Chapter

VII Strategies for Euro Adoption

Author(s):
International Monetary Fund
Published Date:
April 2005
Share
  • ShareShare
Show Summary Details

Countries preparing to adopt the euro will need strategies for meeting both the Maastricht criteria and any conditions that from a domestic viewpoint are needed to ensure a successful experience in the euro area. Macroeconomic policy before euro adoption will, therefore, need to focus on several key tasks: reducing government budget deficits; lowering inflation, or keeping it low; and choosing a central parity and minimizing variations from it. All this must be done in an environment that remains vulnerable to volatile capital flows and demand booms.

A central consideration will be how countries can prepare for euro adoption in a way that minimizes the risks—especially those stemming from the vulnerabilities discussed in Section V—inherent in a major regime change. Three particular challenges stand out. First, plans for supporting policies—monetary, fiscal, and structural—need to be clearly communicated to markets and focused on the objective of smooth integration in the euro area. Second, monetary frameworks need to be designed to provide maximum protection from exogenous sources of capital market volatility. Insofar as existing flexible exchange rate frameworks—a key element in several countries’ risk management strategies in recent years—will be incompatible with ERM2 and the exchange rate stability criterion, monetary policy frameworks must be revamped within the constraints of the Maastricht Treaty. The lessons from the experience of emerging market countries during the 1990s suggest that avoiding soft peg/narrow band (“middle ground”) exchange rate regimes will be important. Third, timing should be considered carefully before entering ERM2. Particularly if ERM2 pushes countries toward “middle ground” exchange rate regimes, short stays with well-focused policies would reinforce a potentially stabilizing role of the central parity, while long stays with policies that are not in line with requirements for euro adoption would test the mettle of almost any monetary policy framework.

Conditional on the existence of positive net gains from joining the euro area, success in developing policy strategies to address these issues should be the ultimate determinant of decisions on the precise timing of euro adoption. Not only must a country galvanize the political will to meet the Maastricht criteria, but also it must establish its credibility sufficiently strongly to garner the support of the European institutions—a key ingredient to securing the confidence of markets.

Taming Fiscal Deficits

Fiscal adjustment is likely to be the most onerous hurdle for the CECs apart from Slovenia. Following rather rapid expansions in recent years, general government deficits relative to GDP in 2003 ranged from 1.8 percent of GDP (Slovenia) to 6.6 percent of GDP (Czech Republic) (Table 7.1). Although cyclical weakness contributed to this widening, the bulk of the change was structural and will require structural measures to reverse. Thus, in 2003, even if output gaps had not existed, deficits relative to GDP would have ranged from 1.4 percent of GDP in Slovenia to 5.9 percent of GDP in the Czech Republic.

Table 7.1.CECs: Cyclical and Structural Fiscal Balances, 20031(In percent of GDP)
Czech Republic2HungaryPoland3Poland4Slovak RepublicSlovenia
Headline balance (ESA-95 basis)–6.6–5.9–4.0–5.6–3.6–1.8
Structural balance–5.9–5.4–2.7–4.3–3.1–1.4
Memorandum items
Interest payments1.23.93.03.12.41.9
General government debt31.359.145.449.142.826.7
Sources: IMF staff estimates.

Fiscal balances on an ESA-95 basis. General government debt on a GFS basis.

Excludes from the deficit and debt government guarantees that have not been called.

Including second-pillar pension scheme in general government.

Second-pillar pension scheme outside of general government.

Sources: IMF staff estimates.

Fiscal balances on an ESA-95 basis. General government debt on a GFS basis.

Excludes from the deficit and debt government guarantees that have not been called.

Including second-pillar pension scheme in general government.

Second-pillar pension scheme outside of general government.

How Much Fiscal Adjustment Is Needed?

Clear yardsticks for the appropriate fiscal position in the CECs prior to adopting the euro do not exist. The Maastricht criteria represent essential minimum conditions, but it will be important to determine if the specific conditions in the CECs argue for tighter standards. In this light, several considerations are relevant: prudent debt limits for emerging market countries, risks of credit and demand booms in the run-up to or soon after euro adoption, and the need to remain within SGP limits on the deficit once inside the euro area.

Fundamentally, optimal fiscal balances are linked to judgments about sustainable public debt levels. For several reasons, the maximum prudent public debt ratio in the CECs is likely to be lower than in the existing euro-area countries.66 These are largely related to greater inherent volatility and its implications for the sustainability of public debt.67 First, revenues are lower and, owing to greater underlying volatility of activity, more variable than in the euro area. Second, interest costs are likely to be more volatile in the CECs than in the euro area. This is because a substantial portion of domestic debt is still of relatively short maturity, making interest costs more sensitive to changes in policy interest rates. Third, with less of a track record of commitment to debt sustainability than in industrial countries, the CECs cannot increase debt levels much without lenders becoming concerned about sustainability issues.

Actual participation in EMU, however, will make maximum safe debt ratios in the CECs higher than in other comparable emerging market countries. Specifically, while IMF (2003d) concludes that the sustainable public debt level for a typical emerging market economy may be as low as 25 percent of GDP, the CECs, with the security of the euro policy framework, should be able to sustain higher levels of debt. Applying the methodology of IMF (2003d), which considers the mean and standard deviation of fiscal revenues and expenditures, as well as the gap between the real interest rate and real growth rate, suggests a prudent public debt ratio of around 45 percent of GDP for the CECs. This would argue for keeping debt ratios in the Czech Republic, Slovak Republic, and Slovenia near current levels or even allowing some modest increase, while in Hungary and Poland debt ratios should be lowered.

