Greece
Selected Issues and Statistical Appendix
Author:
International Monetary Fund
Search for other papers by International Monetary Fund in
Current site
Google ScholarClose
Publication Date:
31 Oct 1997
eISBN:
9781451816075
Language:
English
Keywords:
ISCR; CR; foreign currency; return on equity; tax rate; market share; revenue enhancement; withholding tax; banking sector; cost of funds; money market
Download PDF (6.7 MB)

This Selected Issues paper and Statistical Appendix reviews fiscal developments in Greece, comparing its revenue and expenditure patterns with a number of comparator country averages, and discusses prospects. The paper focuses on the evolution and tax-determinants of the most widely followed indicator of the health and competitiveness of a banking system, the lending-deposit spread. It also brings to bear a wide variety of indicators and econometric techniques to the problem of understanding the state and prospects of the banking system.

Abstract

This Selected Issues paper and Statistical Appendix reviews fiscal developments in Greece, comparing its revenue and expenditure patterns with a number of comparator country averages, and discusses prospects. The paper focuses on the evolution and tax-determinants of the most widely followed indicator of the health and competitiveness of a banking system, the lending-deposit spread. It also brings to bear a wide variety of indicators and econometric techniques to the problem of understanding the state and prospects of the banking system.

Greece: Basic Data

Sources: Data provided by the authorities; and staff estimates.

Fund presentation (accruals basis); includes capitalized and accrued interest on rescheduled or consolidated debt for 1991-97, which is not included in the official public finance statistics, based on Bank of Greece data. Interest for 1997 excludes 0.6 percent of GDP because of the introduction of two-year zero coupon bonds.

New definition as of 1990, including military debt and short-term liabilities to the Bank of Greece. Break in the series in 1994, when government foreign exchange liabilities to the Bank of Greece, not previously included in central government debt, were replaced by government bonds (an increase in the debt equal to 10 percent of GDP); and in 1993, when liabilities to the central bank were included in the stock of debt (see SM/95/179, 7/25/95).

Data on a national accounts basis.

M3 is defined as the sum of currency, private deposits, bank bonds, and repos; M4 also includes private sector holdings of T-bills and government bonds of maturity of up to one year.

Includes capitalized and accrued interest on consolidation bonds held by commercial banks.

On the basis of national accounts data excluding EU transfers to the Investment Budget, unless otherwise indicated.

Based on Ministry of Finance data on accrual basis.

End-of-period.

I. Introduction and Overview

1. Since 1994, the first year of the Revised Convergence Program, impressive strides have been made toward macroeconomic stabilization and fiscal adjustment. However, with Greece the only EU country that currently satisfies none of the Maastricht convergence criteria, much remains to be done. This background document presents a number of studies designed to probe in more depth issues of fiscal adjustment, the public sector’s creditworthiness as judged by foreign lenders, and the banking system. Each issue will prove important as the country strives to fulfill the criteria necessary to enter into the European Monetary Union.

2. Fiscal adjustment has followed a gradual path, relying to a significant extent on revenue enhancement and eschewing primary expenditure reduction. Conventional wisdom has asserted that Greece is relatively under-taxed in certain revenue categories, with significant segments of the population largely escaping taxation, and important social expenditures lagging behind other EU countries. Chapter II (prepared by M. Lutz) reexamines these assumptions, bringing to bear a cross-sectional data set which includes EU, OECD, and other countries with comparable income per capita. One conclusion is that while revenue as a share of GDP is only slightly higher than that of the average of economies with similar incomes, when compared to the relevant tax bases effective tax rates are higher than in those of comparator economies, and in some cases even higher than the OECD and EU averages. Regarding expenditures, Greece is found to maintain a level of transfer payments (mainly old-age and disability pensions) that are significantly higher than in countries with similar incomes. Expenditures on general government services and defense is also seem to claim large shares of resources while expenditures on health care and education are comparatively low. The paper also discusses current and prospective consolidation efforts in light of recent work analyzing how the composition of fiscal adjustment influences its likelihood of “success.”

3. The public sector, prompted by high domestic interest rates engendered by the high debt-to-GDP ratio, has supplemented its domestic borrowing program with foreign borrowing, often reducing its cost of borrowing by selling instruments denominated in foreign currency while assuming the exchange rate risk. The key Greek-specific element of the cost of these instruments is the foreign currency risk premium that foreign lenders demand. Chapter III (prepared by C. Christofides) presents a data set of these instruments in order to measure the risk premium component of the foreign interest cost and to link it to a set of macroeconomic and policy variables. Estimating benchmark spreads in order to correct for a systematic reduction in the maturity of these instruments, it is found that the Greek risk premium has increased over the period 1990–96, primarily because of the rise in its foreign debt and a concomitant worsening of its credit rating by external rating agencies. Conversely, it is found that the spread is lower than it otherwise would be as a result of the successful implementation of the hard drachma policy. This finding has implications for the post-EMU period, when exchange rates will be irrevocably fixed for those countries meeting the necessary criteria.

4. The banking system has undergone wide-ranging reform since its liberalization in the early 1990s. Nevertheless, significant areas of weakness remain, with state-controlled banking institutions inheriting bad debts while retaining a large market share. Chapters IV-VI examine the status and prospects of the banking system from a number of viewpoints.

5. Chapter IV (prepared by C. Christofides) focuses on the evolution and tax-determinants of the most widely followed indicator of the health and competitiveness of a banking system, the lending-deposit spread. Since the liberalization of the banking system in the early 1990s, and the concomitant rise in domestic interest rates, the spread has been one of the highest in Europe. By controlling for the impact of various tax factors on the lending-deposit spread, it is found that Greece has in fact had a spread closer to that of other high-spread EU economies over most of this period. One conclusion may be that the banking system is relatively highly taxed, a situation that may not be sustainable in the post-EMU regime as competition is expected to intensify.

6. Chapters V and VI (prepared by D. Hardy), bring to bear a wide variety of indicators and econometric techniques to the problem of understanding the state and prospects of the banking system. The evolution of the structure of the banking system is described in some detail. It is found that, although significant problems remain, much progress has already been made. On the negative side, state-controlled banks still enjoy a high market share despite their high operating costs. Some of these banks continue to enjoy beneficial tax treatment, while other aspects of the tax code tend to discourage adequate provisioning for bad loans. Widespread recourse to foreign borrowing, generally intermediated through the banking system, has also raised systemic risks. On the plus side, competition has intensified, the regulatory and supervisory framework largely conforms to international norms, and banks are increasingly subject to market discipline. Medium-size private banks have expanded their market share, and appear to earn high profits. All banks are diversifying their base of operations, entering sectors such as mortgage lending that in the past experienced little competition. A number of state-controlled banks have been cleaning up their portfolios, increasing provisioning as well as own capital both through new issues and through retained earnings. The econometric results suggest that medium-size banks have a profit advantage over both small and large banks: this finding may have implications for the ongoing reorganization of the banking system that is occurring through sales of small state-controlled banks and through mergers.

II. Fiscal Developments and Prospects: An International Perspective1

A. Introduction and Fiscal Developments

7. This paper reviews fiscal developments in Greece, comparing its revenue and expenditure patterns with a number of comparator country averages, and discusses prospects for continued successful consolidation. The paper is organized in five sections. The first provides a brief review of fiscal developments. The second section discusses appropriate international comparators, including EU and OECD partners, as well as economies whose per-capita income levels and economic structures are closer to that of Greece. The third section analyzes revenue shares and their composition while the fourth does the same for expenditure shares. While not intended to be prescriptive, the comparative exercise is useful in pointing out certain features of the fiscal system that bear further examination, and offers suggestions concerning areas in which further fiscal consolidation could focus. In doing so, it is found that while Greek revenues as a share of GDP are only slightly higher than that of the average of economies with similar incomes, when compared to the relevant tax bases its effective tax rates are higher than in these comparator economies, and in some cases even higher than OECD and EU averages. Regarding expenditures, Greece is found to maintain a level of transfer payments (mainly old-age and disability pensions) that are significantly higher than in countries with similar incomes. Expenditures on general government services and defense are also seen to claim large shares of resources, while expenditures on health care and education are comparatively low.

8. The final section discusses the current and prospective consolidation efforts in light of recent work analyzing how the composition of fiscal adjustment influences its likelihood of “success.” This literature suggests that consolidation efforts focusing on reducing expenditures, in particular reducing social transfers and the government wage bill, rather than on increasing revenues are more likely to be successful. These findings are in contrast with adjustment efforts to date, which have focused largely on increasing revenues.

9. Prior to the 1980s, the economy exhibited strong growth and the authorities pursued prudent financial policies.2 As seen in Table 1, the debt-to-GDP ratio was only 23 percent as late as 1980. Between 1980 and 1993, however, financial discipline weakened significantly. Transfer payments rose from 9½ percent of GDP to almost 16 percent of GDP in 1993, as eligibility requirements were liberalized and payments increased in real terms. Government employment also increased, with the wage bill rising from 11½ percent of GDP to over 15 percent of GDP in 1990. Fiscal revenues rose significantly as well, reflecting higher indirect taxes and social insurance contribution rates, as well as more progressive personal income taxes. Nevertheless, the increase was insufficient to contain the deficit, and by 1985 the debt-to-GDP ratio more than doubled to almost 50 percent of GDP, with a further doubling by 1993 to over 110 percent of GDP.

Table 1.

Greece: General Government Accounts, 1960-97

(Percent of GDP)

article image
Sources: OECD Analytical databank; Greek authorities; and staff calculations.

10. In June 1994 a Convergence Plan was adopted which targeted meeting Maastricht criteria by 1998. The fiscal deficit targets were met during 1994–96, and the general government deficit fell for three straight years. Of the 6¾ percentage points of GDP reduction in the deficit during this period, 3 percentage points were accounted for by increased revenues, and by a 3¾ percentage point of GDP improvement in net capital transfers. In fact, these two sources of fiscal improvement, combined with a fall in interest payments as a share of GDP, allowed for a 1½ percentage point of GDP increase in primary expenditures (mainly for higher social transfers and a further rise in the wage bill).

11. The authorities must engineer a further reduction of almost 4½ percentage points of GDP to meet the Maastricht criteria. However, Greece already has the highest revenue-to-GDP ratio it has ever experienced, and which is budgeted to increase in further in 1997. Expenditure is also near record levels, with debt servicing in excess of 10 percent of GDP. While the debt ratio may have peaked in 1996, and declining real interest rates, accelerating growth and an improved primary balance should introduce a virtuous cycle in debt dynamics, the nation’s demographics are anticipated to change in the coming decade, with Greece facing one of the more rapid agings of population among OECD members, with deleterious fiscal implications.3 Moreover, the EU presently provides current and capital transfers equivalent to about 5 percent of GDP, which are scheduled to fall over the medium term. Therefore, substantial additional measures are needed if consolidation goals are to be achieved on a permanent basis.

B. International Comparators

12. It is useful to compare Greece’s fiscal structure with other economies. This does not imply that its accounts should mirror those displayed by others on average, as differences will naturally arise from, inter alia, variation in geographic, demographic, cultural, and developmental histories and settings. It is important not to downplay the role these factors have in influencing variations in revenue and expenditure patterns.4 Nevertheless, determining areas in which fiscal data are “outliers” in some sense may suggest the need for adjustment.

13. The approach taken in this paper in determining appropriate comparators is rather agnostic. Possible groupings include the EU, of which Greece has been a member since 1981, and the OECD, of which it is a founding member. One issue arising from these comparisons, however, is that Greece has the lowest per-capita income in the EU and among the lowest in the OECD. Numerous studies have shown that fiscal patterns often vary predictably across economies depending upon their levels of per-capita income, the most famous of which is “Wagner’s Law,” which posits that as economies grow, there is a tendency for the scope and scale of government activities to increase.5 Moreover, economies with similar incomes often display other economic similarities, such as the share of activity comprised by agriculture, the self-employed, small nonincorporated business and other difficult-to-tax sectors. This paper extends comparisons to other countries that have similar per-capita incomes and for which comparable data are available.6 Thirty-eight countries in addition to Greece are used, although data are not available for all countries for all comparisons (Figure 1).7

Figure 1.
Figure 1.

Greece: Per-Capita Income at 1995 PPP US$

Citation: IMF Staff Country Reports 1997, 089; 10.5089/9781451816075.002.A001

Source: IMF, World Economic Outlook.

14. Table 2 provides comparative fiscal data for Greece and 33 other countries for whom general government data on a national accounts basis are available. The data are averaged over 1990–95, as available, to abstract from business cycle influences. There are a number of notable differences between Greece and some comparator averages. First, while it has lower current revenues than either its EU or OECD partners on average, its revenue share is slightly higher than in economies with comparable income levels. Second, as for the composition of revenues, Greece relies to a much lower extent on direct taxes (especially personal direct taxes) than in any comparator group, with more reliance on social insurance contributions, and to a lesser extent on indirect taxes, when compared to economies with similar income levels.8 Third, total current expenditures are relatively high in light of its debt servicing burden, but primary current expenditures (excluding debt servicing) are also higher than in other economies with similar incomes. Fourth, this difference in primary current expenditures reflects a large ratio of transfer payments to GDP, which exceeds that for OECD economies on average. Fifth, Greece has net capital receipts rather than transfers, reflecting, inter alia, EU transfers for investment purposes.

Table 2.

Greece: General Government Accounts, 1960-95 Average

(Percent of GDP; unless otherwise specified)

article image
Sources: OECD Analytical databank; United Nations National Accounts; WEO; and staff calculations. Note: All aggregate figures are weighted by relative GDPs at 1995 PPP US$.

Excluding Luxembourg for data reasons.

Excluding Luxembourg, Mexico, New Zealand and Turkey for data reasons.

Includes the countries listed in footnotes 4-6.

Includes: Colombia, Costa Rica, Czech Republic, Hungary, Korea, Malta, Mauritius, Panama, Poland, Portugal, Slovenia, Spain, Thailand and Uruguay.

Includes: Australia, the Bahamas, Belguim, Finland, Germany, Iceland, Ireland, Israel, Italy, the Netherlands, Sweden and the United Kingdom.

Includes: Austria, Canada, Denmark, France, Japan, Norway and the United States.

In 1995 PPP US$.

15. Figures 2–5 provide scatter plots of the current revenue, current expenditure, current primary expenditure, and transfer payments shares of GDP. In all four figures it is clear that ratios for the four Central European economies undergoing transformation into market-based economies are much higher than in other economies with similar income levels. Also notable is the low shares registered by the United States. Abstracting from these outliers, there appears to, be a positive correlation between revenue and expenditure shares and PPP per-capita incomes. Therefore, simple regressions were run, with component shares regressed on PPP per-capita incomes.9 The figures include the shares predicted by the regressions, as well as bands reflecting one standard deviation above or below the predicted shares.

Figure 2.
Figure 2.

Greece: General Government Current Revenues, 1990-95

(Period average)

Citation: IMF Staff Country Reports 1997, 089; 10.5089/9781451816075.002.A001

Sources: OECD Analytical Databank; United Nations National Accounts; IMF, World Economic Outlook; and staff calculations.
Figure 3.
Figure 3.

Greece: General Government Current Expenditures, 1990-95

(Period average)

Citation: IMF Staff Country Reports 1997, 089; 10.5089/9781451816075.002.A001

Sources: OECD Analytical Databank; United Nations National Accounts; IMF, World Economic Outlook; and staff calculations
Figure 4.
Figure 4.

Greece: General Government Current Primary Expenditures, 1990-95

(Period average)

Citation: IMF Staff Country Reports 1997, 089; 10.5089/9781451816075.002.A001

Sources: OECD Analytical Databank; United Nations National Accounts; IMF, World Economic Outlook; and staff calculations.
Figure 5.
Figure 5.

Greece: General Government Transfers, 1990-95

(Period average)

Citation: IMF Staff Country Reports 1997, 089; 10.5089/9781451816075.002.A001

Sources: OECD Analytical Databank; United Nations National Accounts; IMF, World Economic Outlook; and staff calculations.

16. According to these results, Greece’s current revenue share is only slightly higher than predicted, but well within one standard deviation. In contrast its current expenditure share is more than one standard deviation higher than its predicted share, consistent with the 12¾ percentage point difference between the share in countries at similar income levels shown in Table 2. However, this is the result of exceptional debt servicing obligations, as its primary expenditure share is not significantly higher than predicted. In contrast, transfer payments, the bulk of which go to households, are significantly higher than predicted.

C. Revenues

17. In the previous section, it was determined that while total current revenues were not significantly different than what was predicted based on Greece’s income level, it relied to a greater degree on social insurance contributions and to a lesser extent on direct taxes. Moreover, it had relied in recent years on revenue increases as the main vehicle for fiscal consolidation. This subsection explores the tax structure in more detail, and examines the scope remaining for further revenue efforts.

