during their most productive years at work, and negative savings when they retire. Thus, saving rates would tend to be higher in countries with a higher share of the population of working age. Empirical studies (e.g., Edwards, 1996) find, indeed, that the dependency ratio—the ratio of non working to working-age population—is negatively correlated with saving rates.

Financial system

14. A well-developed financial system plays a key role in channeling savings to the best available investment opportunities, and in doing so, it may offer higher interest rates or superior combinations of risk and return. The effect of higher interest rates on the level of savings, however, is theoretically ambiguous because of the different sign of income and substitution effects. The empirical evidence suggests that higher interest rates have only a very small positive effect on savings—Srinivassan (1993). The consequences of financial sector development for savings, however, go well beyond the effects through the interest rate. On the one hand, some authors indicate that easier access to borrowing could reduce savings for the purchase of durables or for a precautionary motive. On the other hand, some have argued that a number of institutional developments, including efficient prudential supervision and—in some rapid-growth East-Asian countries—the encouragement of postal savings, positively affected savings. Others have also stressed the importance of some aspects that go hand in hand with financial development, and that promote higher savings, such as increases in confidence in financial intermediaries, or more convenience when, for example, more bank branches open up. Empirical studies have usually found a positive effect of proxies of financial development on savings (Edwards, 1996, and UNECE, 2001) with negative effects found in subsamples of countries with relatively sophisticated financial systems (IMF, 1996).

Government and social security

15. The level of government spending, its composition between current and capital expenditures, and the way it is financed—via higher taxes or borrowing—may all have an effect on the level of national savings. For a given level of budget deficit, higher government consumption tends to decrease national savings—especially if this spending is an imperfect substitute for private consumption—and conversely with higher capital expenditure. For a given level of government expenditure, an increase in the budget deficit would have no impact on national savings if private sector savings increased one-to-one in anticipation of future taxes (that is, if Ricardian equivalence holds). Empirical studies (Corbo and Schmidt-Hebbel, 1991; or Masson, Bayoumi, and Samiei, 1995) tend to find, however, that the increase in private savings does not fully offset a decline in government savings. More generous social security payments will also lower the savings of working-age individuals, and funded systems may increase national savings compared with pay-as-you-go systems. Even the types of taxes the government uses to raise revenue have an impact on savings, for example, taxes on consumption would tend to raise savings, relative to income taxes.

D. Empirical Analysis

16. A simple econometric analysis is used to estimate the quantitative effects of several of the explanatory variables discussed above. A panel data set from the World Bank World Development Indicators database covering the years from 1970 to 1999 is used for the estimation. To clear the impact of purely cyclical effects on savings five-year averages were taken for each of the variables in each country. The panel dataset consists of more than 120 countries, and each country has a maximum of 6 observations—that is averages for 1970–74, 1975–79, and so on until 1995–99. Transition countries are included for the most recent years when data are available. Some of the variables mentioned in section C are not included due to lack of data. As a “proxy” for financial development, the ratio of broad money to GDP is used, and a quadratic term is introduced to capture non-linear effects suggested above.

17. The empirical results are consistent with the explanations outlined in the previous section (Table 2). The first column shows the coefficient estimates in a regression using the entire sample.3 All are significantly different than zero (at the 1 percent significance level) and have the expected sign. The level of income has a positive effect on savings—a 10 percent increase in GDP per capita is estimated to raise savings by about 0.2 percentage point of GDP. One additional percentage point in the growth rate of GDP is also estimated to increase savings by about 0.2 percentage point.4 An increase in the proportion of the population of non-working age decreases savings, with the effect being stronger for increases in the proportion of retirees: One additional percentage point in the proportion of the population above 65 years of age lowers the savings rate by about 1 percentage point, three times as much as an increase in the ratio of the population below 14. The proxy used for financial development indicates that this initially has a positive effect on savings but at a decreasing rate, with the sign only reversing for levels of broad money above 110 percent of GDP—for a country with a 40 percent ratio of broad money to GDP (the median value for the entire sample in the 1995–99 period), a 1 percentage point increase in this ratio raises the saving rate by 0.1 percentage point. And finally, an increase in the government’s deficit by 1 percent of GDP is estimated to lower national savings by less than ½ percent of GDP.

Table 2.

Regression results

(standard errors in parenthesis)

article image

18. The results are fairly similar when separate regressions are run by decade. However, for recent times the effect of the level of income and growth seems to weaken, whereas the effect of demographic variables is stronger. The coefficient estimate on GDP is even negative for 1990–99; while for the most recent period of 1995–99 the coefficient is positive again, it implies only that a 20 percent increase in per capita GDP raises savings by 0.1 percent of GDP.

19. For Bulgaria, predicted savings for 1995–99 are about 15 percent of GDP, or 2 percentage points higher than actual savings. It is in this sense that one can claim that Bulgaria has a relatively low savings rate.