Lending further support to keeping public debt around 45 percent of GDP is the finding that, in emerging markets, public debt exhibits a stabilizing tendency only if the debt ratio lies below 50 percent of GDP (IMF, 2003d). This reflects a fiscal policy reaction whereby a rise in public debt is countered by a higher primary surplus. Sustaining a debt stock of this magnitude would require the CECs to run a primary surplus of ½ of 1 percent of GDP on average. Factoring in interest costs, this would imply a headline deficit of about 1¾ percent of GDP over the cycle.

From a shorter-term standpoint, fiscal policy will be the central macroeconomic policy tool for managing the effects of possible credit and demand booms. In view of the difficulty of shifting the stance of fiscal policy rapidly when discretionary spending is a rather small share of the total, this will require that fiscal balances be positioned well in advance of such booms to exert the necessary restraint on demand to prevent overheating and other unsustainable strains on the economies.

Finally, from a more practical standpoint, prudence argues for fiscal balances well below the Maastricht/SGP limit of 3 percent of GDP. Two considerations are relevant here. First, conservative targets would prevent adverse cyclical conditions from derailing schedules for attaining the Maastricht deficit criterion. And while budget sensitivity to the cycle is relatively small in the CECs, output variability can be large, creating the potential for sizable variations in budget balances. Second, entering the euro area with a structural budget deficit at the SGP limit of 3 percent of GDP or, worse yet, a headline deficit at the limit during a cyclical upswing, would put the early years in the euro area at risk of requiring a procyclical reduction in the deficit.

These considerations point to the need to calculate a maximum prudent fiscal deficit. This is the level of the structural balance that would, for a large range of past variations in output growth, allow full play of automatic fiscal stabilizers while keeping the headline deficit at or below 3 percent of GDP (Box 7.1). Simple calculations suggest that a structural fiscal deficit of about 1–2 percent of potential GDP would accommodate all but the most severe cyclical downturns while keeping headline deficits below 3 percent of GDP. Such structural deficits would also help offset excessive demand pressures attendant on credit or demand booms and over time would stabilize debt at safe levels. Based on estimated 2003 outturns, the CECs (excluding Slovenia, where the structural deficit is currently below this level) would need to reduce their structural deficits relative to GDP by between 1½ percentage points (in the Slovak Republic) and 4¾ percentage points (in the Czech Republic) by the time of their test periods to achieve their prudent fiscal deficits.

Box 7.1.Calculating Prudent Structural Fiscal Deficits

Calculating maximum prudent fiscal deficits involves three steps: estimating potential GDP and output gaps; calculating the cyclical sensitivity of the budget; and, based on the historical volatility of GDP and the budget sensitivity to the cycle, calculating a buffer that could accommodate automatic stabilizers in all but the most severe economic downturns. Each of these steps can be contentious. The approach here is to apply widely used techniques uniformly for each country, despite some well-known shortcomings.

A Hodrick-Prescott (HP) filter is used to estimate potential GDP, the basis for calculating output gaps. This was the primary method used by the EC to calculate output gaps before it switched to a production function approach in 2002. Data limitations rule out the latter for the CECs. To reduce the end-point bias inherent in the HP filter, the time series for GDP are extended through 2008 with forecasts from the IMF’s World Economic Outlook (WEO).

Calculations of the cyclical sensitivity of the budget assume that only tax revenues are sensitive to the economic cycle. With low unemployment benefits relative to GDP in the CECs, cyclical influences on spending are small. Following the WEO methodology for estimating cyclically adjusted deficits, current and lagged elasticities of tax revenue (including social security contributions) to GDP are assumed to be 1.0 and 0.1, respectively. No allowance is made for cross-country differences in the composition of tax revenues or progressivity of taxes.1 Output shocks are assumed to be evenly distributed over individual GDP components so that their effect on revenues is given by the average elasticities.

The prudent fiscal deficit is defined as the structural deficit that would allow full play of automatic stabilizers if the output gap widened to two standard deviations below its post-1995 average.2 Whether this average is a good gauge for the future is debatable: transition-related shocks are likely to diminish, but output variability could increase after euro adoption if own monetary policy had previously been effective in smoothing output swings. The buffer is calculated as the product of the cyclical sensitivity of the budget and this “bad draw” output gap. The prudent fiscal deficit is defined as the Maastricht deficit ceiling minus the prudent buffer. Results (see table) imply that adopting the euro with structural fiscal deficits of between 1¼ and 2⅓ percent of potential GDP (levels vary by country) would accommodate all but the most severe cyclical downturns without breaching the Maastricht ceiling.

Cyclical Sensitivity of Fiscal Deficits and Prudent Fiscal Buffers
Cyclical Sensitivity1Prudent Fiscal Buffer2Maximum Output Gap Accommodated3Implied Maximum Structural Fiscal Deficit upon Euro Adoption4
(In percent of potential GDP)
Czech Republic0.421.814.31.19
Hungary0.471.102.41.90
Poland0.431.854.41.15
Slovak Republic0.421.433.41.57
Slovenia0.440.671.52.33
Source: IMF staff calculations.

Impact of a 1 percentage point increase in the output gap on the fiscal balance (as a percentage of GDP).

The budgetary effect of a 2 standard deviation widening of the output gap from the historical mean.

Historical mean output gap (1995–2002) plus 2 standard deviations.

Three percent of GDP Maastricht and SGP headline deficit ceiling minus the prudent fiscal buffer.

Source: IMF staff calculations.

Impact of a 1 percentage point increase in the output gap on the fiscal balance (as a percentage of GDP).

The budgetary effect of a 2 standard deviation widening of the output gap from the historical mean.

Historical mean output gap (1995–2002) plus 2 standard deviations.

Three percent of GDP Maastricht and SGP headline deficit ceiling minus the prudent fiscal buffer.