18. In determining the effects of taxation on resource allocation and economic efficiency, it is marginal, rather than average, tax rates that are important. Greece’s statutory tax rates are similar to those in other OECD economies, with the exception of social insurance contributions, which, at 43 percent, are among the highest. However, properly taking into account numerous tax deductions, credits and exemptions, and weighting various marginal tax rates by income distributions to obtain effective marginal tax rates has proven to be a daunting task, especially if done in an international comparative context.10 As a result, Mendoza, Razin and Tesar (1993) suggest a methodology for computing effective average tax rates on factor incomes and consumption using national accounts data.11 They constructed these rates for G7 countries, and in comparing their effective average tax rates for the United States with effective marginal tax rates obtained by other authors, they conclude that their rates are within the ranges of marginal tax rate estimates and that they display very similar trends. They also find that their estimated tax rates are generally consistent with some key predictions of equilibrium models. In particular, in the majority of countries, the saving rate was inversely related to the tax rate on capital income, the average number of hours worked was negatively correlated with the sum of labor and consumption taxes, and the rate of unemployment was positively correlated with the labor income tax rate. Finally, the investment rate was also inversely related to the capital income tax rate.12

19. Table 3 contains data on simple averages of aggregate effective average tax rates for 24 countries for which sufficient data are available, as well as for Greece. Greece’s consumption rate is found to be higher than for any of the averages presented, although this masks a good deal of variation among other countries.13 This reflects a higher reliance on both VAT, which was introduced in 1987, and excises than in other EU or OECD countries.14 The reason for the higher revenue yield appears to be that it has a relatively “clean” VAT structure, with no exemptions for privileged sectors (apart from the standard exemption for firms below the threshold sales level, as in all VAT systems). The range of specific sectors to which a lower rate is applied, namely to books, culture, food, medicine and newspapers, also appears to be narrower than in other most other OECD VAT structures. Similar differences appear concerning excises. They accounted on average for 7 percent of GDP and 17.4 percent of total tax revenues during 1990–94 while these figures were 4.9 percent of GDP and 12 percent of total revenues in the EU and 4.6 percent of GDP and 12.5 percent of GDP among OECD economies on average.15

Table 3.

Greece: Effective Average Tax Rates, 1990-95 Average

(In percent; unless otherwise specified)

article image
Sources: OECD National Accounts (detailed tables); OECD Revenue Statistics; United Nations National Accounts; OECD Analytical database; GFS1996 Yearbook; WEO; and staff calculations. Note: All aggregate figures are weighted by relative GDPs at 1995 PPP US$.

Excluding Ireland and Luxembourg for data reasons.

Excluding the Czech Republic, Iceland, Ireland, Luxembourg, Mexico, New Zealand, Poland and Turkey for data reasons.

Includes the countries listed in footnotes 4–6.

Includes: Colombia, Hungary, Korea, Malta, Mauritius, Portugal, Spam and Thailand.

Includes: Australia, Belgium, Finland, Germany, Italy, the Netherlands, Sweden and the United Kingdom

Includes: Austria, Canada, Denmark, France, Japan, Norway, Switzerland and the United States.

In l995 PPP US$.

20. Although reliance on consumption taxes has risen among OECD member economies in general (while excise yields have remained stable as a share of GDP), they appear to have increased in Greece, at least since 1980.16 Moreover, while taxes on final consumption are generally thought to be less distortionary than taxes on factor incomes, they nevertheless distort the labor/leisure choice, inducing less labor supply. Therefore, there appears to be little scope for additional revenues from further indirect tax rate increases.

21. Greece also has a much higher effective average tax on labor income compared to economies with similar incomes, and even exceeds the OECD average.17 Initially, the latter comparison is quite surprising in light of the low share of direct personal taxes in GDP as shown in Table 2. The apparent inconsistency is resolved, however, by noting the slightly higher share of social insurance contributions and, more importantly, by noting that the share of wage income in GDP is much lower than the average in the OECD, while the shares of household operating surplus and property and entrepreneurial income are correspondingly much larger. This reflects a larger share of self-employed (even excluding the continued relative importance of agriculture), and a smaller share of dependent employment in the labor force relative to most other OECD economies.18

22. As seen in Figure 6, labor income taxes have increased over the past 30 years, and quite sharply since 1990, reaching the EU average in 1994 (the latest year for which data are available). Reforms of the social security system began in 1990, with increased contributions by private-sector employers and employees. In addition, new civil servants were thereafter required to pay employee contributions. A further round of reforms was introduced in 1993, in order to address the rising costs of the increasingly insolvent social expenditure system, with an additional increase in contributions. Despite these increases, contribution evasion continues, which would otherwise result in even higher effective labor tax rates (OECD 1997b). Health and unemployment insurance contributions were also raised in 1993.19 In May 1994 the highest marginal personal income tax rate was raised from 40 to 45 percent. In order to curb tax avoidance, the authorities instituted in 1995 a system of presumptive taxation on income from agriculture, the self-employed and from unincorporated businesses.20 Levies for farmers are determined on the basis of the crop and size of land being cultivated, while for the latter two groups the tax base depends on the national minimum wage, commercial property rents, location, number of years in operation, the type of university degree (for professionals), and the sector.21 The rates of presumptive taxation were increased in both 1996 and 1997, although separate revenue data from this source are not available. Another factor raising the effective labor income tax rate has been the failure to adjust personal income tax brackets for inflation since 1993, while the cumulative increase in consumer prices has been 50 percent through the end of 1996; nor were the brackets adjusted in 1997. Also in 1997, the government limited or closed a large number of tax expenditures, which along with presumptive taxation has helped to reduce the inequitable burden between the salaried and self-employed, but also further increased the labor income tax rate.

Figure 6.
Figure 6.

Greece: Effective Average Tax Rates, 1965-94

(In percent)

Citation: IMF Staff Country Reports 1997, 089; 10.5089/9781451816075.002.A001

Sources: OECD Analytical databank; United Nations National Accounts; IMF, World Economic Outlook; and staff calculations.

23. The discussion above clearly indicates that the average effective tax on labor income has risen significantly since 1994. Zee (1996a) reviews the nexus between unemployment and taxation. Mendoza, Razin, and Tesar (1993) found econometric support that higher taxes on consumption and labor income are associated with lower hours worked, and that higher labor income taxes are associated with higher unemployment.22 While data on hours worked are not available, there has been a clear increase in the unemployment rate since 1990, coinciding with the sharp rise in labor income taxes.23 Undoubtedly other factors have also affected the labor market as well (including rising real wages and an increase in labor shedding in the manufacturing sector in the last few years), and careful econometric work would be necessary to disentangle the various influences.24

24. In contrast to labor taxes, the effective average tax rate on capital is in line with its comparator economy average, while significantly lower than the OECD and EU aggregate rates.25 This reflects the large share of self-employed and those working for small enterprises, as well as poor tax compliance. The presumptive taxation system discussed above was introduced in order to more folly tax households’ operating surplus and property and entrepreneurial incomes. In order to generate further revenues, a number of other capital income tax measures were also adopted in recent years. In May 1994 the standard corporate income tax rate was increased from 35 to 40 percent. At the same time, the 15 percent withholding tax on interest and dividend incomes was extended to repos and dividends from mutual funds, and was further extended in January 1997, albeit at only a 7½ percent rate, to income from government securities. A reduction in tax expenditures through the limitation and closing of exemptions and deductions in 1997 further increased the effective average capital tax rate.

25. In light of these comparisons, there appears to be little room for generating significant additional revenues through further increases in tax rates without adversely affecting efficiency and growth prospects. The marginal personal income tax rate facing the average production worker was, at 5 percent in 1993, admittedly low, but has undoubtedly risen given the subsequent failure to adjust tax brackets for inflation. Moreover, especially in light of high rates for social insurance contributions and the continued disproportionate burden on salaried employees, any further revenue efforts should focus on improving tax compliance and administration, bringing those individuals that are presently subject to presumptive taxation into the general tax system (see OECD (1996e)).

D. Expenditures

26. This subsection examines two components of current expenditures in more detail, government consumption and transfer payments, in order to identify areas in which fiscal consolidation could focus.

27. Government consumption expenditures as a share of GDP are only slightly higher than in Greece’s comparator economies, though lower than in the EU and OECD (Table 2). It is nevertheless interesting to examine the composition of these expenditures across different functional classifications (Table 4). Although care must be taken in forming conclusions due the fact that the for Greece national accounts-based government consumption data are derived from an old methodology, and due to the relatively small number of countries in some averages, two points are worth noting.26

  • The share of consumption expenditures devoted to general public services (i.e., excluding health, education and other specific tasks) is much higher than in comparator averages. Christodoulakis (1994) reports that the share of civil servant employment in total employment rose from under 14 percent in 1980 to almost 20 percent by 1990, and the number of civil servants have increased by a further 7.5 percent by end-1996, compared to a 4.2 percent increase in total employment.

  • The share of consumption expenditures spent on defense, estimated at about 3¾ percent of GDP,27 is much higher than in all comparator averages (Hewitt (1991, 1993) and Happe and Wakeman-Linn (1994)).

  • In contrast, Greece devotes a relatively low share of public resources to education. Education has always been held in high regard, and by standard measures it is a well-educated country.28 However, the share of public resources presently devoted to education, once adjusted for the upward revision in GDP, is lower than in economies with similar income levels, and sharply lower than in the EU and OECD.29 Although it allocates a relatively small share of public resources to education in total, a relatively high proportion of this is devoted to higher education (Patrinos, 1992). The low level of public expenditure is in part offset by large private expenditures, estimated at 2¼ percent of GDP (OECD 1996b). The bulk of these comprise expenditures for supplemental tutorials (frontisterid) required to improve the likelihood of obtaining one of the limited positions at university. A survey conducted by Kostakis (1990) reveals that direct private educational costs averaged 8.1 percent of family income for upper-secondary General Lycea, and averaged 2.1 percent of family incomes for Technical-Vocational Lycea.30

Table 4.

Greece: General Government Final Consumption Expenditures, 1990-94 Average

article image
Sources: OECD Analytical databank; United Nations National Accounts; and staff calculations. Note: All aggregate figures are weighted by relative GDPs at 1995 PPP US$.

Average of 1990-91 figures.

Includes: Austria, Denmark, Finland, France, Italy and Spain, Sweden; other member states, except Greece, excluded for data reasons.

Includes: EU, plus Australia, Iceland, Japan, New Zealand and Norway; other member states, except Greece, excluded for data reasons.

Includes the countries listed in footnotes 5-7.

Includes: Colombia, Korea, Malta, Mauritius Slovenia and Spain.

Includes: The Bahamas, Finland, Iceland, Italy, New Zealand and Sweden.

Includes: Austria, Denmark, France, Japan and Norway.

Includes housing and community amenities, and recreational, cultural and religious affairs.

Includes other functions.

In 1995 PPP US$.

28. A second point about the pattern of expenditures is the significantly higher share of transfer payments (Table 2 and Figure 5). This reflects in part cultural and geographic differences between Greece and other comparator economies. The social insurance system, like its counterparts in other European nations, has been in existence longer than in most non-European economies. In addition, rising transfer payments have required increasing social insurance contributions, with potential harmful efficiency implications.

29. Over 95 percent of transfer payments are to households, covering a variety of social expenditures. The OECD (1996f) has recently compiled comparable data on public social expenditures among its member countries (Tables 5–6). A number of important features are apparent.31

  • The share of social expenditures in GDP has risen significantly, by 6.4 percentage points of GDP since 1980. While beginning at a relatively low base, the cumulative growth since 1980 has been exceeded by only two other OECD economies. The sharpest increase occurred in the early 1980s, with a 3.5 percentage point of GDP increase in old-age cash benefits, following liberalization of eligibility requirements and increase in real pension benefits, and with a 1.7 percentage point of GDP increase in disability and survivors’ benefits. More recently, unemployment compensation expenditures have risen significantly, reflecting restructuring efforts by manufacturers. Nevertheless, they remain less than 1 percent of GDP, an amount smaller than in any OECD economy except the United States.

  • With one exception, Greece is the only country to devote more than 50 percent of total social expenditures to old-age cash benefits—only three other OECD economies devote more than 40 percent of total expenditures to this category. This unusual pattern reflects a number of factors: the slightly higher share of the elderly in the population,32 generous eligibility requirements for pension entitlements, strong incentives for early retirement and, in the case of civil servant pensions, no restrictions of continued employment in the private sector, and the fact that the pension system is relatively speaking the most generous in the EU, with some after-tax replacement ratios exceeding 100 percent.33 Moreover, the pension system is fragmented into more than 300 separate funds, with poor administrative control. As a result, the OECD (1997b) reported that there is extensive abuse of the system.

  • The share of social expenditures devoted to disability pensions is significantly higher than EU and OECD averages. This reflects both a effective liberalization of the system in the early 1980s, as well as past systemic abuse.34

  • In contrast to the high share of public expenditures devoted to pensions, the share devoted to health care is among the lowest in the OECD. This result is somewhat surprising given its more elderly population and the well-known positive relationship between increased health care and aging. The reason for this pattern is possibly the result of an increasing share of health services being provided by the private sector.35

Table 5.

Greece: OECD Total Social Expenditures, 1980-93

article image
Sources: Social Expenditure Statistics of OECD Members Countries (Provisional Version), Table 1.1; and staff calculations. Note: All aggregate figures are weighted by relative GDPs at 1995 PPP US$.

Excludes Greece, and, for lack of data Germany.

Excludes Greece, and, for took of data, Germany, Mexico end Switzerland

Table 6.

Greece: OECD Social Expenditure Shares, 1993 Social Expenditure Shares, 1993

(Share of total)

article image
Sources: Social Expenditure Statistics of OECD Members Countries (Provisional Version), Table 3; and staff calculations. Note: All aggregate figures are weighted by relative GDPs at 1995 PPP US$.

Data are for 1992.

30. At present, the social security system is actuarially in deficit. Despite high statutory contribution rates, evasion and overly generous pensions relative to contributions have resulted in operating shortfalls of about 5 percent of GDP in recent years, which have been financed through earmarked taxes and fees, and through central government transfers. Moreover, the population is projected to age rapidly in the coming decade, placing further strain on the system.36 37

E. Successful Fiscal Adjustments

31. Recently there have been a number of studies examining similarities among “successful” and “unsuccessful” fiscal adjustments. Definitions of “success” have varied slightly, but all have included a sizable and sustained reduction in the overall or primary deficit- and/or the debt-to-GDP ratio (Alesina and Perotti (1995b, 1996) and McDermott and Wescott (1996)). While it is important to consider the role of related or at least simultaneous economic influences in attaining success,38 it is possible to distill general conclusions that appear to have emerged from the literature, and, as importantly, to examine where the fiscal adjustment to date falls in the successful/unsuccessful dichotomy. The most important, and most robust, conclusion concerning those factors relating to “successful” adjustments concern the distribution of fiscal consolidation between increased revenues and reduced expenditures. All of the studies to date have found that “successful” fiscal adjustments rely primarily on expenditure reductions, and, more specifically, on reductions in the government wage bill and on transfers.

32. Alesina and Perotti (1995b) identify a significant asymmetry in the behavior of expenditure and taxation that emerges between fiscal expansions and contractions among OECD countries in the 1970s and 1980s. In the former, there were virtually no changes in taxes, with significant expansion in government spending (especially transfers and government wages), while in the latter, both spending was cut and taxes were increased. “Successful” fiscal adjustments, however, are characterized by large cuts in transfers and in government wage bills. Moreover, they find that reducing government employment, rather than government wages, is likely to have longer-lasting effects because they send a stronger signal concerning their determination to effect a significant fiscal adjustment.39 They also find that very tight fiscal policies initiated in nonrecession years are twice as likely to be successful as those initiated during recessions.40 Finally, the authors find that major fiscal adjustments do not cause major recessions.41 Among the possible reasons are that: tax increases can have expansionary effects if they induce expectations of less dramatic and disruptive future tax increases; fiscal consolidation, particularly if of sufficient size in economies with high debt ratios, may have important credibility effects on interest rates by reducing risk premia (which then crowds in investment and durable goods consumption); and, that changes in the governments wage bill or in the level of labor taxation in economies that have unionized labor markets affects firms’ unit labor costs and competitiveness.

33. Table 7 compares the changes in revenue and expenditure shares in successful and unsuccessful fiscal adjustments, as defined by Alesina and Perotti (1996b), and displays comparable data for Greece during 1993–97.42 Whereas the overwhelming bulk of consolidation in successful fiscal adjustments falls on primary expenditures, especially on government wages and transfers, in Greece these specific components and total primary expenditures have increased. Moreover, about half of the projected 10 percentage point improvement in net lending is the result of higher revenues, especially higher direct taxes. While this does not definitively rule out the ultimate success of Greece’s consolidation efforts were they to continue to focus largely on additional revenues, it would stand in sharp contrast to other successful consolidation efforts. However, recent strong growth, and a rebound in activity in Europe, would bolster the likelihood that additional fiscal adjustment is translated into a sustained improvement in both the debt ratio and in growth prospects.

Table 7.

Greece: Successful and Unsuccessful Fiscal Adjustments

(Percent of GDP; standard deviations are in parentheses)

article image
Sources: Alesina and Perotti (1996b); Greek authorities; and staff calculations.