20. How can the differences between Bulgaria and other transition countries be explained? A set of countries including Croatia, the Czech Republic, Estonia, Hungary, Latvia, Lithuania, Poland, Romania, and Slovakia, had an average saving rate in 1995–99 of 20 percent of GDP (Figure 3). Of a difference of 7 percentage points with Bulgaria, differences in observables account for 4–5 percentage points (depending on whether we use the coefficients from the regression for 1995–99 or the one for 1970–99): A lower income level and lower growth in Bulgaria explain 2–3½ percentage points, and an adverse demographic structure explains 1½–2 percentage points (Bulgaria has the highest ratio of population over 65). Differences in fiscal deficits were small and thus may not be part of the explanation.

Figure 3.
Figure 3.

Actual vs. Predicted Saving Rates in Selected Transition Countries

(actual in vertical axis; averages 1995-99)

Citation: IMF Staff Country Reports 2002, 173; 10.5089/9781451804454.002.A001

Source: Staff estimates, using regression shown in column 6 of table 2.

21. While differences in the “proxy” for financial development are small and therefore do not explain much, one could argue that before and after the banking crisis of 1996–97 in Bulgaria, confidence in the financial system seriously deteriorated—even beyond what would be captured in the monetization ratio that we use as a crude proxy. This effect may account for a large share of what remains “unexplained.”

22. Compared with most of these countries, Bulgaria has also lagged in efforts at reforms in key areas (see Figure 4). Delays in structural reforms and the resulting lack of restructuring have limited the availability of good investment projects—and by lowering opportunities for lending, the banks have had limited incentives and ability to attract savers.5 Banks have also been cautious in their domestic lending because of an imperfect legal environment, including problems with the enforcement of creditor rights.

Figure 4.
Figure 4.

EBRD Indicators of Transition Reform

Citation: IMF Staff Country Reports 2002, 173; 10.5089/9781451804454.002.A001

Source: EBRD transition report 2001; Privatization index is an average for small and large scale privatization.

Looking ahead

23. Projections for the medium-term values of the explanatory variables indicate that the saving rate could increase by about 5 percentage points over the medium term—from its average of 13 percent in 1995–99. The interaction of the coefficient estimates from the regressions above and the projected evolution of the explanatory variables suggest that: an increase in the income level—as a result of sustained growth of around 5 percent a year—would raise the savings rate by about 3 percentage points of GDP; continued monetization—the ratio of broad money to GDP is projected to increase to more than 50 percent of GDP by 2007—would imply higher savings by almost 1 percentage point of GDP; the consolidation of a balanced government budget would increase savings by about ½ percentage points of GDP (the fiscal deficit is projected at 0.8 percent in 2002 and zero by 2005 and beyond, compared with an almost 3 percent average in 1995–99). Although the ratio of the population below 14 years of age is expected to decline as a result of low past and current fertility rates, the continued aging of the population will imply an increase in the ratio of those above 65 years of age from 15.1 percent in 1995–99 to 16.7 by 2007. The net demographic effect would subtract about ½ percentage point of GDP from the savings rate.

24. In addition to the effects of the expected evolution in the covariates used in the regression above, other important variables could contribute to an additional increase in the savings rate. The improvement in confidence in the financial system and more effective financial intermediation could have favorable effects. If unobservables related to this phenomenon were responsible for the lower-than-projected savings in 1995–99, a gradual improvement would contribute to higher savings. However, most of this effect is already reflected in our projections for the monetization ratio, which implicitly reflect the staff’s assessment on the favorable prospects for the development of the financial system. The Bulgarian authorities also intend to increase the share of capital expenditures, from about 2 percent of GDP on average in 1995–99 to 4 percent in 2004 and beyond, in part to meet EU-accession-related spending. In the context of a balanced budget, this shift towards capital spending would result in an increase in public and overall savings. The short- and medium-term effects of pension reform on savings will depend, in part, on how the transition costs of introducing a funded pillar are financed—the impact will be larger and more positive if these costs are met with cuts in current expenditure. Tax reform—mainly the authorities’ plans to lower direct personal and corporate income taxes, and raise indirect taxes—would increase incentives to save as well. Continued restructuring, in addition to lower corporate income taxes, would lead to higher corporate profitability, and provide a boost to corporate savings.

25. Given the uncertainty surrounding the estimates presented above, the staffs current macroeconomic framework projects a more conservative increase in the saving rate, to slightly more than 17 percent of GDP by 2007. With investment rates projected to reach 22 percent of GDP over the medium term, consistent with growth projections, the current account deficit would gradually decline to 4½ percent of GDP. As this deficit is expected to be covered by FDI inflows, the gross external-debt-to-GDP ratio would decline to 50 percent, from almost 80 percent in 2001.