1 Several CECs will change their tax systems in the next few years to harmonize with the EU and improve competitiveness. This will probably move the composition of tax receipts and the cyclical sensitivity of the budget closer to the OECD average.2 Assuming shocks are normally distributed, only 2½ percent of shocks would fall below this level of the output gap.

These estimates are similar to recent EC reestimates of “minimum cyclical safety margins” for existing euro-area members. They find that on average countries should run deficits of 1.4 percent of potential GDP to stay under the 3 percent limit: their estimates of fiscal positions consistent with minimum cyclical safety margins range from a surplus of 0.1 percent of potential GDP for Luxembourg to a deficit of 2.1 percent for Austria.68 But estimates by Corricelli and Ercolani (2003) using a similar methodology on a slightly different group of CECs and allowing for larger cyclical swings indicate the need for fiscal deficits averaging some 0.7 percent of potential GDP.

Reaching Prudent Fiscal Deficits: The Immediate Implications for Demand and Output

Reducing fiscal deficits to prudent limits will have significant effects on aggregate demand. Large fiscal imbalances have produced substantial increases in public debt and in several cases reflect large and poorly targeted social transfers and subsidies, and reining them in should have positive effects on medium- to long-term growth. But the short-term effects on aggregate demand, in and of themselves, will almost inevitably be negative.69 At the same time, however, approaching euro adoption should entail an offsetting easing of monetary conditions and, therefore, a shift in the fiscal-monetary policy mix that puts downward pressure on exchange rates relative to a no-policy-change scenario. This section explores the effects on demand and output from needed fiscal adjustments and examines the influences that might simultaneously offset any negative effects.

Estimates of the impact of fiscal consolidation on growth vary widely from strongly negative (traditional Keynesian view) to zero or even positive (Box 7.2). Background work for this paper included a panel estimation of fiscal multipliers for the current euro-area countries, explicitly allowing for the impact of foreign demand, exchange rates, and interest rates. The estimates, based on data for 1980–2001, suggest that fiscal multipliers averaged around 0.45 in the adjustment year (Table 7.2 and Appendix 7.1). This is roughly in the middle of the range of multipliers found in the literature and reported in Box 7.2. They also suggest a strong impact of partner countries’ import growth on own output growth and favorable output effects from a depreciation of the real exchange rate. No statistically significant effect could be found for real short-term or long-term interest rates.70 Individual country estimations find (as expected) weaker fiscal multipliers and stronger effects from partner countries’ growth in smaller and more open countries (see Appendix 7.1). These results are likely sensitive to the composition of adjustment—a feature that could not be included in these aggregate estimates. Detailed empirical studies of OECD countries in the 1980s and 1990s suggest that credible expenditure restraint—especially on transfers—increases the likelihood that adjustments are sustained and lessens the growth-dampening effects.

Table 7.2.GDP Growth and Its Sources in EMU Countries, 1980–20011
Explanatory Variables
Lagged GDP growth0.27(3.0)
Fiscal impulse0.46(4.5)
Partner countries’ imports0.25(5.3)
Real short-term interest rateNot significant
Real long-term interest rateNot significant
Real exchange rate (RER)–0.05(–2.3)
Lagged RER0.02(1.9)
Sources: OECD Analytical Database (June 2003); and IMF staff estimates.

GDP growth is the dependent variable; t-statistics in parentheses. See Appendix 7.1 for details. Results reported here are the averages of regressions II and III.

Sources: OECD Analytical Database (June 2003); and IMF staff estimates.

GDP growth is the dependent variable; t-statistics in parentheses. See Appendix 7.1 for details. Results reported here are the averages of regressions II and III.

Without sufficient recent history in the CECs from which to glean the effects of fiscal consolidation, the experience of the noncore EMU countries is used to draw some lessons for the CECs. These countries entered the three years leading up to their assessment period (1997 for all except Greece, and 1999 for Greece) with general government deficits of 2–9.3 percent of GDP and debt of 68–134 percent of GDP. The fiscal adjustment that Greece, Italy, and Spain undertook was substantial, in terms of both headline and structural primary deficits (Figure 7.1)—structural primary general government deficits were reduced by 1.3 percentage points of GDP a year on average in the following three years (Figure 7.2). Portugal, which had a more favorable initial position, had a much more moderate negative fiscal impulse during this period. Ireland, which did not have to reduce its fiscal deficit to meet the Maastricht criteria since the deficit was 3 percent of GDP or lower throughout the 1990s, is not included in the analysis. Alongside significant fiscal consolidations, real short- and long-term interest rates in the noncore countries continued a sharp decline, which had begun in the early 1990s.

Figure 7.1.Noncore Euro-Area Countries and CECs: Fiscal Situation Before Euro Adoption1

(In percent of GDP)

Sources: OECD Analytical Database (June 2003) (for euro-area countries); and IMF staff estimates (for CECs).

1 On ESA-95 basis. For noncore euro-area countries, three years before the fiscal criterion assessment year (1994 for Ireland, Italy, Portugal, and Spain; 1996 for Greece). For the CECs, 2003.

2 Including second-pillar pension funds in the general government.

3 Second-pillar pension fund outside of general government.

4 First bar indicates debt including second-pillar pension funds in the general government. Second bar indicates debt with second-pillar pension fund outside of general government.

Figure 7.2.Southern Euro-Area Countries: Policy Mix, External Setting, and Growth

Source: OECD Analytical Database (June 2003).