1993-97.

III. Foreign Currency Risk Premium: Evolution and Determinants43

A. Introduction and Summary

34. Access to foreign lending sources is an important element of the borrowing strategy of any country. Domestic borrowing costs have risen significantly since the liberalization of its banking system, and understanding the evolution and determinants of its foreign borrowing costs is a crucial issue. Looking ahead, given the desire to ultimately enter EMU, the size of the risk premium accruing to Greek foreign currency denominated paper may become even more important.

35. This paper attempts to provide a direct measure of the risk premium (representing credit risk) by collecting and analyzing a new data set of foreign currency denominated fixed rate loans and bonds issued or guaranteed by the Hellenic Republic over the period 1980-96. The data set shows that, as the amount of foreign borrowing increased and macroeconomic fundamentals deteriorated during the 1980s and early 1990s, the maturity of this instrument has systematically declined. Thus, it is not possible to directly compare the spread paid by Greece across time. To correct this problem, this paper estimates spreads for constant maturity issues—the so-called benchmark spreads. Using benchmark spreads, it is possible to meaningfully track the historical evolution of the foreign debt risk premium.

36. An important finding is that the benchmark spread has increased by more than has the actual spread. Based on the observed spread one would misleadingly conclude that the risk premium has experienced no net increase since the early 1980s (a period of relatively high observed spreads). However, the benchmark spread shows otherwise, as it indicates a rise of about 50 basis points. Thus, the loss of macroeconomic stability, the rise in foreign debt, and the worsening of its credit rating, have proved costly.

37. Another finding is that, during 1996, the observed spread has declined below the level suggested by the fundamentals. This effect has also been observed for other “high yielders,” and has been attributed to the greater likelihood of EMU accession perceived by the markets.

38. The paper then tries to identify variables that influence the spread using econometric techniques, both to gauge more precisely the reasons for the increase in the risk premium over the sample period, and to identify factors that could potentially help to reduce it. Macroeconomic variables that proved statistically significant include the ratio of foreign debt to GDP, the primary deficit of the central government as a ratio to GDP, the rate of depreciation of the nominal effective exchange rate, and the level of foreign interest rates (measured by the U.S. dollar 6-month LIBOR). A number of characteristics of the instruments also proved statistically significant, including the maturity of the instruments and their credit rating (as determined by Moody’s rating agency). The estimated effect of all the above-mentioned variables was to increase the spread, with the exception of the primary deficit. The deficit surprisingly appeared to affect the spread negatively: this result is probably indicative of reverse causality, meaning that the deficit expanded because lower spreads allowed more foreign borrowing to take place. A number of other plausible variables did not seem to have a statistically significant effect on spreads. These included international reserves, the non-interest current account, the issuer, and the loan amount.

39. The “hard drachma” policy is seen to have a beneficial effect on interest rate spreads: between 1985, when the drachma was devalued, and 1996, the stabilization of the drachma has reduced spreads by about 25 basis points. On the other hand, although foreign debt can appear cheaper than domestic debt under conditions of exchange rate stability, a higher ratio of foreign debt to GDP tends to increase the credit risk component of interest rates. Between 1990 and 1996, this factor alone contributed about 20 additional basis points to the spread. The results also appear to suggest that the central government primary deficit as a ratio to GDP has tended to expand with cheaper availability of foreign funds: this strategy could not be continued under a convergence program which includes quantified fiscal targets.

B. The Measurement and Evolution of the Spread

40. The starting point for the decomposition of interest rate spreads is usually a risk-adjusted version of the uncovered interest parity condition, which can be approximately written as:

i t − i t * = E t ( S t + τ / S t ) + π t + ∊ t ( 1 )

where I denotes the yield on a domestic instrument and I* the yield on a comparable foreign instrument of maturity τ, Et (St+τ/St) the expected rate of depreciation of domestic currency vis a vis the foreign currency over a time period τ using information available as of time t, π the risk premium—by definition positive for a “high-yielder,” and ∊ an error term that is assumed to be orthogonal to all relevant information available as of time t. Typically, most of the extensive literature that attempts to estimate versions of equation (1) considers π to be unobservable, and relies on more or less sophisticated econometric methods to estimate it. However, a major problem with this method is that expected depreciation is also unobservable, which forces researchers to attempt to separate two unobservable components from each other, with results that tend to be even more assumption-dependent than usual.

41. This paper will instead measure π directly, by focusing on a specific subset of instruments for which expected depreciation is not a factor—namely, foreign currency denominated paper. In addition to its intrinsic interest, this exercise would clearly facilitate future attempts to estimate equation (1), as it would no longer be necessary to simultaneously disentangle two unobservable components. For many countries, and especially those countries for which π would tend to be the largest in size, this exercise is typically difficult to carry out because of the thinness of the relevant financial markets, the absence of formal secondary markets, or even the unavailability of appropriate instruments. Greece is an interesting case precisely because even though it is a “high-yielder,” it has been issuing foreign currency denominated instruments since 1980. Although a formal secondary market with significant trades over these instruments does not exist, the large number of issues allows the construction of a spread series—measured at issue. Because of the relatively large number of eligible issues (numbering about 100) it will also prove possible to perform an econometric analysis of the determinants of the spread.44

42. Foreign currency denominated instruments represent a promise by the issuing authority to repay a given amount, plus interest, in foreign currency. Because the proceeds of the loan are used domestically, the issuer assumes all the foreign exchange risk: this is the opposite of the typical transaction, where a foreign investor purchases domestic bonds with the intention of converting the proceeds back to the foreign currency at some later date. Thus, for these instruments, expected depreciation is no longer a factor directly affecting return.45

43. Figure 7 depicts the evolution of the foreign currency denominated spread, measured at issue, for the period 1980–96. In general, the spread is small—almost never larger than 100 basis points. Other, often less developed, countries find they have to pay a spread in the hundreds of basis points, even when they are willing to guarantee repayment in foreign exchange. This phenomenon, where countries belonging to the EU generally pay a smaller spread than other economies that appear to have equally favorable macroeconomic conditions, has not gone unnoticed. Conventional wisdom has ascribed it to the implicit “bailout clause” that markets believe applies to EU members.

Figure 7.
Figure 7.

Greece: Evolution of the Spread and Maturity, 1980–96

Citation: IMF Staff Country Reports 1997, 089; 10.5089/9781451816075.002.A001

Sources: Eurocurrency Loanware and Noteware, Eurocurrency Bondware; and Fund staff estimates.

44. The behavior of the spread across time can be grouped in several distinct periods. At the beginning of the sample, which coincides with Greece’s entry in the EU (in 1981), spreads were generally low, at around 50 basis points. After 1985 the spreads increase noticeably, often reaching 75 basis points. Toward 1989 spreads appear to plunge substantially, in some cases falling as low as 25 basis points. After 1991, when Greece’s foreign exchange debt was downgraded by Moody’s rating agency from Baal to Baa3 (the lowest investment grade rating), spreads increased significantly, to 75 basis points or more. Finally, during 1996, spreads appear to have again declined significantly, following expectations that Moody’s would upgrade the rating of the foreign exchange debt.

45. Looking at the raw spread can be very misleading, however, because other factors that affect the spread need not have remained constant. Figure 7 also shows that as spreads increased, Greece felt it necessary to reduce the maturity of the issued instruments.46 During the early 1980s maturities ranged between 8—12 years. In the latter half of the 1980s maturities were very rarely more than 8 years long. During the 1990s maturities never exceeded five years. Thus, the actual increase in the spread underestimates the true increase in the costs faced by Greece when issuing foreign currency denominated debt.

46. In looking for a first explanation for the increase in the spread, it is natural to examine a number of key macroeconomic variables, grouped in Figure 8. There, it is found that the nominal effective exchange rate was depreciating throughout most of the sample period, with a big spike coinciding with the 1985 devaluation (which was a part of a stabilization program effected by the authorities under the auspices of the EU). The exchange rate stabilized only after 1992, when the Bank of Greece adopted a “hard drachma” policy. Following EU accession the primary deficit (excluding EU transfers) grew significantly, until the fiscal reforms of the early 1990s finally made an impact. Throughout the period the foreign exchange debt also continued to increase, with a pause following the 1985 stabilization program. The next section will present econometric evidence of the impact of these and other relevant variables on the spread.

Figure 8.
Figure 8.

Greece: Macroeconomic Developments, 1980–96

Citation: IMF Staff Country Reports 1997, 089; 10.5089/9781451816075.002.A001

Sources: Greek authorities; and Fund staff calculations.

C. The Determinants of the Spread

47. The previous section provided an outline of the evolution of the spread and some of its plausible determinants. However, the spread will in general be affected by a host of variables, and their impact can only be disentangled and evaluated using econometric methods. Unfortunately, economic theory provides little explicit guidance on which variables will prove the most promising.47 In general, there will be three sets of variables that can be expected to affect the spread:

  • • Variables intrinsic to the issued instruments, and to the financial markets where they are traded. Such variables include the maturity of the instruments, the amount of the loan, whether they possess option (derivative) clauses, their credit rating, the issuing authority, whether they are bonds or syndicated loans, the fees paid, and the underwriter country and currency in which they are issued.

  • • Variables related to the country’s ability to pay, together with general macroeconomic variables. Such variables include the level of foreign debt (suitably normalized), the level of foreign exchange reserves, the exchange rate, the foreign interest rate, and the government deficit. Some of these variables may only have a signaling effect, in that they may have no direct impact on ability to pay but may raise questions about the authorities’ credibility and willingness to pay.

  • • Variables related to demand factors. It is widely believed that the markets go through phases where they engage in a wholesale reevaluation of the prospects for repayment of foreign exchange debt (often prompted by some event, like Mexico’s 1982 debt crisis, and again following Mexico’s recent devaluation and emergency borrowing from the United States and IMF sources). Thus, it is possible that spreads could change with no apparent change in any fundamental.

48. The data set collected for the purpose of this paper has extensive information on variables intrinsic to the issued instruments.48 Unfortunately, not all relevant macroeconomic variables are available at the same frequency: variables such as the foreign exchange debt and the government deficit were therefore interpolated to a monthly frequency. Finally, there was no easily available variable capturing global demand factors.49 With these preliminaries, an empirical investigation was carried out to see which variables from those listed in the above paragraph had a statistically significant impact on the spread. To organize the presentation of the results, the preferred specification is collected in the following tabulation, which also presents a series of diagnostic statistics.

Determinants of the Spread

(Regression Output)

article image
Key: MAT=maturity in years; MAT2=maturity squared; DOPTION1=dummy for instruments with option clauses; D12E=annual rate of NEER (nominal effective exchange rate) depreciation; GBPRI_EU=central government primary deficit excluding EU transfers as percent of GDP; GCD_FGDP=foreign currency public debt as percent of GDP; LIBOR6M=six month dollar LIBOR; CRAT1=dummy for credit rating (1 after lowered).

49. The results are for the most part conventional and statistically and economically significant. Longer maturity is consistent with higher spreads. Here, a quadratic approximation is used to capture the curvature of the maturity effect. A higher rate of depreciation of the NEER is consistent with higher spreads: this effect is likely related to the ability to pay (since depreciation would create adverse valuation effects for the foreign exchange part of the public debt), while also serving as a signal of the government’s ability to manage the economy. Similarly, a higher foreign interest rate is consistent with higher spreads, as is higher foreign currency debt as a percent of GDP.

50. The result on the significance of Moody’s credit rating is interesting. The coefficient has the correct sign, and is quite large, although imprecisely measured in this specification. There is some controversy in the literature that discusses the effect of credit rating agencies on the pricing of corporate debt: it is not obvious whether such a rating should have any effect additional to that of the fundamentals. The argument that it does not is that since the market has the same access to fundamentals as do the rating agencies, the ratings should have no additional information. The opposite argument is that since there are costs to collecting the information, ratings do have independent value. When the entity under review is a small country, one can argue that the information collection aspect of ratings is even higher than it is for domestic corporate entities. Consequently, it makes sense that Greece’s rating appears to affect its spread.

51. The result on the option clauses is included in the regression output, even though it is economically and statistically insignificant in this specification, because it was economically and statistically significant for some subsamples. There were relatively few issues with option clauses, and the data did not include information that would allow the pricing of the embedded options directly. For those subsamples, it was found that the existence of an option clause would reduce the spread by as much as 25 basis points. Thus, it is likely that options were occasionally included to sweeten the deal for the underwriters, consequently generating contingent liabilities that were off of the government’s balance sheet. This effect disappeared over the larger sample, because no option clauses were reported over the past few years.

52. A final effect that deserves mention is the counterintuitive result whereby a higher deficit is consistent with lower spreads. This effect holds even after subtracting interest payments from the deficit (to purge spurious correlation) and EU transfers (which are unrelated to government policy and have no signaling value as to the government’s credibility). The result can be interpreted, however, if reverse causality is considered. During most of the sample Greece was running high deficits, which were difficult to finance from domestic sources (especially after the financial system was liberalized, which took away the state’s ability to directly appropriate financial savings for its needs). Under such conditions, it is reasonable to consider that the government was liquidity constrained. Thus, it is possible that the government would resort to external financing to a larger extent whenever, and for whatever reasons, its spreads were lower. To examine the plausibility of this hypothesis, an auxiliary regression was run relating the deficit variable to the lagged spread. It was indeed found that a lower lagged spread was related to a higher current deficit, which is consistent with the interpretation of liquidity constraints advanced here.

53. It is also interesting to consider those variables that were examined and found statistically insignificant. The issuer did not seem to matter, since a dummy variable for the Bank of Greece (arguably the most credible issuer) was statistically insignificant. International reserves were also insignificant. Bonds were more expensive to issue than other types of issue, but this result held only for some specifications. There was some evidence that issuing in yen was cheaper for Greece. This is reasonable, given that Greece has developed good long term relations with Japanese underwriters. Unfortunately, most of the issues in yen had to be dropped from the sample because they lacked some key data. It is likely that, if these issues could have been included, the spread in yen would have been lower by a statistically significant degree.50 The amount of the loan was also insignificant, a result that makes sense given the deepness of global financial markets. Finally, the fee paid by Greece proved insignificant. This could be due to one of two things: either the fee paid was exogenous, determined primarily by market conventions, or the fee data was problematic (a proxy had to be constructed for the fee variable, since total fees paid were not available).

54. Using the regression presented earlier, Figure 9 exhibits the estimated spread against the actual spread. It is seen that the estimated spread tracks the actual spread quite well, even during some apparently puzzling dips in the period 1994–95. An interesting exception is 1996, where it appears that the actual spread is lower than the estimated spread by more than 25 basis points. This phenomenon has been observed also for some other high-yielders, where it was found that equations explaining interest rate differentials have been over predicting since early 1996. The explanation given by most observers is that the bond markets have come to believe that EMU will happen, and that the high-yielders are now expected to join. It is interesting that a similar phenomenon can be observed for the case of Greece, which is not expected to join EMU as part of the first wave of countries.

Figure 9.
Figure 9.

Greece: Actual Vs. Predicted Spread and Residual, 1980–96

Citation: IMF Staff Country Reports 1997, 089; 10.5089/9781451816075.002.A001

Sources: Eurocurrency Loanware and Noteware, Eurocurrency Bondware; and Fund staff estimates.

55. As earlier indicated, perhaps the most problematic aspect of the regression specification presented above is the absence of a global demand variable. To ascertain the extent of this problem, a number of alternative time series specifications were experimented with, on the assumption that such a variable would likely exhibit positive autocorrelation properties (over the frequencies considered in this study). The residuals are shown in Figure 9, which provides some weak evidence in favor of this assumption. It is clear from that chart that a good amount of the residual variation can be attributed to isolated issues (no more than five in number), which is not surprising given the variety of issuers and currencies in the data set. A specification was tested that included a lagged spread variable, but that variable proved statistically insignificant. On the other hand, a MA(1) specification tended to overcorrect for autocorrelation, resulting in an even worse D-W statistic than the preferred specification.51 Thus, it is likely that a more satisfactory treatment of the global demand conditions will have to wait until an appropriate sample can be collected.

D. Constructing Benchmark Spreads

56. Previous discussion has made it clear that the actual spread has some serious shortcomings as an indicator of the true cost of foreign currency denominated debt because the maturity of the issued instruments has not been held constant over time. For countries with large and liquid financial markets constructing benchmark yields is a routine matter, involving little more than splicing together the yields of bonds with the desired residual maturity. However, for the case of Greece available data simply do not allow the use of this method.

57. Using the estimated regression, however, it is possible to construct benchmark spreads that better capture the changes to Greece’s cost of funds over time. This involves constructing estimated spreads at different maturities. Figure 10 presents the actual spread, together with benchmark spreads at 1–, 5–, and 10–year maturities. It is clear from the figure that Greece’s true cost of funds has increased by significantly more than the evolution of the actual spread would indicate. The evolution of the benchmark spreads indicates a significant decline as the markets first became familiar with Greece in the period 1980–82, a rise following increasing macroeconomic uncertainty in the period 1982–85, a decline following the stabilization program of 1985 and until 1990, and finally significant increases in two steps, which raised the benchmark spreads by almost 50 basis points in the period 1990–96.