E. Conclusions and Policy Implications

26. The seemingly low rates of saving in Bulgaria can be explained, to a large extent, by the country’s characteristics, in particular, its low level of income and an adverse demographic structure. Still, actual saving rates are lower than what would be predicted solely on the basis of Bulgaria’s characteristics. We have argued that two reasons may be largely responsible for this outcome: the adverse shock to confidence in the Bulgarian financial system as a result of the banking crises in 1996–97, and the lack of economic restructuring and legal problems constraining domestic lending, which in turn have limited the returns to saving.

27. On the basis of projected values for variables that influence savings, the chapter estimates that the savings rate could increase by about 5 percentage points in the medium term. Unfavorable developments in the demographic structure, as the population continues to age, would be more than offset by higher income levels, more effective financial intermediation, and continued tight fiscal policies. Higher savings would increasingly finance necessary investments to reach growth objectives and accede to the EU, gradually bring down the current account deficit, and allow for a sharp decline in the ratio of gross external debt to GDP.

28. In addition to continued prudent fiscal policies, the projections for higher savings rely on the implementation of other policies that would provide the proper environment for sustained high rates of income growth, and the development of the financial system. The most important policies include:

  • An acceleration of structural reforms that encourage restructuring and increase productivity and, thus, the returns on savings.

  • Policies that foster sound financial intermediation, including finalizing the privatization of state-owned banks; improving the legal framework to better protect creditors’ rights—especially relating to bankruptcy and liquidation procedures; and strengthening banking supervision—particularly important in a context of rapid credit growth.

  • A decline in current expenditure on the part of the government will be needed to finance increased accession-related capital spending and remove the pressure that the transition costs of pension reform will exert on the fiscal accounts.

References

  • Carrol, C. and D. Weil, 1993. “Savings and Growth: A Reinterpretation,” NBER Working Paper n.4470, September.

  • Collins, Susan M, 1991. “Saving Behavior in Ten Developing Countries,” in B. Douglas Bernheim and John B. Shoven, eds., National Saving and Economic Performance, Chicago: Chicago University Press for NBER, 34972.

    • Search Google Scholar
    • Export Citation
  • Corbo, V. and K. Schmidt-Hebbel, 1991. “Public Policies and Saving in Developing Countries,”Journal of Development Economics, n36, 89115.

    • Search Google Scholar
    • Export Citation
  • Deaton, Angus, 1999. “Saving and Growth,” in Klaus Schmidt-Hebbel, Luis Servén, eds., The Economics of Saving and Growth: Theory, Evidence, and Implications for Policy, Cambridge: Cambridge University Press, 33370.

    • Search Google Scholar
    • Export Citation
  • Denizer, Cevdet, and Holger C. Wolf, 1998. “The Savings Collapse during the Transition in Eastern Europe,” Working Paper, World Bank “Saving Across the World” research project.

    • Search Google Scholar
    • Export Citation
  • Edwards, Sebastian, 1996. “Why Are Latin America’s Saving Rates So Low? An international comparative analysis,” Journal of Development Economics, vol.51, 544.

    • Search Google Scholar
    • Export Citation
  • Honohan, Patrick, 1999. “Financial Policies and Saving,” in Klaus Schmidt-Hebbel, Luis Servén, eds., The Economics of Saving and Growth: Theory, Evidence, and Implications for Policy, Cambridge: Cambridge University Press, 71105.

    • Search Google Scholar
    • Export Citation
  • International Monetary Fund. 1996. “Savings and Investment Performance in International Perspective,” in Selected Economic Issues, United Kingdom Staff report for the 1996 Article IV consultation.

    • Search Google Scholar
    • Export Citation
  • Masson, Paul, Tamim Bayoumi, and Hossein Samiei, 1995. “Saving Behavior in Industrial and Developing Countries,” in Staff Studies for the World Economic Outlook, IMF, 127.

    • Search Google Scholar
    • Export Citation
  • Srinivasan, T.N, 1993. “Saving in the Development Process,” in J. H. Gapinski, ed., The Economics of Saving, Dordrecht: Kluwer.

  • United Nations Economic Commission for Europe, 2001. “Domestic Savings in the Transition Economies,” Economic Survey of Europe n.1.

1

Prepared by Enric Fernandez.

2

In Bulgaria, for example, almost 50 percent of household income is spent on food.

3

In the presence of heteroscedasticity we used generalized least squares, and all standard errors are computed using White’s procedure. The second column in Table 2 shows that the coefficient estimates using ordinary least squares are virtually identical; although standard errors are, as one would expect, much higher, all variables remain significant at the 1 percent significance level.

4

Similar results were obtained using lagged growth as an instrument to avoid simultaneity problems. The results are not presented to economize space. Carrol and Weil (1993) conduct causality tests and find that growth “causes” savings.

5

Only about one-third of banks’ assets were lent domestically in 2001. The remaining were mostly placed abroad.

Bulgaria: Selected Issues and Statistical Appendix
Author: International Monetary Fund