1 Cyclically adjusted primary general government balance.

2 Three-month money market rate adjusted for 12-month lagged consumer price inflation. Year average.

The experience of the late-adjusting noncore countries suggests that, even with a negative impact on growth, significant fiscal consolidation need not produce a slowdown. The adverse impact of fiscal withdrawal on growth can be mitigated by favorable foreign demand: indeed, respectable GDP growth in the late-adjusting noncore countries owed much to particularly strong growth in partner countries’ import demand during that period (Table 7.3; Figure 7.2), with some further support from falling interest rates. At the same time, the feasibility and success of fiscal consolidation depends on general economic conditions (von Hagen, Hallett, and Strauch, 2001); for small open economies, these are largely externally driven. Thus, fiscal consolidation in the late-adjusting noncore countries—measured by changes in the structural primary deficit—was largest when robust economic growth in Europe swelled partner countries’ import growth, as in 1990 and 1991, and again in the second half of the 1990s (Table 7.3; Figure 7.2). In contrast, with Europe in recession in 1993, budgets prepared for 1994 contained little or no fiscal consolidation.

Table 7.3.Southern Euro-Area Countries: Policy Setting and Growth in the Run-Up to Maastricht

(Three-year average, in percent)1

Fiscal Impulse2Partner-Country Import GrowthReal Short-Term Interest RatesGDP Growth
Three-year averageThree-year averageAverage of previous three yearsThree-year averageAverage of previous three yearsThree-year average
Greece–1.28.78.2–1.2–1.83.5
Italy–1.64.39.55.06.32.0
Portugal–0.32.68.24.46.13.9
Spain–1.04.18.74.05.93.1
Sources: OECD Analytical Database (June 2003).

The three-year period ends in the assessment year (1997 for Italy, Portugal, and Spain; 1999 for Greece).

Change in the structural primary deficit (in percent of GDP).

Sources: OECD Analytical Database (June 2003).

The three-year period ends in the assessment year (1997 for Italy, Portugal, and Spain; 1999 for Greece).

Change in the structural primary deficit (in percent of GDP).

Box 7.2.Empirical Estimates of Short-Term Fiscal Multipliers

Traditionally, estimates of the short-run impact of fiscal policy on growth come from simulations using structural macro models. Such models, with a prominent role for Keynesian effects, suggest expenditure multipliers for European countries ranging from 0.6 to 1.5, with an average of 0.9, and somewhat smaller revenue multipliers (see Brunila, Buti, and in ‘t Veld, 2002; Hunt and Laxton, 2003; and, for a survey, Hemming, Kell, and Mahfouz [HKM], 2002). While a strength of these structural macro models is that they shed light on the channels through which effects take place, a drawback is that some of the relationships are imposed—based on economic theory—rather than reflecting empirical results.

In circumstances of high government debt, the credibility effects of a fiscal contraction can offset the traditionally assumed Keynesian effects. Indeed, the experience of some European countries that undertook fiscal consolidation in the 1980s and 1990s generated a literature on “expansionary fiscal contractions” (see Alesina and Ardagna, 1998; Perotti, 1999; and EC, 2003). Surveys of these studies by HKM and the EC conclude that expansionary fiscal contractions have occurred and are most likely when (1) initial conditions are “appropriate” (high public debt and interest rate premia); (2) the adjustment is accompanied by monetary easing and/or a depreciation; and (3) the adjustment is based on expenditure cuts rather than tax increases. The impact of the composition of adjustment is assessed in detail by Alesina and Ardagna (1998); Alesina, Perotti, and Tavares (1998); and IMF (2003d). In line with this literature, von Hagen, Hallett, and Strauch (2001) find in a panel estimation that fiscal multipliers have become insignificant since 1990 for EU countries. They attribute their result to a “Maastricht effect”: as fiscal consolidation improved the prospects for EMU and thereby for macroeconomic stability, the Keynesian effects of fiscal contractions were offset by positive confidence effects.

Recent estimates of fiscal multipliers from structural VAR models, obviously with a strong empirical element, cover a wide range of values. While studies focusing on the United States typically find large multipliers in line with Keynesian theory (with peak multipliers between ½ and 1½), recent studies of large European countries reach mixed conclusions. Aarle, Garretsen, and Gobbin (2001) find considerable variation in the size and signs of multipliers for EU countries. In line with von Hagen and others, Perotti (2002) finds that the effects of a fiscal expansion on GDP have become weaker over time. He finds that the spending multiplier has come down from a peak of 1.7 during 1960–74 to 0.8 during 1974–2000 in Germany and from 0.9 in 1960–79 to –0.1 in 1980–2000 in the United Kingdom. IMF (2003b) finds for Spain spending and revenue multipliers of 1.4 and 2.8, respectively. IMF (2002) finds for the United Kingdom a multiplier for social transfers of around 0.7, but statistically insignificant multipliers for other fiscal policy variables.

What conclusions can be culled from these estimates for the CECs? Estimates of fiscal multipliers for the existing EU countries range from below zero to more than unity. In contrast to the EU countries that experienced “expansionary fiscal contractions,” however, public debt ratios in the CECs are relatively low, reducing the likelihood that non-Keynesian effects would dominate. But Keynesian effects themselves are likely to be smaller in the more open CECs than in the large EU countries on which most of the results from macro-model simulations are based. This would suggest that multipliers in the CECs are significantly larger than zero but significantly smaller than unity.

These results suggest a few conclusions for the CECs. First, the direct effect on growth of the fiscal adjustment needed to position countries for euro adoption will almost inevitably be negative. Extrapolating the estimation results for the euro area using openness indicators for the CECs suggests, first, that fiscal multipliers could range from around ⅓ in the smaller CECs to almost ½ in Poland. Illustrative calculations based on the estimated required reduction in structural primary fiscal deficits suggest that the cumulative impact on GDP would range from 0.4 percent in the Slovak Republic to over 2 percent in Poland (Table 7.4). Second, these direct effects should be partly offset by an easing of monetary conditions as interest rates converge to euro-area levels. The expansionary effect of lower interest rates appears to be small in practice, but a more depreciated real exchange rate would generate additional growth. For example, a 5 percent real depreciation would raise cumulative growth by 0.25 percent. Third, the success of fiscal consolidation will depend significantly on the EU economy as the major source of demand for CEC exports. Each additional percentage point of partner-country import growth would add 0.2–0.3 percentage point to GDP growth. Fourth, countries should be opportunistic in implementing fiscal adjustments: a cyclical recovery in Western Europe would provide ideal circumstances for decisive adjustment. Fifth, fiscal adjustment that relies on credible expenditure reductions rather than ad hoc measures and revenue increases is likely to minimize growth-dampening effects.