Figure 10.
Figure 10.

Greece: Actual Vs. Benchmark Spread, 1980–96

Citation: IMF Staff Country Reports 1997, 089; 10.5089/9781451816075.002.A001

Sources: Eurocurrency Loanware and Noteware, Eurocurrency Bondware; and Fund staff estimates.

58. The construction of the benchmark spreads, aside from its intrinsic interest, also makes it easier to interpret the movements of the cost of funds across time. The two-step increases observed during 1990–96, which were not visible in the evolution of the actual spread, clearly coincide with similar step increases in the foreign currency debt variable (already presented in Figure 8). Thus, the increase in foreign borrowing over the period 1990–96, which must have seemed significantly cheaper than domestic borrowing (at least given some stability in the exchange rate), carried with it rather large costs—an additional 50 basis points in interest rate costs per unit borrowed.

59. The limitations of the above analysis should also be kept in mind. It is possible that the linear model that was estimated in order to explain the determinants of the spread could break down at higher levels of foreign currency debt. It is also possible that the true spread/maturity relationship, which for longer maturities could only be estimated at relatively low levels of foreign currency debt (early in the sample period), is significantly worse than the estimated relationship for longer maturities. To that extent, Greece might have to pay a spread exceeding 150 basis points in order to borrow at 10-year maturities, even if Greece were willing to guarantee the foreign exchange risk (which would be quite large over such a long maturity).

ANNEXI DATA SOURCES AND SOME DATA ISSUES

60. The source for the raw data is Euromoney Loanware & Noteware ©, and Euromoney Bondware ©. This data is proprietary, and the contract under which it is provided precludes its being redistributed. For each issue, extensive information is provided, including: date of issue, borrower type, type and amount of the loan (also expressed in US$ if the currency of issue is not the US$), maturity, interest rate spread, underwriters, fees, and miscellaneous information (e.g., whether the issue had an option clause attached). The issues were screened, and only those issues that were fixed term and had comprehensive information were included.

61. Unfortunately, issues in yen were disproportionately represented among those issues that had to be excluded from the sample, because of missing information. This made it impossible to test the hypothesis that yen spreads were lower than other currency spreads, although the partial information available suggested that this hypothesis is true.

62. Also, information on fees was presented in such a way that it was impossible to construct a variable that exactly matched the true cost of each issue. Typically, fees belonged to different types, and were often presented in tranches. Since the information did not reveal the actual amount sold per tranche, it was not possible to compute the actual fees paid. An attempt was made to construct a fee variable by averaging over tranches (implicitly assuming that the distribution of tranches was uniform). Obviously, the constructed variable is at best a proxy for the true fees paid to managers and underwriters.

63. Some early issues had multiple maturities and spreads. For such issues, the longest maturity was selected, on the grounds that it is the longest maturities over which information is hardest to collect.

64. One data point was collected per month. For some months, more than one issue was provided. For these cases only one issue was included in the sample. The criteria of selection included maturity and amount, which were lexicographically applied in that order.

65. Since the data were collected at issue, not every month is represented in the sample. Because some of the explanatory variables were intrinsic to each issue, the data was inherently asynchronous. Thus, a number of techniques that are often used for the processing of asynchronous/unobservable data (e.g., the Kalman filter) could not be applied here.

IV. The Lending-Deposit Spread and its Tax-Related Determinants52

A. Introduction

66. The lending-deposit interest rate differential is generally considered one of the key indicators of the health and competitiveness of a banking system. Greece, with its legacy of large, high operating cost state-controlled banks, has had lending-deposit spreads that have been high relative to those of other EU countries. Over the past few years the lending-deposit spread has shown considerable variability.

67. This note focuses on the tax-related determinants of the lending-deposit interest rate differential. Taxes on banks are relatively high, and could therefore partially explain why the lending-deposit spread is relatively high. Moreover, the applicable tax rates have shown considerable variation over time, so that changes in the lending-deposit spread may reflect their influence rather than changes in market conditions or other factors.

The main findings are:

  • Tax factors have accounted for a significant part of the lending-deposit spread, ranging from 30 to 40 percent of the spread.

  • The direct impact of tax factors has generally declined over time.

  • The part of the spread not accounted for by tax factors has generally been much more stable than the observed spread, and in early 1996 fell to levels close to those of other high-spread EU countries.

  • Since the middle of 1996 the spread has increased to the point where Greece is again an outlier in the EU. This rise is largely unexplained by tax factors.

B. Recent Developments in the Lending-Deposit Spread

68. The lending-deposit spread became relatively high following the liberalization of the banking system in 1991–92, hovering at around 10 percent (Figure 11). It started coming down significantly during 1995, falling from 9.1 percent at end-1994 to 6.5 percent at end-1995.53 In the first half of 1996, there was a slight increase in the spread, which reached about 7 percent. At these levels, the lending-deposit spread was no longer a striking outlier for the EU. For example, the Italian spread was 6.5 percent, while that of Ireland was 5.7 percent and that of Portugal 5.4 percent. Typical EU mid-range values ranged from 4.1 percent for the Netherlands to 4.6 percent for Belgium, while a number of other countries hovered lower, at around 3 percent (for example, Spain).

Figure 11.
Figure 11.

Greece: Lending-Deposit Interest Rates

(In percent)

Citation: IMF Staff Country Reports 1997, 089; 10.5089/9781451816075.002.A001

Sources: Greek Authorities; and Fund staff calculations.

69. At around the middle of 1996 the spread began to rise again, reversing the trend described above. By early 1997, the spread reached some 9 percent, a trend opposite from that observed in the EU. Indeed, the Italian spread fell to 5 percent, the Irish spread remained steady at 5.7 percent, the spread in Portugal fell to 4.6 percent, in the Netherlands to less than 3 percent, in Belgium to 4.3 percent, and in Spain to 2 percent.

70. Although this note focuses on the banking system tax factors, two other factors that could account for the recent rise in the spread might be mentioned here. One is the significant provisioning for past losses undertaken by Greek state banks since 1996.54 The other is the imposition of a 7.5 percent tax on interest earned from government paper, effective in January 1997. As government paper is a close substitute for deposits, whose tax rate has remained unchanged at 15 percent, the lower after-tax interest available on deposit substitutes may have allowed a significant reduction in deposit rates without a commensurate reduction in lending rates.

C. Tax Factors Influencing the Lending-Deposit Spread

71. Earlier studies emphasized the noncompetitive market structure of the banking system, its high operating costs, and significant non performing state sector loans in explaining the spread. This study focuses instead on a number of tax-related factors that affect Greek banks. These factors include the following:

  • Required reserves now comprise 12 percent of deposits (a relatively high percentage, given that many EU countries have a required reserve ratio of less than 2 percent), and are not fully remunerated.55 Over the sample period, required reserves increased from 9 to 12 percent.56

  • Loans are also subject to a tax-at-issue (EFTE—now 4 percent).57

  • Banks are subject to capital adequacy ratios.

    This information will be used in the following section to provide a quantitative estimate of each tax-related factor on the lending-deposit spread.

D. Decomposing the Lending-Deposit Spread

72. The lending-deposit spread can be decomposed into its constituent components (with separate contributions from each tax related factor) using the “break-even” equation for lending rates. Let X denote a nominal interest rate or a return, S a share, τ a tax rate, p the probability of default on a loan, L a loan, D a deposit, E the EFTE tax (a transaction tax on loans), R required reserves, and K capital adequacy. Then, the equation for the break-even lending rate, denoted by X*L, can be written as:

(1-p)(1+SK-SR)(1-τE)X*L + SR XR = XD + SK XK

This equation provides the lending-rate that would be needed so that the bank could break even given that it has to provide required reserves, and that it has to raise funds by paying depositors or by paying equity rates.

73. X*L can be solved for and compared to the actual lending rate, with the residual denoting that part of the spread that is not explained by the tax-related factors. The information and assumptions made in order to calculate the various contributions to the spread include the shares and tax rates on a monthly basis, and data on lending and deposit rates. The probability of default is not known: in the calculations, a 5 percent default rate was used, interpreted as a precautionary provisioning by the banks. The other return which is unavailable is the return to equity capital. For the calculations, it was assumed that the return on equity capital was 10 percent higher than the deposit rate.

74. The solution for X*L obtained thereby is shown in Figure 12, which also shows the residual between the break-even and the actual lending rate. Because the solution is nonlinear in terms of the variables and parameters of interest, there is no simple decomposition of its determinants in terms of individual tax related factors. However, using linearization techniques,58 it is possible to provide the required decomposition of the lending-deposit spread. Letting XL=X*L+∊, with ∊ denoting the residual shown in Figure 12, the decomposition of the actual lending-deposit spread becomes:

Figure 12.
Figure 12.

Greece: Break-Even Lending Rate

(In percent)

Citation: IMF Staff Country Reports 1997, 089; 10.5089/9781451816075.002.A001

Sources: Greek Authorities; and Fund staff calculations.

XL-XD = τE XD+pXD+SR(XD-XR)+SK (XK-XD)+∊

where the first term to the right of the equals sign arises from the EFTE tax, the second from default-risk, the third from required reserves, and the fourth from capital adequacy regulations. Figure 13 collects each of these components and their cumulative contribution to the lending-deposit spread. Because of the linearization technique used, the sum of the components is slightly different from the actual spread. However, the difference is clearly very small.

Figure 13.
Figure 13.

Greece: Lending-Deposit Spread and its Determinants

(In percent)

Citation: IMF Staff Country Reports 1997, 089; 10.5089/9781451816075.002.A001

Sources: Greek Authorities; and Fund staff calculations.

E. Results

75. Looking at Figures 12 and 13 it is seen that the break-even lending rate tracks the actual lending rate quite well over time, with significant turning points coinciding with changes in the tax-related factors. For the period 1993–mid 96, the residual between the actual and the break-even lending rates is relatively constant over time: this should be contrasted with the large decline in the lending-deposit spread. Clearly, the nontax factors that affect the lending-deposit spread (overall market conditions, rates of return on substitute financial products, the oligopolistic structure of the banking system, the high operating costs, and the nonperforming loans in the state controlled banks) have not changed significantly over this period.

76. In fact, during 1993–mid 96, changes in tax-related factors have accounted for a reduction in the lending-deposit spread of about 1.5 percentage points, which is more than half of its observed decline. The reduction in the EFTE tax rate has contributed directly to this outcome while the reduction in the deposit rate has generally more than compensated for the increase in the reserve ratio.59 In terms of the relative contributions between the different tax-related factors, required reserves accounted for the largest part, followed by the EFTE tax. The effect of the capital-adequacy ratio was negligible. Finally, the effect of default risk was considerable, but was based on certain assumptions that would need to be further verified.

77. The increase in the lending-deposit spread since mid-1996 is generally not explained by changes to the tax related factors, despite a reduction in the required reserve remuneration rate in 1997. Factors that are likely to have contributed to this increase include the significant increase in provisions for past bad loans owned by the large state controlled banks during 1996, and the imposition of a withholding tax on interest earned on government paper in January 1997.

V. Recent Developments in the Banking System60

78. The environment in which the Greek banking system operates has been transformed over the last 20 years as formerly pervasive and detailed state controls on interest rates and credit allocation were progressively dismantled. Interest rates on loans were largely liberalized by 1987, and deposit rates by 1989. Requirements on banks to invest in treasury bills and lend to specific sectors were abolished in 1993, although the requirement on banks to redeposit funds placed in the public’s foreign currency accounts with the Bank of Greece was partially relaxed only in 1995. Regulations were issued over time permitting the introduction of a wider range of financial instruments and services, starting with leasing, factoring, and repurchase agreements, and eventually covering a variety of derivative products. Greece accepted the obligations of Article VIII in 1992. International movements in long-term capital were freed of restrictions in 1991, and remaining restrictions on capital account transactions were lifted in May 1994.

79. The scope and pace of the liberalization process have required the authorities to adapt prudential regulations and supervisory practice, and set in motion far-reaching changes in the structure and performance of the banking sector. In this background paper some of these recent developments will be described, starting with the current system of prudential regulation and supervision. The second section discusses the evolution of the structure of the banking market and ownership relationships. The third examines banks’ balance sheets in more detail, and the fourth presents a breakdown of banks’ income, expenses and profitability.

80. The evidence suggests that the Greek system now operates within a regulatory framework that largely conforms to common European practice, and displays signs of responsiveness to market forces, but is still dominated by state-controlled institutions that are burdened by a history of political interference in their operations and credit allocation.

A. The Regulatory and Supervisory Framework

81. The primary piece of legislation governing banking activities in Greece is now the Banking Law of 1992. This law sets out the principal conditions to be met for the establishment and operation of deposit-taking institutions, and gives the Bank of Greece a mandate to issue detailed regulations, license and supervise banks, take enforcement actions, and cooperate with supervisory authorities abroad. It incorporates the relevant EU Directives, notably on the establishment of branches of banks from other EU member states. The law does not apply to certain specialized financial institutions, which are governed by separate legislation and are under the direct control of the government.

82. Licensing requirements and prudential regulations meet EU and BIS standards in all substantial respects. The Capital Adequacy Regulation, which came into force in its present form in November 1996, is the most comprehensive, because it sets out capital requirements for all principal banking activities including foreign and domestic trading positions and off-balance sheet items. Regulations issued in 1992–93 place limits on large exposures, lending to “insiders,” and equity participation in nonfinancial enterprises. However, a number of financial institutions enjoy “grandfathering” derogations, whereby they have been granted a lengthy transition period within which to meet capital adequacy requirements or to reduce holdings in nonfinancial enterprises. Banks’ foreign currency positions and bank liquidity are monitored, but no separate limits are applied. Consumer lending practices are subject to special regulation with a view to ensuring that households do not become financially over-extended. Specific provisions in the Banking Law and regulations require banks to be vigilant in the detection of money laundering and to report suspect transactions.

83. Most banks have to submit detailed and comprehensive prudential returns every half year, with a lag of up to five months. Certain information, for example on the main elements of the balance sheet, liquidity and foreign currency positions, must be submitted quarterly or monthly, and the Bank of Greece may demand the submission of additional data when it sees fit. Information contained in returns is used to construct a variety of indicators of bank profitability, solvency, risk taking, and liquidity. Comparability of data across banks has been enhanced by the establishment of uniform accounting practices in 1994. Individual banks’ liquidity and borrowing costs in the interbank market is monitored on a daily basis. In addition the Bank of Greece has automatic access to the reports of banks’ external auditors, including confidential reports.

84. The Bank of Greece is responsible for the prudential supervision of Greek commercial banks on a consolidated basis, but not of the nonbanking activities of other financial institutions such as insurance and leasing companies with which banks may be affiliated. The Supervision Department has approximately 130 staff, who mostly spend their careers at the Bank. Normally a supervisor is assigned a bank or pier group of banks to be monitored through both on- and off-site supervision.

85. Every bank is subject to a general on-site inspection at least once every three years, and in addition specific, ad hoc inspections are carried when data from returns or other sources indicates that a potential problem has arisen. The Bank of Greece emphasizes the review of management practices and internal controls in its on-site inspections, and also of the verification of how individual loans are classified. On this basis it makes recommendations to banks on how they can improve the soundness of their operations, including through the injection of new capital, and has on occasion levied fines or issued cease and desist orders where infractions have been found. According to most commercial banks, supervision by the Bank of Greece is both thorough and useful to them.

86. In 1995 a deposit guarantee scheme became operational, under which deposits are guaranteed up to Ecu 20,000 per depositor per institution. Greek banks are obliged to make contributions to a central fund as a proportion of their deposits along a sliding scale; initially the average rate of contribution was about 0.6 percent; supplementary contributions can be required if the need arises. The Bank of Greece does not have a published rule for deciding when and how to intervene in banks that display signs of illiquidity or deteriorating capital adequacy. It has twice during the last decade forced the closure of small insolvent institutions (one of which was subsequently rescued), in one case having to move rapidly to seize assets.

87. Banks are subject to a variety of taxes on lending, besides reserve requirements, which are high by European standards. Several specialized credit institutions receive favorable tax treatment, even though they are now engaged in a broader range of banking activities (see Chapter IV for more details). The tax treatment of provisions is also unusual by European standards (see SM/97/1, Supplement 3, 1/3/97, Tax Treatment of Loan Losses of Banks for a comparison), general provisions of up to 1 percent of loans without a government guarantee can be deducted from income, as can specific provisions on condition that the underlying problem loan is fully written off within three years, failing which back taxes and a penalty must be paid. Reportedly bankruptcy and other court cases can easily be delayed, and partial write-offs cannot be made, so banks are reluctant to make specific provisions. It might be added here that the bankruptcy law appears to be ambiguous on the treatment of netting of collateral used in repurchase and derivatives contracts.

B. Financial Market Structure

88. The Greek financial sector is dominated by commercial banks, and is in large part government controlled. Twenty one commercial banks are currently licensed by the Bank of Greece.61 Each major bank is at the center of a conglomerate encompassing also institutions offering insurance, factoring, leasing, and mutual fund services, and in some cases smaller banks.62 Besides the commercial banks, six specialized financial institutions exist to finance industrial development, provide residential mortgages, or collect retail deposits, although most now offer almost all the banking services available at commercial banks. Some of the specialized financial institutions, and certain state-controlled banks hold controlling or significant equity stakes in numerous nonfinancial companies. Twenty three foreign banks, mostly from the United States, the United Kingdom, the Netherlands, and France, have branches, and a number of other foreign banks have representative offices. Since 1990 several new private commercial banks (some with foreign equity participation), two branches of foreign banks, and four cooperative banks have been established. One bank, which was partially foreign-owned, has been closed and its depositors compensated. Three pairs of banks are known to be holding talks on partial or full merger, but progress is being delayed by contradictions between the various laws that seem to apply; a revised law clarifying the conditions for bank mergers and break-ups is being prepared.

89. Nine commercial banks, including the three largest by asset size, and all specialized financial institutions are partially or wholly government controlled. Government ownership can be direct or indirect, as when one bank is owned by a government-controlled bank or pension funds. Despite the presence of representatives of government and interest groups such as trade unions on the boards of these banks, most claim now to be fully commercially oriented. In recent years the government has (partially) privatized a few small banks. Individual banks belonging to one conglomerate reportedly operate autonomously, but in some cases managerial links between banks and affiliated nonbank financial institutions are reinforced by cross-ownership.

90. A summary of banks’ balance sheets and the structure of the banking system is provided in Table 8. Data is presented for 1990, when financial markets were still relatively tightly controlled, and for 1995, the latest period for which data is available and when the liberalization process was largely complete. The broadest aggregate covers Greek and foreign commercial banks and specialized credit institutions, except the industrial development banks, the Post Office Savings Bank, and several minor institutions, each of which is sui generis. Data on the three largest banks (which are all state controlled), the four largest private banks in 1995, and foreign banks are aggregated; the two smallest of the large private banks were not operational in 1990.

Table 8.

Greece: Elements of Bank Balance Sheets

article image
Source: Greek authorities, and Fund staff estimates.

Excluding the investment banks, the Post Office Savings Bank, and several small specialized institutions.

Excluding foreign banks.

91. Indications can be found of a decline in the relative importance of banks, and especially that of the traditional large banks, in the process of intermediation and credit allocation. While credits to the nonfinancial sector remain the largest item among banks’ assets, the share has been declining, as has been the ratio of such credits to nominal GDP, and there has been a corresponding increase in banks’ holdings of government securities. The growth in the money markets is reflected in the rapid rise in banks’ gross claims on and liabilities to financial institutions. While the accounts are not fully consolidated, it appears that the banking sector went from being a net borrower to being a net lender to other financial institutions.

92. The three largest, and state controlled, banking institutions (ranked by 1995 assets) retain just over half of all assets in the sector, and the largest five government-controlled institutions retain approximately two thirds, but relative to GDP their overall balance sheet size has remained stable, and their share of the banking sector has been declining.63 The decline in their share of credits to the nonfinancial sector and of deposits has been larger, although they continue to hold a disproportionate amount of deposits. These banks, together with the industrial development banks, hold almost all equity in nonfinancial enterprises. Market share has been gained by private Greek banks and foreign banks, which also both expanded their balance sheets relative to GDP.

C. Bank Balance Sheets

93. According to the Bank of Greece, all major banks meet the 8 percent minimum weighted average capital to asset ratio, and almost all capital is considered to fall into Tier I. Unweighted bank capitalization, shown in Table 8, increased by more than 1 percent of GDP between 1990 and 1995; the increase is largely attributable to the private Greek banks. The process is continuing and widening as banks try to reduce their funding costs, and finance the introduction of new technology and the restructuring of their operations. For example, one major government-controlled bank recently announced that it would pay no dividend for 1996 despite record profits in order to add to its general reserves. Both state-controlled and private banks have recently undertaken rights issues or are planing issues in the near future. Most banks have been able to increase the proportion of “free” capital, that is, capital minus fixed assets and equity participation, although for some banks (including some very large state-controlled banks) “free” capital remains very low or negative. A number of institutions are arranging to be assessed by international credit rating agencies in the hope of receiving better terms on borrowing and being able to raise more capital, possibly abroad, and it is generally recognized that increasing involvement in international capital markets will require greater transparency and conformity to international practice.

94. Bank capital has also been effectively augmented by the availability of government guarantees that were granted to support the financing of particular sectors by government-controlled institutions. The exercise of these guarantees contributed a significant part to the increase in bank holdings of government bonds since 1990, and the decline in the share of loans held by government-controlled banks. A stock of guaranteed loans remains outstanding, and every year some guarantees are called. One major specialized credit institution is undergoing recapitalization and restructuring, whereby it receives government bonds as past policy-loans are written off. This institution has also been cutting staff and is under the close supervision of the Bank of Greece.

95. Nonetheless, a significant proportion of some banks’ loan portfolios remains impaired or nonperforming. Recorded impaired loans of Greek and foreign banks, including all specialized financial institutions, increased from 4.3 percent of GDP (4.3 percent of all assets) in 1990 to 6.8 percent of GDP (6.1 percent of assets) in 1995. Impaired loans of Greek commercial banks increased over the period from 3.5 percent to 4.6 percent of GDP (from 4.6 percent to 5.7 percent of their assets). Some fraction of this increase may be explained by tighter accounting standards and a greater willingness on the part of banks to report problem loans, rather than by a deterioration in the financial position of borrowers or loan evaluation procedures by banks. Approximately two thirds of impaired loans are more than one year overdue and therefore nonaccruing. While a finer break-down is not available, most of these impaired loans are reported to be in the state-controlled institutions.

96. Despite the increase in impaired loans, provisioning of specific loan losses and other risks remains low by international standards, as can be seen in Table 8, and indeed it is striking that foreign banks in Greece, which are guided by their parent institutions, hold disproportionately high provisions. This tendency may in part be accounted for by the availability of government guarantees on some of a substantial stock of loans, and tax considerations (see above). Provisioning as a share of assets has even declined as a share of assets since the early 1990s, and as a result, by 1995 capital plus provisions for doubtful claims barely exceeded the book value of problem loans.

97. The relative stability of credit and deposit aggregates masks considerable change in individual financial markets. As can be seen from data contained in Table 8, lending in foreign currencies has become much more widespread in recent years. Foreign currency denominated lending other than to financial institutions rose from less than 4 percent of banks’ loan portfolio in 1990 to 15 percent in 1995, and amounted to more than 5 percent of GDP. Since then foreign currency lending has continued to grow rapidly, reaching one quarter of all loans outstanding in March 1997, and for some banks may represent the vast majority of new lending. Much of the foreign currency lending is reportedly denominated in Japanese yen. The development of this market has been driven on the demand side by the stability in the external value of the drachma, the premium on drachma interest rates, and certain tax provisions; on the supply side foreign currency lending has been facilitated by capital account liberalization, which was completed in May 1994, and the release of 30 percent of banks’ foreign currency deposits with the Bank of Greece in 1995. In addition direct borrowing abroad by enterprises has also increased, and most of the derivatives traded in Greece are foreign currency related. Thus, although most banks apparently do not take large foreign currency positions, the household and enterprise sectors may now be much more exposed than they were to exchange rate risk, especially when they do not themselves enjoy a source of export earnings.

98. Consumer lending is seen as a growth area, and many banks have tried to expand in this market. The dominant position of the housing banks in the provision of home mortgages has been weakened, but in real terms aggregate real estate loans have declined, and house prices are fairly stable.64 An array of new financial products, including derivatives, are now available either from banks in Greece and their affiliates in the local market or in the main European financial centers. The growth of nontraditional financial products in Greece has recently been dealt with in depth in Papaioannou, Michael, and E.K. Gatzonas, “Financial Innovation Involving the Greek Drachma,” IMF Working Paper WP/97/14, January 1997. Currently some banks are considering whether to introduce asset-backed securities, but regulatory clarification would be required.

D. Bank Operations and Profitability

99. The published profit and loss statement for different categories of banks are presented in Table 9, and Table 10 includes summary ratios on banks’ costs. Reported aggregate bank profitability has been comparatively high in Greece in recent years, although individual institutions, and especially one state owned specialized credit institution, suffered significant losses in the earlier part of the 1990s.65 Because nominal interest rates have been above those prevailing elsewhere in Europe, both interest revenue and interest expenditure have represented relatively large fractions of assets. The ratio of fee and other income to assets has historically been higher for Greek banks than for banks in most other European countries.

Table 9.

Greece: Bank Profit and Loss Statement

article image
Source: Greek authorities, and Fund staff estimates.

Excluding the investment banks and the Post Office Savings Bank.

Table 10.

Greece: Bank Branching and Staff Levels

article image
Source: Greek authorities, and Fund staff estimates.

Excluding the investment banks and the Post Office Savings Bank.

Deflated by the GDP deflator.

100. Generally the private commercial banks are proportionately more profitable than the government controlled institutions. They tend to have somewhat higher interest income relative to total assets, perhaps reflecting the smaller weight of government securities in their portfolio, and also higher earnings from fees and other income, which in turn is derived mostly from trading activities. The private Greek banks and foreign banks seem however to have relatively high costs of depreciation and other operating expenses, and to pay proportionately more in taxes.66 The relationship between profitability, cost and revenue structures, and ownership is considered in more detail in another background paper.

101. Bank branching and overall employment in the banking sector is somewhat lower than in several other European countries, though allowance should be made for the availability of retail banking services at post offices.67 During the 1990s banks in Greece, and especially private Greek banks, have continued to add staff and expand their branch networks, whereas banking sector employment has tended to contract in other industrialized countries. Such indicators of the input mix as the inflation-adjusted asset to staff or asset to branches ratios have remained stable, but real nonpersonnel operating costs have been rising, perhaps in part because of increased investment in computerization. The foreign banks have relatively high and rising assets and costs per branch and per employee, which may reflect their concentration on wholesale activities.

102. Several banks, including the major state-controlled banks and some of the specialized credit institutions, are attempting to redirect their business strategies and management practices. Most banks seem to intend to concentrate on the provision of financial services as part of conglomerates, and largely to dispose of their holdings in nonfinancial enterprises. Some disposals have already taken place, but the process is slow because banks wish to receive adequate compensation and the market for corporate control is not well developed.

103. Generally banks are aiming not so much to reduce staffing, which is difficult given labor market regulations, as to redeploy resources toward the provision of services to corporate clients and household depositors, and the marketing and distribution of financial products through their extensive branch networks.68 However, redeployment of staff may be hampered by skill shortages, which in some cases can be traced to the past history of state-controlled banks hiring without due regard for qualifications; managers at state-controlled banks are now required to have banking experience. In some instances banks’ organization and procedures are being reviewed with the help of outside consultants or major foreign financial institutions, although most banks in Greece regard their current internal control and audit procedures, including those applying to trading activity, as broadly satisfactory. A few banks are in the process of acquiring computer-based models of market risk.

104. Greater optimism about the future of the Greek banking system is reflected in the share prices of banks. As can be seen from Figure 14, since 1989 bank stocks have tended to out-perform others on the Athens Stock Exchange, where they constitute some of the largest and most actively traded issues. The trend has continued: in the first four months of 1997 the composite stock price index climbed 56 percent, but the index of banking stock prices rose 84 percent.

Figure 14.
Figure 14.

Greece: Banking System Share Prices

Citation: IMF Staff Country Reports 1997, 089; 10.5089/9781451816075.002.A001

Source: Bank of Greece, and staff estimates.

105. A striking change in the Greek financial sector as a whole over the past five years has been the growth in nonbank financial institutions, notably mutual funds, pension funds, and insurance companies. As can be seen from Table 11, the number of such institutions and the stock of assets they control has multiplied, and they now represent a major channel of intermediation. In particular the mutual funds, which are currently invested overwhelmingly in Greek government drachma-denominated securities, must be regarded as close substitutes for bank deposits; by the end of 1996 assets in mutual funds reached almost Dr 3.9 trillion, approximately one-fifth the stock of bank deposits or 15 percent of GDP. The growth in mutual funds, which are often subsidiaries of banks, was in part motivated by the advantage of avoiding reserve requirements on deposits and transforming interest income into capital gains, which are taxed at a lower rate.

Table 11.

Greece: Nonbank Financial Institutions

article image
Source: Greek authorities, and Fund staff estimates.

Data for 1996.

106. Awareness is increasing that Greece’s participation in the European Monetary Union could have considerable implications for banking operations and competition in the sector. Banks are concerned in the near future by the costs of transition, rather than by loss of income from reduced foreign exchange trading, which is estimated at only 2 to 4 percent of revenue on a gross basis. In the longer term there is some concern over the prospect of increased cross-border competition from banks in the main European financial centers. In most banks technical preparations are at a diagnostic stage, though progress has been made in developing a real time gross settlement system to be integrated into the European TARGET system.

VI. Indicators of Competition and Efficiency in the Banking System69

107. The banking sector is characterized by a relatively high level of concentration, especially in certain market segments.70 The largest banks and many smaller banks are state-controlled institutions that, at least until recently, were not fully commercially oriented either in their lending policies or in their cost management practices. The existing market structure and banks’ current portfolios are in large measure the inheritance of a long history of heavy regulation and political interference. Anecdotal evidence suggests that oligopolistic tendencies and inefficiencies were widespread, but that competition has become sharper in the past few years. It is widely expected that European economic integration and technological innovation will create market pressures on all banks to cut costs and further improve services to their customers, in part through mergers.

108. Given this evolution, it is of interest to investigate more thoroughly and quantatively the degree of competition and efficiency displayed by the Greek banking system, with a view to discerning likely trends for the future. To this end the relationship between bank profitability and market position will be examined. It is found that certain banks have achieved persistently higher levels of profitability than others, and in particular that midsized banks that are not state-controlled have been able to sustain much higher rates of return. Some insight into the causes of these differences is obtained through a decomposition of banks’ revenues and costs. This decomposition makes apparent the importance for revenues of bank specialization in different market segments such as intermediating domestic savings, wholesale banking, providing transaction services, and proprietary trading. Certain banks have successfully adopted a strategy of spending more on personnel and other operating items in order to generate high income from fees and trading activity. Tax rates differ significantly across banks, and state-controlled banks tend to earn relatively little fee and other income given their cost structure.

109. There have been numerous studies of banking sector competition and its cost and allocative efficiency in other countries.71 The multitude and variety of such studies in part reflects the difficulty of the undertaking. Banking is a multi-product service industry, where many inputs and outputs are difficult to measure or even define. By its very nature, banking involves asymmetric information, the intensive employment of human capital, and certain fixed costs, for example, in operating a branch network. It is not surprising to find different banks specializing in different segments of the banking market, and therefore having very different portfolios and cost structures, so it is not easy to identify “best practices.” For Greece any study is further complicated by the changing regulatory and accounting framework, and the relatively small number of banks, which limits the amount of data available. These caveats should be born in mind when assessing the results given below.

A. Rates of Return and Market Share

110. In a competitive industry one would expect profitability across firms to converge due to entry and exit and because firms can copy the most successful technology or pricing behavior. Persistent differences in profitability may be due to a number of factors: unusually profitable firms may have monopoly power, either protected by barriers to entry or due to a strategic advantage over their rivals, for example in the introduction of innovations. Less profitable firms may be operating with inferior technology and management, or they may have an objective other than profit maximization, such as the maximization of revenue or managerial perquisites, or the fulfillment of certain politically-dictated social obligations. The magnitude and persistence of differences in profitability across firms in an industry are thus indicators of the degree of competition, and the relationship between such differences and firm characteristics may suggest a causal explanation. Here the profitability of different Greek banks will be compared and related to their market shares and ownership structure.

111. In the case of banking, profitability is usually measured in terms of return on equity (ROE) or return on assets (ROA). Annual data were available for 1985 and 1990–95 on the after-tax profits, total assets, loans, deposits, and capital of 18 domestic Greek banks, including the principal commercial banks and the two housing banks.7273 The ROE series was constructed as the ratio of profits to capital, and the data were also used to construct several measures of market concentration, and in particular, the ratio of each bank’s assets to the total in each year (its asset share, ASTSH). One bank displayed a very high ROE in all years, and two other banks made exceptional losses in individual years; dummy variables were constructed to deal with these outliers. In addition, dummy variables were constructed to capture year-specific effects.

112. The ROE series was regressed on a constant (CONST), a function of ASTSH, and the dummy variables. After some experimentation it was found that the logarithm of ASTSH and the square of the logarithm provided a robust and flexible functional form. Only the year dummies for 1990 and 1991 proved significant; the other parameter estimates do not make a significant difference whether or not they are included. “Pooled” OLS estimates were generated for the whole sample together, and then a “fixed effects” model was estimated, where the data are stratified by bank.74 Summary results are presented in Table 12, and the actual and predicted values from specification (2) are shown in Figure 15.

Table 12.