Table 7.4.CECs: Impact of Fiscal Consolidation
Required Primary Fiscal Adjustment1 (percent of GDP)Fiscal MultiplierCumulative Impact (percent of GDP)
Czech Republic5.40.321.7
Hungary2.00.350.7
Poland2,33.10.461.4
Poland2,44.60.462.1
Slovak Republic1.40.310.4
Slovenia
Sources: OECD Analytical Database (June 2003); and IMF staff estimates.

The difference between the structural primary deficit relative to GDP in 2003 and the structural primary deficit relative to GDP consistent with the maximum prudent deficit calculated in Box 7.1.

Compared with 2004, assuming an increase in the deficit to 6.9 percent of GDP as projected in budget proposals.

Including second-pillar pension in general government.

Second-pillar pension outside of general government.

Sources: OECD Analytical Database (June 2003); and IMF staff estimates.

The difference between the structural primary deficit relative to GDP in 2003 and the structural primary deficit relative to GDP consistent with the maximum prudent deficit calculated in Box 7.1.

Compared with 2004, assuming an increase in the deficit to 6.9 percent of GDP as projected in budget proposals.

Including second-pillar pension in general government.

Second-pillar pension outside of general government.

These considerations suggest that it should be possible to devise a strategy that significantly curtails possible adverse effects of fiscal adjustment on growth. But in any event, delaying fiscal adjustment, even if euro adoption is indefinitely put off, is unlikely to be a viable option. It is true that most CECs’ moderate debt ratios and relatively high potential growth rates hold at bay serious imminent threats to fiscal sustainability. But external current account deficits that will flirt with the limits of market tolerance as output approaches potential make it unlikely that fiscal deficits of recent magnitudes can continue indefinitely outside a currency union.

Reaching Prudent Fiscal Deficits—How Can It Be Done?

The adjustment required to reach the prudent fiscal balance in the four high-deficit CECs will need to come mainly from expenditure restraint. Two main considerations militate toward this conclusion. First, revenues as a share of GDP—while below the euro-area average—are (with the exception of the Slovak Republic) close to or somewhat above ratios in the EU adjusted for per capita income differences (Figure 7.3). Moreover, tax competition, particularly as regards corporate income taxation, is likely to drive rates down in the future. While mandated harmonization of indirect taxes with the EU will raise some revenue, the addition will be small in most countries, except in the Czech Republic, where it will add about 1½ percentage points. Scope for further indirect tax increases will be limited by the inflation objective. Second, primary expenditure relative to GDP in several CECs is close to or above the euro-area average, nearly 4 percentage points on average higher than in the southern noncore countries, and well in excess of ratios adjusted for per capita income. As channeling fiscal adjustment through expenditure restraint invariably requires enormous political will and leadership, it will test the mettle of governments. Credibly pursued, however, it will also be the strategy that both commands the strongest support of markets and minimizes adverse effects on growth. The remainder of this section examines the options for such adjustment.

Figure 7.3.CECs and Euro-Area Countries: General Government Revenue and Expenditure, and per Capita GDP1

(In percent of GDP)
(In percent of GDP)

Sources: Eurostat; IMF, World Economic Outlook; and IMF staff estimates.

1 For the euro-area countries, average for 1995–2003. For the CECs, 2003 estimates. Regression line based on OLS estimation over the EU-15 countries. Classifications of expenditure items for the CECs may not be fully comparable across countries. For Hungary, investment spending includes capital transfers.

Interest savings

The potential fiscal windfall from nominal interest rate convergence and declining public debt is substantially smaller in the CECs than it was in 1994 for several euro-area countries (Table 7.5 and Figure 7.4). Lower interest rates are estimated to have cut interest payments relative to GDP by 2½–5 percentage points in Greece, Italy, and Portugal, and by about ½–1 percent of GDP in Belgium, France, Ireland, and Spain during 1994–97.71 In France and Spain, increasing debt levels fully offset the effect of these gains on gross interest payments, but in Ireland a sharp drop in the debt ratio further contributed to the 2¼ percent reduction in interest payments relative to GDP. Of course, with inflation falling, not all the drop in interest rates reflected lower real interest rates. Thus, adjustments in headline deficits—the key test for the Maastricht criteria—were not always fully matched by lower operational deficits, in this context probably a better measure of true fiscal adjustment. In all, lower interest payments accounted for between half and four-fifths of actual adjustment in headline deficits from 1994 to the Maastricht test year in Belgium, Greece, Ireland, and Portugal.

Table 7.5.Selected Euro-Area Countries: Contribution of Interest Payments to Fiscal Adjustment(In percent of GDP, unless otherwise indicated)
BelgiumFinlandFranceGreeceIrelandItalyPortugalSpain
General government debt
1993138.255.945.3110.196.5118.159.158.4
1996130.257.157.1105.8474.1122.162.968.1
Interest payments
19949.64.23.613.96.111.46.64.7
19978.04.33.68.353.89.44.24.8
Change in interest rates, 1994–97 (in percentage points)
Benchmark bond yield1–2.0–3.1–1.6–14.46–1.6–3.7–4.1–3.6
Effective rate2–0.80.0–1.6–4.86–1.2–2.0–4.5–1.0
Change in interest payments–1.60.10.0–5.66–2.3–2.0–2.40.1
Of which: Contribution of–1.00.0–0.9–5.16–0.9–2.4–2.8–0.7
interest rate change3
Memorandum items
Total fiscal adjustment, 1994–973.04.22.57.663.26.63.63.4
Primary fiscal adjustment, 1994–971.44.32.51.160.94.61.23.5
Spread on benchmark bond yield relative to Germany (in percentage points)
19940.92.20.413.91.13.73.63.1
19970.10.3–0.11.850.71.20.70.8
Sources: Eurostat; and IMF staff calculations.