Greece: Return on Equity and Asset Shares of Greek Banks

article image
Source: Bank of Greece, Commercial Bank of Greece, and staff estimates. Standard errors in parentheses. **: significant at 1 percent. *: significant at 5 percent. +: significant at 10 percent.
Figure 15.
Figure 15.

Greece: Banks’ ROE and asset shares

Citation: IMF Staff Country Reports 1997, 089; 10.5089/9781451816075.002.A001

Source: Bank of Greece and staff estimates.

113. The ROE of Greek banks displayed a highly significant, nonlinear relationship to asset shares, such that both relatively small and relatively large banks achieved a relatively low ROE. Given the existing market structure, and based on the “pooled” regressions, ROE reaches a peak of about 24 percent for banks with a little over 6 percent of total assets. When average differences between banks are accounted for with “fixed effects,” a qualitatively similar, albeit statistically less significant relationship between ROE and asset shares is obtained, with a maximum ROE obtained for a market share of about 12 percent. This result suggests that, while differences in profitability between banks are persistent (the fixed effects taken together are highly significant), banks that evolve to be moderate in size can raise their profitability.

114. Several explanations for the relationship between profitability and asset shares can be put forward. It could be that the smallest banks had not exhausted all economies of scale, and that large banks suffered diseconomies. Cost functions may have been such that the large and medium-sized banks pursued different price-setting strategies. The medium-sized banks may also have found it managerially easier to introduce new technologies and products. Since most of both the small and the large banks are state-controlled, they may not have been striving only to maximize profits, but were also obliged to fulfill certain policy objectives. Therefore they may have lent at more favorable terms or suffered unusually high levels of default. They may have also tolerated greater cost inefficiencies, either because they were under less market discipline or because political pressure hindered managerial action.

115. Very similar results were obtained when profitability was measured as the ratio of a bank’s profits to its assets, loans, or deposits, and when market share was measured on the basis of these other balance sheet items. The results proved robust to changes in the functional form of the specification, and exclusion of the dummy variables. Results for different sub-samples were also similar; there seems to have been little convergence in profitability rates over the period.

B. Revenue and Cost Components

116. Greek banks are heterogeneous in their portfolio structure, their sources of revenue, and their cost structures. This heterogeneity may be due in part to differences in efficiency, market power and technology, to the strategic decisions of bank management, or to the historical circumstances of individual banks. By relating revenue to expenses one may obtain indicators of relative efficiency and even of the direction in which banking business is evolving, although it will not normally be possible to disentangle demand, cost and efficiency factors.

117. To this end regressions were run with various revenue items as the dependent variable and cost components and other variables on the right-hand side. The data set, based on published annual accounts, covered ten Greek commercial banks and two housing banks in 1990,1993, and 1995.75 Revenue was broken down into interest income, net fee revenue, and other revenue, which consists mostly of trading income. Expenses includes interest (INTER), personnel (PERSL), other operating items including depreciation (OOPER), other expenses (OEXPS), taxes (TAXES), and provisions for bad loans (PROVS). All variables were normalized by converting them into percentages of the bank’s total assets. To allow for possible non-linearities both the level and the square of the explanatory variables were included, and some experimentation with the specification and other functional forms was undertaken. A dummy variable was constructed that takes the value unity for a bank that is state controlled and zero otherwise (STATE), and two other dummy variables were included to deal with outliers. Both “pooled” and “fixed effects” models were estimated. Results of the final specification are presented in Table 13. The results suggest the presence of a number of relationships.

  • Revenue, and in particular interest income increases close to one for one with interest expenses, but when comparing across banks in the “pooled” regressions interest expenses enter with a coefficient greater than one in the interest revenue equation and a negative coefficient in the fee income equation. Thus, high interest expenses are associated with banks that specialize in lending and receive relatively little fee income.

  • A one drachma increase in personnel costs, relative to assets, is associated with more than one drachma in extra revenue. In the “pooled” regressions, almost all of this extra revenue takes the form of fee and other income; perhaps certain banks are more specialized in providing services that generate fee income or in proprietary trading that generate other income, both of which activities require more and better qualified staff. In the “fixed effects” regressions, personnel costs are positively correlated with interest and other income; the statistical relationship may the result of a common (downward) trend in the three variables.

  • The results from the “pooled” regressions suggest that an increase in other operating expenses is associated with increased revenues, especially interest and fee income, but the relationship is non-linear, and the squared term enters with a negative coefficient. Thus, only up to a certain point do banks that spend more on other operating items such as branching and computerization tend to earn more on their lending activities and earn higher fees; the maximum is achieved when other operating expenses equal approximately 1.7 percent of assets. The curvature of the relationship is reversed in the “fixed effects” regressions, indicating that extra spending by individual banks is associated first with lower and then with constant or higher revenues, possibly due to economies of scale or demand factors.

  • Other expenditures are closely associated with other revenues, and again the relationship is nonlinear. Both across banks and, to a lesser extent, across time for individual banks one sees that only up to some maximum do higher expenses, due for example to revaluation losses, go together with higher income, including higher trading income. Possibly very high other expenses represent exceptional losses, correction of past hidden losses, or nonproductive outlays, none of which generate income.

  • Provisions for loan losses, which are mostly small, are not strongly related to interest or fee income. Provisioning is estimated to be negatively associated with other revenues and, more weakly, fee income, perhaps because of some omitted factor; possibly certain banks may be shifting their strategies from risky lending, which occasions provisioning, to the provision of services and proprietary trading.

  • As expected, tax payments are strongly related to revenue, especially interest revenue. The “pooled” regression coefficient is more than double the “fixed effects” regression coefficient, indicating that some banks are systematically more “tax efficient” than others.

  • The estimated coefficient on the dummy variable capturing state control is significantly negative. State-controlled banks seem to earn consistently lower fee and other revenue than one would expect based on their cost structure.

Table 13.

Greece: Income and Expenses of Greek Banks

article image
Source: BOG, and staff estimates.

Includes two dummy variables to account for outliers.

Includes one dummy variable to account for an outlier.

Includes one dummy variable to account for another outlier.

118. In summary, the regression results suggest that many of the differences between Greek banks are systematic, representing their various specializations and levels of efficiency. Higher costs, for instance for better qualified personnel, can be justified by higher revenues. However, since market niches may already be occupied and in some areas diseconomies of scale may become significant, one bank may not be able to replicate the strategy of another.

List of Recent Staff Studies

1996: SM/96/188, 7/19/96

Growth Performance: A Survey of the Literature

Unemployment: A Survey of the Issues

The Capital Inflows Episode: 1994-96

The Greek Experience with Exchange-Rate Based Stabilization

Electoral and Partisan Effects on Macroeconomic Policy

1995: SM/95/179, 7/25/95

An Index of Coincident Economic Indicators for Greece

Private Saving Behavior in Greece

Inflation, the Budget Deficit and Debt Dynamics

Competitiveness and External Performance

1994: SM/94/173, Supplement 1, 7/6/94

Greek Manufacturing 1980-92 in a European Perspective

Money Demand in Greece

Exchange and Trade Issues

Data Issues

1993: SM/93/112, 5/24/93

Social Security Reforms

Financial Liberalization

1991: SM/91/91, 5/14/91

Investment and Potential Growth

Domestic Structural Issues

International Issues

Inflation and the Private Savings Ratio: Theory and an Application to the Case of Greece

STATISTICAL APPENDIX

Table 1.

Greece: Aggregate Demand

(At constant prices of the previous year)

article image
Sources: Ministry of National Economy; and Fund staff calculations.
Table 2.

Greece: Aggregate Demand

article image
Source: Ministry of National Economy.
Table 3.

Greece: Private Sector Income Account 1/

(In billions of drachmas; at current prices; percentage changes in parentheses)

article image
Source: Ministry of National Economy.

Including public enterprises.

Table 4.

Greece: Saving-Investment Balance

(In billions of drachmas, at current prices; in percent of GDP in parentheses)

article image
Sources: Ministry of National Economy; and Fund staff calculations.

On a national accounts basis; government statistics refer to the general government.

Adjusted to include PIP transfers as current revenue.

Current account deficit.

Table 5.

Greece: Agricultural Production

(In thousands of tons)

article image
Sources: Ministry of National Economy; and National Statistical Service of Greece.

Oriental, burley, and Virginia varieties.

Table 6.

Greece: Manufacturing Production

(Percentage Changes)

article image
Sources: National Statistical Service of Greece, Monthly Statistical Bulletin; Ministry of National Economy; and IOBE.

Estimate by IOBE.

Table 7.

Greece: Price Developments

(Average percentage changes over preceding period, except as indicated)

article image
Source: Bank of Greece, Monthly Statistical Bulletin.

Weights are based on 1980 for the wholesale price index and 1988 for the consumer price index prior to 1995, and based on 1994 for data for 1995–96.

This category includes health and personal care and education and recreation, along with other goods and services.

Unit value of imports.

Table 8.

Greece: Implicit Price Deflators

(Percentage changes)

article image
Source: Ministry of National Economy.
Table 9.

Greece: Cost-Push Indicators of Inflation

(Percentage changes)

article image
Source: Ministry of National Economy.

Per unit of output.

Table 10.

Greece: Labor Force, Employment, and Unemployment

(In thousands, unless otherwise noted)

article image
Source: National Statistical Service of Greece.

Period average.

Based on the annual labor force survey by the National Statistical Service of Greece.

By the Labor Force Employment Organization (OAED).

14+ age group.

Table 11.

Greece: Employment in Selected Sectors

(In thousands)

article image
Sources: Ministry of National Economy; National Statistical Service of Greece; and Union of Banks.

Seventy thousand persons employed in the repair of vehicles and home appliances have been reclassified into the service sector.

Permanent and temporary employees of the central administration, local authorities, and other budgetary organizations.

Table 12.

Greece: Wages and Salaries in the Nonagricultural Sector

(Percentage changes over previous period)

article image
Sources: Bank of Greece; and National Statistical Service of Greece.

National accounts basis (ESA).

Bank of Greece estimates; differences in rates of change between wage bill and average earnings are due not only to changes in employment, but also to statistical discrepancies.

Gross remuneration (including overtime) in establishments with ten or more employees.

Preliminary estimates.

All sectors excluding the civil service, public enterprises, and banking.

Table 13.

Greece: Employment, Productivity, and Unit Labor Costs in Manufacturing

(Annual percentage changes)

article image
Sources: Bank of Greece; and National Statistical Service of Greece.

Preliminary estimates.

For wage earners.

Production per manhour.

Estimate.

Table 14.

Greece: Collective Labor Agreements, Compulsory Arbitration, and Impact of Labor Disputes

article image
Sources: Bank of Greece; and Ministry of Labor.

Starting in 1992, arbitration decisions are not issued by courts, but by the newly-established (under Law 1876/90) Organization for Mediation arid Arbitration.

Table 15.

Greece: Summary of Central Government Finances 1/

article image
Sources: Ministry of Finance; and Bank of Greece.

Data not directly comparable to those on a national accounts basis in Tables 22 and 23.

Special Account for Guarantees of Agricultural Products.

Table 16.

Greece: Ordinary Budget Revenue

(In billions of drachmas)

article image
Source: Ministry of Finance.
Table 17.

Greece: Ordinary Budget Revenue

(In percent of GDP)

article image
Source: Ministry of Finance.
Table 18.

Greece: Ordinary Budget Expenditures

article image
Source: Ministry of Finance.

Does not include capitalized and accrued interest.

Bank of Greece data.

Table 19.

Greece: Investment Budget Expenditure by Sector

(In billions of drachmas)

article image
Source: Ministry of Finance.

Does not include Dr 48.5 billion paid to the Greek Telecommunication Organization against loan from the European Investment Bank, and Dr 19 billion for increase of Olympic Airways share capital in 1995.

Table 20.

Greece: Budget Transfers from and to the European Union

(In billions of drachmas)

article image
Source: Ministry of Finance.
Table 21.

Greece: Central Government Expenditure, Functional Classification

(Accrual basis)

article image
Source: Ministry of Finance.

Including military debt service.

Including capitalized interest.

Table 22.

Greece: Summary of General Government Finances 1/

(In billions of drachmas)

article image
Source: Ministry of National Economy.

Data on a national accounts basis; central government accounts not directly comparable to those compiled by the Ministry of Finance.

Table 23.

Greece: Summary of General Government Finances 1/

(In percent of GDP)

article image
Source: Ministry of National Economy.

Data on a national accounts basis; central government accounts not directly comparable to those compiled by the Ministry of Finance.

Table 24.

Greece: Public Entities Balance 1/

(In billions of drachmas)

article image
Source: Ministry of Finance.

Covers seven major public entities.

Excluding amortization payments.

Table 25.

Greece: Public Enterprise Balance 1/

(In billions of drachmas)

article image
Source: Ministry of National Economy.

Covers 46 major public enterprises.

Breakdown into components are estimates.

Surplus (+) or deficit (−).

Excluding amortization payments.

Table 26.

Greece: Operating Balance of Selected Public Enterprises

(In billions of drachmas)

article image
Sources: Ministry of Finance and Ministry of National Economy.

Including Thermic Buses Corporation (since June 1994), Athens Piraeus Trolley Buses, and Athens Piraeus Electric Railways.

Thirty-one additional public enterprises.

Table 27.

Greece: Financing of the PSBR

(In billions of drachmas)

article image
Source: Bank of Greece.

Including treasury bills and bonds held by the Bank of Greece, as well as changes in the balance of the petroleum account through 1992.

Table 28.

Greece: Gross General Government Debt

(In billions of drachmas; end of period)

article image
Sources: Ministry of Finance; and Bank of Greece.

Replaced by long-term bonds at end-1993.

Bonds issued in 1993 to cover valuation differences.

Table 29.

Greece: Monetary Program and Outturn 1/

(End of period)

article image
Source: Bank of Greece.

The definition of net domestic credit and credit to the public sector in the monetary program is different from that in the monetary survey, it includes borrowing by the public sector directly from abroad, as well as capitalized interest. Also, for all credit aggregates the data do not reflect the exchange of government-guaranteed credit for government bonds.

Percentage changes in credit to the public sector and net domestic credit are calculated as the flows during the year excluding valuation adjustments over the stock of debt outstanding at the end of the previous year.

M4 was revised to include secondary market transactions from 1995 on.

NDC to the public sector in 1995 is affected by the inclusion of secondary-market sales of government paper from bank portfolios to the nonbank public.

Table 30.

Greece: Changes in Minimum Reserve Requirements and Investment Ratios for Commercial Banks

(In percent of deposits)

article image
Source: Bank of Greece.

Half of the required reserves are unrenumerated.

Applicable on drachma and foreign exchange deposits, surrendered to the Bank of Greece against drachmas.

Of which, as of June 1988 3 1/2 percentage points, and as of September 1991 4 1/2 percentage points, are non-interest-bearing.

The base was also broadened to include deposits of and bank borrowing from nonresidents as well as certain new categories of liabilities resulting from the creation of new financial products with maturities up to two years. For these, the reserve requirement was 3 percent in July, 7 percent in August, and 11 percent in September 1995.

The base was also broadened to include foreign exchange borrowing with maturities exceeding two years.

Share of the increase in deposits that must be invested in treasury bills.

In May 1993 this requirement was abolished.

Applicable only to drachma deposits.

As of November 1992, this requirement applied only to changes of deposits.

In July 1993 this requirement was abolished.

Table 31.

Greece: Monetary Survey 1/

(In billions of drachmas; end of period)

article image
Source: Bank of Greece.

Revised data not comparable to previous years, due to a change in the reporting system. Data reflect the exchange of government-guaranteed credit for government bonds. Also, net credit to the central government in 1991-95 includes capitalized interest on government bonds held by commercial banks.

Includes securities and loans in foreign currency.

Excluding long-term loans in foreign currency by the Bank of Greece.

Net domestic credit to the central government now includes Bank of Greece foreign exchange differences.

The monetary aggregates are defined as follows: narrow money (M1) is currency plus private sight deposits (excluding blocked deposits); broad money (M3) is Ml plus time and savings deposits, bank bonds and repurchase agreements; total drachma financial assets (M4) is M3 plus private sector holdings of T-bills and government bonds of maturity up to one year.

Table 32.

Greece: Growth of Money and Credit Aggregates 1/

(In percent per annum; end of period)

article image
Sources: Bank of Greece; and Fund staff calculations.

Figures include capitalized interest on government bonds held by commercial banks. Data also reflect the exchange of government-guaranteed credit for government bonds.

Excluding long-term loans to government in foreign currency by the Bank of Greece.

Including securities and loans in foreign currency.

Table 33.

Greece: Issues of Securities Through the Athens Stock Exchange

(In millions of drachmas)

article image
Source: Bank of Greece; and Athens Stock Exchange.
Table 34.

Greece: Distribution of Bank Credit to the Private Sector 1/

(End of period)

article image
Source: Bank of Greece.

Without taking into account the reduction in outstanding bank credit caused by the conversion of loans guaranteed by the government into government bonds. These conversions were: 1991 Dr 54.3 billion; 1992 Dr 185.0 billion; 1993 Dr 492.1 billion; and 1994 Dr 31.9 billion.

Table 35.

Greece: Short-term Interest Rates

(In percent)

article image
Source: Bank of Greece.
Table 36.

Greece: Official Interest Rates

(In percent)

article image
Source: Bank of Greece.

According to the size of the overdraft.

In addition, a penalty surcharge of 0.4 percent per day was imposed on bank overdrafts.

In addition, a penalty surcharge of 0.3 percent per day was imposed on bank overdrafts.

In addition, a penalty surcharge of 0.1 percent per day was imposed on bank overdrafts.

Table 37.

Greece: Bank Interest Rates

(End of period; in percent per annum)

article image
Source: Bank of Greece.

The Lombard facility was introduced on June 16, 1993.

Table 38.

Greece: Interest Rates on Government Paper

(End of period, in percent per annum)

article image
Sources: Bank of Greece; and IMF, International Financial Statistics.
Table 39.

Greece: Foreign Exchange Government Bond Yields

(In percent)

article image
Source: Bank of Greece.
Table 40.

Greece: Exchange Rates

(Percentage changes) 1/

article image
Sources: Bank of Greece; IMF, International Financial Statistics; and Fund staff catenations.

Foreign currency per drachma; a negative sign denotes a depreciation.

January-April against same period of previous year.

April 30, 1997 against December 31, 1996.

Based on OECD data. Unit Labor Costs in the business sector relative to a group of 18 industrial countries weighted by their share in Greece’s trade in 1992. The countries include all the EU countries, excluding all the EU countries, excluding Luxembourg, the United States, Japan, and die EFTA counties excluding Ireland.

Non-oil trade weighted vis-a-vis 15 competitor countries (1981-84 weights).

January-May against same period of previous year.

April 30, 1997.

January-April, 1997.

Table 41.

Greece: Official Reserves

(In millions of U.S. dollars; end of period)

article image
Sources: IMF, International Financial Statistics; and Bank of Greece.
Table 42.

Greece: Balance of Payments

(In millions of U.S. dollars; on a settlement basis)

article image
Sources: Bank of Greece, Monthly Statistical Bulletin; data provided by the authorities; and IFS.

Private and official capital, excluding errors and omissions.

Includes direct investment and enterprise borrowing abroad.

Includes official suppliers’ credits.

Borrowing by the Hellenic Industrial Development Bank, the Agricultural Bank of Greece, and the National Mortgage Bank of Greece.

Table 43.

Greece: External Services and Transfers

(In millions of U.S. dollars)

article image
Source: Bank of Greece.

Receipts reflect net EU transfers.

Table 44.

Greece: External Current Account Deficit and Financing

(In percent of GDP, settlement basis)

article image
Sources: Bank of Greece; and IMF, International Financial Statistics.

Including residual items.

Table 45.

Greece: Current Account of the Balance of Payments 1/

(In millions of U.S. dollars)

article image
Source: Ministry of National Economy.

National accounts presentation. Converted into U.S. dollars using the annual average exchange rate.

Figures for 1996 are estimates based on January to October customs data.

Excludes official EU transfers to the public investment program.

PIP: Public Investment Program.

Table 46.

Greece: Selected Indicators for Trading Partners 1/

(Annual changes, in percent)

article image
Source: IMF, International Financial Statistics.

Except for non-fuel commodity prices (see footnote 7 below), these composites are averages of percentage changes of data for each trading partner (as specified in footnotes 3 and 4 below) weighted by their share in exports or imports, as appropriate, of Greece.

Weights are proportional to 1992 exports of Greece to partner countries as specified in footnotes 3 and 4 below.

Based on data for partner countries that together account for at least 95 percent of exports or imports, as appropriate, of Greece.

Based on data for industrial partner countries only.

Weights are proportional to 1992 imports of Greece from partner countries as specified in footnotes 3 and 4 above.

That is, weighted averages of percentage changes in indices expressed in national currencies of industrial partner countries.

Based on averages of world market prices for component non-fuel commodities weighted by the 1979-81 composition of commodity trade (exports and imports) of Greece.

Table 47.

Greece: Capital Account

(In millions of U.S. dollars)

article image
Source: Bank of Greece.

Includes some debt-creating capital flows in the form of enterprise borrowing abroad.

Borrowing by the Helenic Industrial Development Bank, the Agricultural Bank of Greece, and the National Bank of Greece.

Medium- and long-term only.

Table 48.

Greece: General Government External Debt 1/

(In millions of U.S. dollars; end of period)

article image
Source: Bank of Greece; Ministry of Finance.

Including external borrowing by the Bank of Greece on behalf of the Central government prior to 1994.

Bonds with maturities of one year or less and treasury bills issued in the domestic market held by residents.

Excluding military debt.

Table 49.

Greece: External Debt Service 1/

(In millions of U.S. dollars)

article image
Source: Bank of Greece.

Excludes private nonguaranteed amortization after 1994.

Medium- and long-term only. Includes both interest and amortization payments.

Debt service (total: A + B + C) in percent of current account receipts.

References

Chapter II

  • Alesina, Alberto, and Roberto Perotti, 1994, “The Welfare State and Competitiveness,” NBER Working Paper No. 4810.

  • Alesina, Alberto, and Roberto Perotti, , 1995, “Taxation and Redistribution in an Open Economy,” European Economic Review, Vol. 39 (May).

    • Search Google Scholar
    • Export Citation

  • Alesina, Alberto, and Roberto Perotti, , 1995b, “Fiscal Expansions and Adjustments in OECD Countries,” Economic Policy No. 21 (October).

    • Search Google Scholar
    • Export Citation

  • Alesina, Alberto, and Roberto Perotti, , 1996, “Fiscal Adjustments in OECD Countries: Composition and Macroeconomic Effects,” IMF Working Paper 96/70, July (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation

  • Alesina, Alberto, and Guido Tabellini, 1990, “A Positive Theory of Fiscal Deficits and Government Debt,” Review of Economic Studies, Vol. 57 (July).

    • Search Google Scholar
    • Export Citation

  • Christodoulakis, Nicos M., 1994, “Fiscal Developments in Greece, 1980–93: A Critical Review,” European Economy No. 3., pp.99–134.

    • Search Google Scholar
    • Export Citation

  • Commission of the European Communities, 1992, “Greece,” Country Studies No. 9, Directorate-General for Economic and Financial Affairs, July (Brussels).

    • Search Google Scholar
    • Export Citation

  • Courakis, Anthony S., Fatima Moura-Roque, and George Tridimas, 1990, “Public Expenditure Growth in Greece and Portugal: Wagner’s Law and Beyond,” Institute of Economics and Statistics Applied Economics Discussion Paper Series, No. 90 (Oxford).

    • Search Google Scholar
    • Export Citation

  • Curtis, Glenn E., ed., 1995, Greece: A Country Study, Federal Research Division, Library of Congress (Washington, 4th ed.).

  • Demekas, Dimitri G., and Zenon G. Kontolemis, 1996, “Unemployment in Greece: A Survey of the Issues,” IMF Working Paper 96/91, August (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation

  • Diamond, Jack, and Alan A. Tait, 1988, “The Growth of Government Expenditure: A Review of Quantitative Analysis,” IMF Working Paper 88/17, February (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation

  • Dowrick, Steve, and John Quiggin, 1997, “True Measures of GDP and Convergence,” American Economic Review (March).

  • Easterly, William, and Sergio Rebelo, 1993, “Fiscal Policy and Economic Growth: An Empirical Investigation,” Journal of Monetary Economics, Vol. 32.

    • Search Google Scholar
    • Export Citation

  • Eurostat, 1993, Old Age Replacement Ratios (Brussels).

  • GFS Yearbook 1996 (Washington: International Monetary Fund).

  • Giavazzi, F., and M. Pagano, 1990, “Can Severe Fiscal Adjustments be Expansionary?,” NBER Macroeconomics Annual.

  • Happe, Nancy, and John Wakeman-Linn, 1994, “Military Expenditures and Arms Trade: Alternative Data Sources,” IMF Working Paper 94/69, June (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation

  • Hewitt, Daniel, 1991, “Military Expenditures—International Comparison of Trends,” IMF Working Paper 91/54, May (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation

  • Hewitt, Daniel, 1993, “Military Expenditures 1972–90: The Reasons Behind the Post-1985 Fall in Military Spending,” IMF Working Paper 93/18, March (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation

  • Hondroyiannis, George, and Evengelia Papapetrou, 1995, “An Examination of Wagner’s Law for Greece: A Cointegration Analysis,” Public Finance, Vol. 50, No. 1.

    • Search Google Scholar
    • Export Citation

  • IBRD, 1997, World Population Projections Database.

  • International Monetary Fund, 1995, Unproductive Public Expenditures: A Pragmatic Approach to Policy Analysis, IMF Pamphlet Series, No. 48 (Washington).

    • Search Google Scholar
    • Export Citation

  • Jouganatos, George A., 1992, The Development of the Greek Economy, 1950–1991: An Historical, Empirical, and Econometric Analysis, Greenwood Press (Westport, Connecticut).

    • Search Google Scholar
    • Export Citation

  • King, M. A., and D. Fullerton, eds., 1984, The taxation of Income from Capital: A Comparative Study of the United States, the United Kingdom, Sweden, and West Germany, NBER (Chicago: University of Chicago Press).

    • Search Google Scholar
    • Export Citation

  • Kostakis, Anastasia, 1990, “Vocational and Academic Secondary Education in Greece: Public and Private Costs Compared,” Economics of Education Review, Vol. 9, No. 4.

    • Search Google Scholar
    • Export Citation

  • McDermott, John C., and Robert F. Wescott, 1996, “An Empirical Analysis of Fiscal Adjustments,” IMF Working Paper 96/59, June (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation

  • Mendoza, Enrique G., Assaf Razjn, and Linda L. Tesar, 1993, “A Comparative Analysis of the Structure of the Tax System in Industrial Countries,” IMF Working Paper 93/14, February (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation

  • Mendoza, Enrique G., Gian Maria Milesi-Ferretti, and Patrick Asea, 1995, “Do Taxes Matter for Long-Run Growth?: Harberger’s Superneutrality Conjecture,” IMF Working Paper 95/79, August (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation

  • OECD, 1996a, Revenue Statistics 1996 (Paris).

  • OECD, 1996b, Educational Policy Review: Greece (Restricted), March (Paris).

  • OECD, 1996c, Special Sessions on Consumption Taxes: Consumption Tax Trends (Restricted), September (Paris).

  • OECD, 1996d, National Accounts: Detailed Tables, Volume II (Paris).

  • OECD, 1996e, OECD Economic Surveys (Paris).

  • OECD, 1996f, Social Expenditure Statistics of OECD Member Countries: Provisional Version (Paris).

  • OECD, 1997a, Analytical Database.

  • OECD, 1997b, OECD Economic Surveys: Greece (Paris).

  • OECD, 1997c, OECD Economic Outlook, June (Paris).

  • Patrinos, Harry Anthony, 1992, “Higher Education Finance and Economic Inequality in Greece,” Comparative Education Review, Vol. 36, No. 3 (August).

    • Search Google Scholar
    • Export Citation

  • Tanzi, Vito, 1993, “Fiscal Issues in Adjustment Programs,” in J. de Melo and R. Fani, eds., Fiscal Issues in Adjustment in Developing Countries, MacMillian (London).

    • Search Google Scholar
    • Export Citation

  • Zee, Howell H., 1996a, “Taxation and Unemployment,” IMF Working Paper 96/45, May (Washington: International Monetary Fund).

  • Zee, Howell H., , 1996b, “Empirics of Cross-Country Tax Revenue Comparisons,” World Development, Vol. 24 (October).

Chapter III

  • Baxter, M., 1994, “Real Exchange Rates and Real Interest Differentials: Have We Missed the Business-Cycle Relationship?,” Journal of Monetary Economics, 33, pp. 5–37.

    • Search Google Scholar
    • Export Citation

  • Canonero, G., 1995, “Spain: Unemployment, Debt Management, and Interest Rate Differentials,” IMF Working Paper 95/107, November (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation

  • Caramazza, F., 1993, “French-German Interest Rate Differentials and Time-Varying Realignment Risk,” IMF Working Paper 93/1, January (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation

  • “Caught in the Debt Trap,” The Economist, April 1, 1995, pp. 59–60.

  • Christofides, C., 1996, “Financial Links Between Portugal and Spain: Evidence from Short-Term Interest Rates,” Portugal—Selected Issues and Statistical Appendix, Chapter III, IMF Staff Country Report No. 96/129, December 1996 (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation

  • Coorey, S., 1991, “The Determinants of U.S. Real Interest Rates in the Long Run,” IMF Working Paper 91/118, December (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation

  • Evans, L., S. Keep, and J. Okunev, 1994, “Modeling Real Interest Rates,” Journal of Banking and Finance, 18, pp. 153–165.

  • Favero C., F. Giavazzi, and L. Spaventa, 1996, “High Yields: The Spread on German Interest Rates,” NBER Working Paper, No. 5408.

  • Ford, R., and D. Laxton, 1995, “World Public Debt and Real Interest Rates,” IMF Working Paper 95/30, March (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation

  • Garcia-Herero, A., and J. Thornton, 1996, “Additional Evidence on EMS Interest Rate Linkages,” IMF Working Paper 96/115, October (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation

  • Halikias, I., 1994, “Testing the Credibility of Belgium’s Exchange Rate Policy,” IMF Staff Papers, 41,2:350-366.

  • Haque, N., M. Kumar, N. Mark, and D. Mathieson, 1996, “The Economic Content of Indicators of Developing Country Creditworthiness,” IMF Working Paper 96/9, February (Washington, International Monetary Fund).

    • Search Google Scholar
    • Export Citation

  • Isard, P., 1991, “Uncovered Interest Parity,” IMF Working Paper 91/51, May (Washington: International Monetary Fund).

  • Karfakis, C., and D. Moschos, 1989, “Testing for Long Run Purchasing Parity: A Time Series Analysis for the Greek Drachma,” Economics Letters, 30:245-248.

    • Search Google Scholar
    • Export Citation

  • Svensson, L., 1991, “The Simplest Test of Target Zone Credibility,” IMF Staff Papers, 38,3:655-665.

  • Taylor, M., 1991, “Intervention, Interest Rates, and Charts: Three Essays in International Finance,” IMF Working Paper 91/106, November (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation

  • Werner, M., 1996, “Mexico’s Currency Risk Premia in 1992-94: A Closer Look at the Interest Rate Differentials,” IMF Working Paper 96/41, April (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation

Chapters V and VI

  • International Monetary Fund, 1997, “Tax Treatment of Loan Losses of Banks,” SM/97/1 Supplement 3, January 3, 1997.

  • International Monetary Fund, 1996, “France–Selected Issues,” SM/96/249 Supplement 1, October 7, 1996.

  • Dziobek, Claudia, 1966, “Regulatory and Tax Treatment of Loan Loss Provisions,” IMF Paper on Policy Analysis and Assessment, 96/6, January (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation

  • Lewis, Alfred, and Gioia Pescetto, 1996, EU and US Banking in the 1990s, Academic Press, London.

  • OECD, Bank Profitability, Paris, 1996.

  • Papaioannou, Michael, and E.K. Gatzonas, 1997, “Financial Innovation Involving the Greek Drachma,” IMF Working Paper 97/14, January 1997 (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation

  • Journal of Banking and Finance, Vol. 17, Nos. 2-3, 1993.

1

Prepared by Mark S. Lutz.

2

Jouganatos (1992) provides a comprehensive description of the Greek economy in the postwar period. See also Commission of the European Communities (1992).

4

For example, the extensive systems of social transfers typical in Western European economies have had a strong influence on Greece’s expenditure patterns in recent decades.

5

See Diamond and Tait (1986). See also Courakis, Moura-Roque, and Tridimas (1990), and Hondroyiannis and Papapetrou (1995) for applications to Greece. Easterly and Rebelo (1993) and Zee (1996b) analyze tax revenue structures in economies at differing income levels.

6

Per-capita incomes are measured in 1995 U.S. dollars using purchasing power parity (PPP) exchange rates (from the WEO database), because nominal exchange rates may differ significantly from their PPP counterparts. However, PPP data compiled using constant price indices are subject to substitution bias, which may underestimate the dispersion of income levels when a relatively rich country’s prices are used to value GDP. See Dowrick and Quiggin (1997).

7

The EU average 1995 PPP per-capita income (excluding Greece and weighted by population) was US$15,781, 75 percent larger than Greece’s per-capita income of US$8,913, while the OECD average income was US$18,664, more than twice that of Greece.

8

Zee (1996a) finds that based on unweighted averages of OECD and of 56 developing countries, average revenues from payroll taxes during the 1975–89 period amounted to about 25 percent of total tax revenue for the former and about 7 percent of the latter. The ratios in Table 2 are 34 percent in Greece and in the EU, 30 percent in the OECD, and 23½ percent in the group of similar-income countries. The latter is higher than Zee’s number in part because it includes the Czech Republic, Hungary, Poland, and Slovenia, all of which have high ratios, in excess of 40 percent of total current revenues in Hungary and Slovenia, as legacies of their previous economic systems.