Yield on 10-year government bond.

Interest payments divided by government debt at the end of the previous year.

Change in the effective interest rate multiplied by the debt stock at end-1996.

1998.

1999.

1994–99.

Sources: Eurostat; and IMF staff calculations.

Yield on 10-year government bond.

Interest payments divided by government debt at the end of the previous year.

Change in the effective interest rate multiplied by the debt stock at end-1996.

1998.

1999.

1994–99.

Figure 7.4.Euro Area: Fiscal Adjustment and Its Contributions1

(Changes in ratios to GDP over the period, in percentage points)

Sources: OECD; and IMF, World Economic Outlook.

1 At the general government level. For Greece, 1994 to 1999; for Portugal, 1995 to 1997; for all other countries, 1994 to 1997, except in bottom panel where the period is 1990–97 for all countries. A positive sign indicates an increase in revenues, a decrease in primary expenditures or interest payments, or a decrease in the deficit.

In contrast, the four high-deficit CECs stand to gain relatively small amounts from interest savings (Table 7.6). Based on staff forecasts of debt ratios in 2008—which show a strong increase in the debt ratio in the Czech Republic and Poland, significant declines in Hungary, and unchanged ratios in the Slovak Republic and Slovenia—and assuming that CEC interest rates converge to about 85 basis points above euro rates (as occurred in the southern noncore euro-area countries in their Maastricht fiscal test years), interest saving relative to GDP would range from about negative ¾ of 1 percentage point in the Czech Republic to 1½ percentage points in Hungary.72

Table 7.6.CECs: Expected Contribution of Government Long-Term Interest Rate Convergence to Fiscal Adjustment(In percent of GDP, unless otherwise indicated)
Czech RepublicHungaryPolandSlovak RepublicSlovenia
General government debt
200219.549.945.744.326.9
2008139.244.960.043.825.0
Interest payments
20031.23.93.02.41.9
20092.02.33.12.21.3
Effective interest rate on government debt (in percentage points) (2003)6.17.86.65.47.1
Assumed change in effective government bond yield (in percentage points)2–1.1–2.7–1.5–0.3–2.0
Reduction in general government interest payments–0.81.6–0.10.20.6
Memorandum item
Spread on benchmark bond yield relative to Germany (in percentage points) (2002)0.12.32.52.1n.a.
Sources: Eurostat; IMF World Economic Outlook (WEO); and IMF staff calculations.

WEO projections, GFS basis. For Hungary, assumes a combination of sustained moderate growth, strong fiscal adjustment in 2004–05, and some debt reduction through privatization.

Assumes that the effective rate converges during 2004–09 to 85 basis points above the euro-area interest rate.

Sources: Eurostat; IMF World Economic Outlook (WEO); and IMF staff calculations.

WEO projections, GFS basis. For Hungary, assumes a combination of sustained moderate growth, strong fiscal adjustment in 2004–05, and some debt reduction through privatization.

Assumes that the effective rate converges during 2004–09 to 85 basis points above the euro-area interest rate.

Fiscal implications of EU accession

During 2004–06, accession-related spending is expected to increase fiscal deficits in the CECs (although revenue gains from indirect tax harmonization will in some cases partially offset this effect). IMF staff scenarios suggest that net costs as of 2006 budgets could range from 1 percent of GDP in the Czech Republic and Slovak Republic to 1½ percent in Hungary and Slovenia. Costs are incurred from the obligation to contribute to the EU’s budget in the amount of about 1¼ percent of GDP. Revenues will come from the EU’s Structural Funds and the Cohesion Fund, support through the Common Agriculture Policy (CAP), and various other EU policies. During the initial three years, the EU will also pay compensating grants intended to offset the net fiscal cost of membership contributions. But at least during this period—for which the EU budget has been approved—additional demands on CECs’ budgets from EU membership will be positive owing to cofinancing obligations (averaging about one-third of the cost of EU-supported projects), loss of customs revenues from imports originating in the EU, the fact that not all EU receipts will pass through the budget, and countries’ decisions to top-up payments to farmers under the CAP.

Beyond 2006—a period for which the EU budget has not yet been agreed—the net impact of EU membership on the CECs’ budgets is more uncertain. If the twin principles of redistribution in favor of the less well-off member states and regions and promotion of economic integration through development of infrastructure networks are upheld (as suggested in the EC’s “Second Progress Report on Economic and Social Cohesion,” 2003), the accession countries could be eligible to receive relatively large grants on a per capita basis. Disbursement would, however, be contingent on meeting certain criteria, and not all the proceeds would benefit the budget. Moreover, the budget would still need to cover cofinancing, contributions to the EU, and indirect accession-related costs. There is therefore some speculation that the net effect on budgets from EU accession-related spending will rise after 2007.73 The more neutral assumption here is that the net effect will be unchanged from the 2006 level.

Infrastructure needs

Infrastructure networks in the CECs still to a large extent reflect their central planning origins, with underprovision in some sectors (for example, roads) and overcapacity in others (for instance, rail). Moreover, the task of building a socially efficient infrastructure network will be costly. Public gross fixed investment on a national accounts basis averaged 4 percent of GDP during 1999–2002, the same as in the southern noncore euro area, but well below the 7½ percent in the Asian “tiger” economies. User fees and public-private partnerships have defrayed only a small part of the cost of infrastructure building. For 2003, budgetary spending on investment by the CECs is estimated at between 3¼ percent of GDP (Poland) and 5½ percent of GDP (Hungary), generally above the euro-area average during 1995–2003 (3¾ percent of GDP) but below the average in the southern non-core countries (5 percent of GDP).

The added demands on infrastructure standards dictated by the acquis and general shortfalls in infrastructure quality make a strong case for sustaining or even increasing public investment spending in the run-up to euro adoption—particularly in countries where infrastructure investment has so far been relatively low. Moreover, taking advantage of time-bound EU funds will lower the cost to the country of a given investment, while slowing or even halting construction for one year may add to the overall cost of the project. It will therefore be important to avoid the experiences of Italy and Spain, for example, which cut public investment spending relative to GDP by 1–2 percentage points prior to euro adoption. Indeed, approximate estimates for the CECs suggest that projected transfers from the EU together with required cofinancing should raise total allocations for general government investment relative to GDP by about 1 percentage point, depending on a country’s specific circumstances.

Current spending

After accounting for prospective nondiscretionary influences on budgets from savings on interest payments, budgetary effects of EU accession, and harmonization of indirect taxes, the CECs will face a sizable residual adjustment that will need to come from primary current savings. The needed saving relative to GDP amounts to some 2½–2¾ percentage points for Hungary and the Slovak Republic; 4 percentage points for Poland; and about 4¾ percentage points for the Czech Republic (Table 7.7). Most of the CECs have already made significant starts to realize these savings in their 2004 budgets: the exception is Poland, where the budget envisaged a large deficit increase in 2004. Moreover, the recent Eurostat decision to view Poland’s second-pillar pension system as outside the general government will increase the medium-term burden of primary current adjustment to 5½ percentage points.

Table 7.7.CECs: Required Fiscal Adjustment Relative to Estimated 2003 General Government Budget Outturns1(Changes in ratios to GDP, in percentage points, except where indicated)
Czech RepublicHungaryPoland 20042Poland 20043Slovak RepublicSlovenia
(1) Structural deficit45.975.44.25.73.11.4
(2) Prudent structural fiscal deficit41.21.91.21.21.62.3
(1)–(2) Required fiscal adjustment relative to 2003 outturn to satisfy prudent fiscal deficit4.73.53.04.51.6–1.0
Nondiscretionary measures–0.10.8–1.1–1.1–0.8–0.8
Indirect tax harmonization1.60.60.20.20.00.0
Saving on interest payments5–0.81.6–0.1–0.10.20.6
Budgetary effect of EU accession6–0.9–1.4–1.2–1.2–1.0–1.4
Primary current adjustment needed4.72.74.15.62.4–0.2
Memorandum item
Headline deficit (ESA-95 basis)6.675.95.46.93.61.8
Source: IMF staff estimates.

Based on IMF staff estimates of 2003 outturns, except where indicated.

Including second-pillar pension scheme in general government.

Second-pillar pension scheme outside of general government.

In percent of GDP.

From Table 7.6.

Excluding indirect tax harmonization.

Excluding from the deficit and debt, government guarantees that have not been called.

Source: IMF staff estimates.

Based on IMF staff estimates of 2003 outturns, except where indicated.

Including second-pillar pension scheme in general government.

Second-pillar pension scheme outside of general government.

In percent of GDP.

From Table 7.6.

Excluding indirect tax harmonization.

Excluding from the deficit and debt, government guarantees that have not been called.

Spending priorities and efficiency considerations would argue against across the board, proportional, expenditure cuts. Rather, each country will need to scrutinize its own budget, looking for inefficient spending items. Measures, of course, will need to be tailored to specific conditions in the individual countries, but cross-country comparisons of the size of government and the composition of expenditure relative to GDP may help shed light on inefficiencies within the CECs’ budgets, and conclusions of recent IMF Article IV consultations suggest that cuts in subsidies and social transfers should be priorities in medium-term adjustment plans. As an illustration, for example, in each of these areas, moving to EU norms adjusted for per capita GDP would yield savings relative to GDP of about 5 percentage points in Poland, 3 percentage points in the Czech Republic and Hungary, and 1 percentage point in the Slovak Republic (see Figure 7.3).

The broad picture that emerges is that for most countries the needed fiscal adjustment could be achieved through changes that could well improve economic efficiency without diverting resources from, and in some cases increasing, infrastructure spending. For Hungary and Poland, in particular, scope exists to meet prudent budget targets wholly or largely through reductions in subsidies and social transfers. In the Slovak Republic, accelerated expenditure reform is the most sustainable way to reconcile the recent tax reform—based on a single uniform rate for income and value-added tax (VAT)—with continued fiscal consolidation. More efficient tax collection would also help in this endeavor. The Czech Republic will need a broader examination of the sources of excessive current spending.

Harnessing fiscal devolution to contribute to adjustment

Fiscal decentralization—ongoing in several of the CECs and driven in part by EU accession—is an added risk for the required deficit reduction if regions and local governments do not have appropriate incentives for prudent fiscal behavior. Local governments account for at least one-fifth of overall government spending in the Czech Republic, Hungary, and Poland, and this share is expected to increase in the future. But if well-designed, fiscal decentralization can coexist with and even help foster fiscal consolidation. Decentralization can mitigate inconsistencies between beneficiaries and funders that can give rise to excessive spending and higher deficits and public debt (von Hagen, Hallett, and Strauch, 2001). Indeed, Drummond and Mansoor (2002) find no statistical evidence linking decentralization with weaker fiscal consolidation in high- and middle-income countries with good governance.