9

In the regressions, the United States was excluded and dummies were used for the four European transition economies.

10

See, for example, King and Fullerton (1984).

11

The tax rates are calculated as follows. The effective average tax rate on consumption is the ratio of general taxes on consumption (e.g., VAT) and excises, divided by private and public (net of the wage bill) consumption less consumption taxes. Thus, the consumption tax rate is taxes/consumption, with the latter at its pre-tax level. The effective average labor income tax rate begins with a calculation of households’ average tax rate on total income (which is necessary as detailed data for household tax revenues from labor and capital income are not separately available). This rate is defined as direct personal taxes divided by total household income (including wages and salaries, property and entrepreneurial income and the operating surplus of private unincorporated enterprises—hereafter, household operating surplus). The average effective tax on labor is calculated by multiplying this rate with wages, and adding social insurance contributions (by both employer and employee) and payroll taxes, with the sum divided by wages plus employer-paid social insurance contributions. Finally, the average effective tax on capital income equals the sum of (i) the sum of household’s operating surplus and household’s property and entrepreneurial income, multiplied by the average household income tax rate, (ii) direct business taxes, (iii) recurrent taxes on immovable property and, (iv) taxes on financial and capital transactions, all divided by the total operating surplus of the economy. Data are not available on the household operating surplus in Austria, Germany, and Switzerland. It is therefore assumed that the ratio of the household operating surplus to the sum of wages and the household operating surplus is the same as the average of the ratios for Denmark, Finland, the Netherlands, Norway, Sweden, and the United Kingdom. In Greece, data for operating surpluses are not sectored by households, government and incorporated businesses. Therefore the average ratio of households’ operating surplus to wages for Italy, Malta, and Spain is used. It is not thought that these assumptions impart any serious bias into the level of estimated tax rates, and certainly should not influence their trends.

12

Mendoza, Milesi-Ferretti, and Asea (1995) extend the sample to 18 OECD economies and obtain similar results.

13

For example, the simple (unweighted) average consumption tax rate among EU economies was 20.7 percent in 1994, with a standard deviation of 5.4 percentage points. Consumption tax rates in Denmark, Finland, and Norway were higher than Greece. Therefore, while Greece’s tax rate is among the highest, it is (barely) within one standard deviation of the simple average. It is more than one standard deviation above the simple average rate of 12.0 among economies with similar incomes (whose standard deviation was 6.9 percentage points). The simple average consumption tax rate for the entire sample of countries was 15.7 percent, with a standard deviation of 8.3 percentage points. The smaller average and larger standard deviation, compared to the EU aggregate, reflects the inclusion of a number of countries (Australia, Japan, Malta, Mauritius, Switzerland, and the United States) in which the consumption tax rate was less than 10 percent.

14

According to OECD Revenue Statistics 1996, during 1990–94 general consumption tax revenues (mainly VAT in the case of Greece, all EU and most other OECD members) were equivalent to an average 9.8 percent of GDP and accounted for 24.4 percent of total revenues. Comparable figures for the EU was 7.5 percent of GDP and 17.9 percent of total tax revenues, and 6.5 percent of GDP and 17.2 percent of total tax revenues among all OECD economies.

15

Another reason for the higher VAT yields is that VAT is usually levied on the duty-paid value of excise products, and Greece has a greater reliance on excises. See OECD (1996c).

16

Taxes on general consumption have increased during 1980–94 by 5.5 percentage points of GDP in Greece (from its 1980 level of 3.9 percent of GDP), compared to a 1.5 percentage point rise in the EU (from a 1980 base of 6.1 percent of GDP), and compared to a 1.7 percentage point rise in the OECD (from a 1980 base of 5 percentage points of GDP). See OECD (1996c), which also provides a comprehensive discussion of the structure of excises in its member economies. Thus, while part of the increase was in compensation for a lower initial tax share, this has been more than offset, especially if compared to other economies with comparable per-capita incomes.

17

It should be noted, however, that the variation of effective average tax rates on labor income in countries with incomes similar to Greece’s is fairly high. The simple average rate is 22.8 percent with a standard deviation of 14.2. Therefore, the effective labor tax rate, at 35.9 percent, is just within one standard deviation.

18

Eurostat data indicate that 86 percent of Greeks employed in the private sector (excluding agriculture) work in small- and medium-sized enterprises (SMEs)—the highest share in Europe, compared to an average of 66 percent for the EU. Moreover, most of these employees work in businesses with 1–9 employees. Greece is estimated to have more than one million SMEs, the sixth highest in the EU, despite that small size of its economy relative to most other EU partners.

19

Zee (1996a) reports that the combined employee and employer social security contribution rate was 43.3 percent, compared to an unweighted OECD average of 30.2 percent, whose standard deviation is calculated at 15.9 percentage points. Clearly, this places Greece at the upper end of the spectrum.

20

Although farmers were always formally subject to income taxation, relevant tax code provisions had not been applied. Commission of the European Communities (1992, p. 19, footnote 27) report that the contribution of tax revenues from agriculture amounted to around 0.3 percent of total tax revenues in 1990, despite the fact that agriculture accounted for about 25 percent of total employment.

21

In 1995, this tax raised an amount equivalent to 8.9 percent of total direct tax revenues and 0.6 percent of GDP.

22

Alesina and Perotti (1995) examine the effects of increasing labor taxes on unit labor costs and external competitiveness, finding that in a sample of 10 European countries an increase by 1 percent in the share of labor taxation in GDP leads to an increase in relative unit labor costs by 1–1.7 percent.

23

The unemployment rate remained stable at about 7½ percent throughout the 1980s, peaking at 8.2 percent in 1983–84, and falling to 7 percent in 1990. Subsequently, however, it has risen steadily, and was 10.4 percent in 1996.

24

Demekas and Kontolemis (1996) examine the reasons behind the increase in unemployment. Although they did not consider explicitly the role of labor taxes, they argue that the rigidity of real wages (which may reflect attempts by households to shift to business higher personal income taxes, and the latter’s attempts to shift to the former higher social security contributions), and, more importantly in their findings, the expansion of public sector employment and the demonstration effects of increasing public/private sector wage differentials were responsible for higher unemployment.

25

The simple average of effective average capital tax rates in the EU was 33.9 percent, with a standard deviation of 13.6 percentage points, while the OECD average was 35.7 percent with a standard deviation of 13.1 percentage points.

26

In 1994, the national accounts methodology was revised in line with EU standards, resulting in a 21 percent increase in measured output in 1988, the benchmark year. However, detailed data, including government consumption expenditures on a functional classification, are not yet available under the new methodology. As a result, data available under the old methodology through 1991, as contained in OECD (1996d), are used. Therefore, conclusions concerning government consumption components should be treated with caution. However, a cursory comparison of data for the same years under the two systems indicate that the figures for public consumption were the least revised among major expenditure components—in fact, they were reduced by 14 percent in the revised data for the benchmark year. A working assumption is that the relative size of the component shares are not affected by the shift in methodology. Therefore, one could simply multiply them by the ratio of the share of government consumption to GDP in the new NIA accounts, namely 14.8 percent, to their share under the old methodology (20.4 percent)

27

Based on the share in Table 4, and using the working assumption that the relative composition of final consumption expenditures is not changed under the new national accounts methodology. This total is about 50 percent higher than the industrial country average.

29

If, as a working assumption, one simply reduced the share of GDP for education by the ratio of consumption expenditures under the new and old methodologies (i.e., by 14.8/20.4), the revised share for education would be 2.3 percent of GDP. The relatively low education share is also apparent in the lower half of Table 4.

30

An additional unintended result of the lack of university slots is the significant number of students that study abroad. Patrinos (1992) reports that in 1990 there were over 28,000 Greek students abroad, equivalent to 15 percent of domestic enrollments, while the OECD (1996b) notes that this rose to 29,000 students in 1994.

31

Some care must be taken in interpreting these data, as they do not include mandatory social expenditures by employers or other private social expenditures, which may be significant in some countries (e.g., for health care in the United States). In addition, the coverage of expenditures at lower tiers of government in some countries may be less than complete. The taxation of social expenditures also varies widely across countries—as a result, two countries with identical expenditure levels may have much different net transfers. Finally, at present there are no comprehensive information on how governments pursue social policy objectives through the tax system.

32

According to the IBRD’s World Population Projections database, over 1990–95 the elderly (i.e., those aged 65 years and above) on average comprised 14¾ percent of the population, and its old-age dependency ratio (the number of elderly divided by those aged 15–64) was on average 22 percent. This compares to an OECD elderly ratio of 13¼ percent and old-age dependency ratio of 19¾ percent. Thus, while the population is somewhat more elderly than the OECD average, it is less than one standard deviation above the averages. Pension expenditure GDP shares were regressed on elderly and old-age dependency ratios. Again, while expenditure ratios were higher than the regressions’ predicted values, they were not significantly higher.

34

Eligibility requirements for new disability pensions were tightened in 1990–91, with a subsequent decline in the share of new disability pensioners, but the overall share of disability pensions in GDP has yet to fall noticeably.

35

See Curtis (1995). Data in the latest OECD Economic Outlook (1997c) also indicate a sharp increase since 1980 in private health care expenditures in Greece as a share of total expenditures.

36

While the Greek old-age dependency ratio of 22.9 percent in 1995 exceeds the OECD average of 20.2 percent by less than a standard deviation of 4.8 percentage points, by 2005 the Greek ratio is anticipated to rise to 28 percent, significantly higher than the 22.2 percent OECD average (whose standard deviation is 5.6 percentage points). Thereafter, while the Greek dependency ratio rises further to 46.7 percent by 2035, the rest of the OECD also ages rapidly, with its ratio rising to 41.8 percent, while the standard deviation widens to 9.7 percentage points.

37

The most recent OECD Economic Surveys: Greece (OECD 1997b) discusses in depth the status of pensions and health care systems and offers detailed proposals for reform.

38

Among the other criteria for success could be reductions in the rate of inflation, the external current account deficit and a sustainable acceleration in the growth of aggregate output, while among the other factors which may either contribute to or detract from the achievement of success, however defined, are, inter alia, changes in the exchange rate and external competitiveness, shifts in monetary stringency, and agreement or reforms in incomes policies.

39

In this context, see Tanzi (1993). Alesina and Perotti (1994) find that reductions in government wage bills tend to depreciate relative unit labor costs, thereby improving competitiveness and the macroeconomic response to fiscal adjustment.

40

McDermott and Wescott (1996) also find that higher domestic and foreign economic activity increase the probability for success.

42

The data for 1997 are official projections.

43

Prepared by Charalambos Christofides.

44

Annex I has a description of the raw data together with the procedures followed in constructing the variables of interest. Because the data is proprietary, it cannot be presented in this paper.

45

Of course, expected depreciation could still affect the risk-premium if investors thought that high expected depreciation was a signal of broader macroeconomic difficulties or a weak government, or an indication of broader repayment difficulties (especially given large foreign currency obligations).

46

Section D below will construct benchmark spreads to correct for this. The benchmark spreads will be comparable across time.

47

There is a paucity of theoretical, and even empirical, papers on the determinants of foreign currency denominated spreads. Related literatures exploring the determinants of the causes of default for consumer credit are almost exclusively empirical, utilizing large databases to correlate defaults with observable characteristics of the consumer. By necessity, this approach is also followed in this paper. In this regard, it is interesting to note that the literature on consumer credit often finds that the best predictor of a continued ability to service debt is whether a consumer has been promptly repaying existing debt. Greece, and other EU countries, have not rescheduled or otherwise downgraded their foreign exchange debt in recent memory, which is an important reason why their spreads tend to be lower than comparable developing countries that have recently rescheduled their foreign exchange debt.

48

See Annex I for a more detailed description.

49

The most promising method would use a panel data set involving a large number of countries and their spreads. Estimating a common trend would capture the global demand factor, which could be included as a variable in the single country regressions. This approach was followed by Favero, Giavazzi, and Spaventa (1996), but even though they restricted their attention to large industrial countries, their sample could only be extended back to 1991. Clearly, assembling a data set for a large number of small industrial and developing countries would be a large project unto itself. Therefore, this paper uses time series techniques to control for the global demand factor.

50

Other currencies were even less represented in the sample than the yen.

51

MA specifications with longer lags could not be estimated given the non-synchronous nature of the data (as the recommended iterative estimation algorithm breaks down in this case).

52

Prepared by Charalambos Christofides.

53

Deposit rates reflect the true cost to banks of raising funds, and posted lending rates refer to the average rate charged by commercial banks on short-term loans to finance working capital for industry.

54

Indeed, the largest state bank has used all of its 1996 profits for such provisioning.

55

One half of required reserves is nonremunerated, while the other half earned 12.5 percent. Since January 1997, the rate of remuneration was reduced to 11 percent.

56

With the inclusion of foreign currency deposits in required reserves, the share of liabilities belonging to the reserve base has also increased. This effect is only indirectly/partially captured in this study, through the effect of this change on the deposit rate.

57

At end–1993, EFTE was reduced from 8 to 4 percent. This explains why the spread dips after end–1993.

58

That is, ignoring all second-order terms, which are terms involving multiples of small shares with interest rates, shares with shares, etcetera.

59

It was shown earlier that the relevant term is the reserve ratio multiplied by the deposit rate.

60

Prepared by Daniel C. Hardy.

61

This number includes one former specialized credit institution that still concentrates on lending to one sector.

62

Commercial banks cannot directly participate in the stock exchange, and instead own stock brokerages. This restriction is maintained under a derogation from the EU’s Financial Services Directive, which, however, is due to expire in 1999.

63

The concentration levels per se are not high compared with those in several other European countries. See for example Lewis, Alfred, and Gioia Pescetto, 1996, EU and US Banking in the 1990s, Academic Press, London.

64

Commercial real estate lending is negligible except for the infrastructure-related loans granted by the industrial development banks.

65

See SM/967249, Supplement 1, 10/7/96, France—Selected Issues; and OECD, Bank Profitability, Paris, 1996 for comparisons.

66

This difference may in part reflect accounting practices. The large state-controlled institutions are long-established and largely own their own buildings, which have already been fully depreciated.

67

The statistics in Table 11 can be compared with those in Table 8 of SM/96/249 Supplement 1.

68

Reportedly the retailing of government securities has been quite lucrative for banks, with commissions of up to 45 basis points.

69

Prepared by Daniel C. Hardy.

70

A fuller description of the current structure, performance and regulatory framework of the Greek banking system is contained in section V.

71

A selection of recent research and survey articles can be found in the Journal of Banking and Finance, Vol. 17, Nos. 2-3, 1993.

72

Some banks were established during the sample period, and for others data are not available for all years. The total sample comprised 106 observations.

73

Some of the banks in the sample were partially or wholly owned by other banks, but because they claim operational independence they are included separately.

74

In a “fixed effects” regression, a dummy variable is included for each individual of the stratification, in this case, for each bank. Including these dummy variables in effect removes the mean of each variable for each bank. Thus, in the “fixed effects” regression the variation across time of the dependent variable is related to the variation across time of the explanatory variables; variation between individuals is captured by the dummies. Consequently, the effect of explanatory variables that are constant across time (such as an intercept or a dummy for state control) are not identified.

75

Three banks had not yet begun operations in 1990, so 33 observations were available.

Same Series

Other IMF Content

Other Publishers

  • Collapse
  • Expand
Cover IMF Staff Country Reports
Greece: Selected Issues and Statistical Appendix
Author:
International Monetary Fund
Volume/Issue:
Volume 1997: Issue 089
Publisher:
International Monetary Fund
ISBN:
9781451816075
ISSN:
1934-7685
Pages:
133
DOI:
https://doi.org/10.5089/9781451816075.002
Figure 1.

Greece: Per-Capita Income at 1995 PPP US$
Figure 2.

Greece: General Government Current Revenues, 1990-95

(Period average)
Figure 3.

Greece: General Government Current Expenditures, 1990-95

(Period average)
Figure 4.

Greece: General Government Current Primary Expenditures, 1990-95

(Period average)
Figure 5.

Greece: General Government Transfers, 1990-95

(Period average)
Figure 6.

Greece: Effective Average Tax Rates, 1965-94

(In percent)
Figure 7.

Greece: Evolution of the Spread and Maturity, 1980–96
Figure 8.

Greece: Macroeconomic Developments, 1980–96
Figure 9.

Greece: Actual Vs. Predicted Spread and Residual, 1980–96
Figure 10.

Greece: Actual Vs. Benchmark Spread, 1980–96
Figure 11.

Greece: Lending-Deposit Interest Rates

(In percent)
Figure 12.

Greece: Break-Even Lending Rate

(In percent)
Figure 13.

Greece: Lending-Deposit Spread and its Determinants

(In percent)
Figure 14.

Greece: Banking System Share Prices
Figure 15.

Greece: Banks’ ROE and asset shares