The experience of the existing euro-area members offers some guidance on the scope for managing fiscal devolution to contribute to adjustment. In general, autonomous financial arrangements of the sub-national governments, often defined in law, impinged on the size of the contribution they made to the overall fiscal adjustment. As a result, two-thirds of fiscal adjustment was shouldered by the central government (see Figure 7.4). The notable exception was Spain, where an “internal stability pact” permitted more than the full adjustment in the general government balance to occur outside the central government.74 Regional tax revenues had been governed by tax-sharing arrangements with the center that were canceled in 1997, forcing greater reliance by regional governments on own revenues and thereby requiring them to implement better medium-term revenue and expenditure planning.

Similar types of reforms to provide more local autonomy over spending and taxation decisions helped, although less dramatically, achieve adjustment in other euro-area countries. In Italy, although only 20 percent of the deficit reduction was achieved at the regional and local levels, a 1992 reform of local public finances played a key role (Bordignon, 1999). The main principle of this reform was to abolish central government grants to regions and to assign local governments their own tax revenues so as to reduce their fallback reliance on central government bailouts. Thus, regions were made responsible for covering the cost of services they provided. Similarly, Finland, during 1993–98, replaced the system of central government grants—which financed 45 percent of local government expenditures—with revenues derived from locally sourced taxes. In addition, transfers from the center were delinked from actual local government outlays and tied to municipal demographic, social, and economic indicators. This reduced the possibility of gap-filling grants from the center and, with financing of local government expenditures mainly through own taxes, incentives for inefficient spending.

The experience in other countries speaks to the importance of strong control mechanisms. In Belgium, a group comprising representatives of the Ministry of Finance, regional governments, and the central bank was given responsibility for monitoring compliance by all parts of the government with Belgium’s Convergence Program. This group allocated the targeted adjustment across each level of government (including the regions and local communities). If one part of government breached its target, it was expected to revise its current budget to compensate for the earlier overshooting. This internal peer pressure proved successful in co-opting the support of local governments in fiscal adjustment (von Hagen, Hallett, and Strauch, 2001). Similarly, in Austria, an internal stability pact distributed the deficit permissible for the lower levels of government among the different entities.

These experiences suggest several steps for the CECs to help harness devolution so that it can contribute to fiscal adjustment. Specifically, the CECs should ensure that fiscal decentralization is accompanied by devolution of spending and revenue-raising authority, while retaining a strong coordination function at the central level.75 Clear mechanisms to ensure budgetary discipline at the lower levels are also needed—borrowing ceilings on local governments and rule-based and transparent grants and transfers from central governments that eliminate expectations of bailouts. Agreement on fiscal policy objectives at all levels is also vital.

Hungary and Poland were the first of the CECs to begin to decentralize—although local tax collection still accounts for less than 10 percent of general government tax revenue—and appear to be conforming to some of these lessons. Both countries established clear rules for revenue sharing, and local governments have some discretion in setting tax rates. Local government borrowing is limited, but in Hungary the restriction applies only to borrowing from banks. Hungary established a municipal bankruptcy law, which has been applied on several occasions, leading to the restructuring of local governments’ debts and helping to prevent bailouts by the central government.

Pension reform and public saving—what is the issue?

Various structural fiscal reforms recently implemented or under consideration—strongly backed by the international financial institutions—pose dilemmas for the CECs. Implementation, which is essential to securing sound fiscal trends over the long term, is likely to involve substantial up-front budgetary costs that could conflict with and therefore discourage needed fiscal reform. Health care, public enterprise restructuring, and labor market liberalization are among these reforms. Pension reform, however, provides the most stark example.

Each CEC either has recently introduced a mandatory defined-contribution pillar to its pension system (Poland and Hungary) or is considering doing so. As in several euro-area countries, second pillars will help diversify assets underlying future pensions and identify needed savings to render pension systems solvent. The up-front fiscal costs of introducing a fully funded pillar—shifting a segment of contributions out of the pay-as-you-go (PAYG) system while payment obligations from the pre-reform system remain intact—are typically substantial: transfers to the second pillar amount to 1¼ percent of GDP per year in Poland and ¾ of 1 percent of GDP in Hungary. Plans in the Slovak Republic envisage transfers to the second pillar of 1 percent of GDP a year in the near term. While these transfers should have little direct effect on economy-wide savings, they raise debt and conventional measures of the fiscal deficit.76 Eurostat recently ruled that pension contributions to the second pillar cannot be treated as intergovernmental transfers under ESA-95. The up-front cost of introducing a multipillar system—which can persist for several years—could have profound consequences for the pace of pension reform in the CECs, even though such reforms would enhance fiscal sustainability (Box 7.3). The European institutions will need to decide how to handle transitional costs in judging fulfillment of the Maastricht criteria and ensuring that countries realize savings needed to render pension systems solvent.

Controlling Inflation While Stabilizing the Exchange Rate: Will It Be Possible?

Current inflation rates in the CECs are broadly in the same range as those of existing euro-area members in the mid-1990s (Figure 7.5). CPI inflation rates in 2002–03 varied from a low of 0.1 percent in the Czech Republic to a high of 8.5 percent in the Slovak Republic, roughly comparable to the range of 1 to 8.9 percent in 1995 in existing euro-area members. Core inflation rates tell a similar story, although at 7.9 percent in 1996, core inflation in Greece was clearly an outlier. Responding to a number of strong anti-inflation measures, headline inflation was reduced to below the Maastricht inflation criterion—an average of 2.4 percent for the year leading to March 2000 for Greece, and 2.7 percent for the year leading to January 1998 for the others. Whether such efforts are necessary or feasible in the CECs is the subject of this section.

Figure 7.5.CECs and Noncore Euro-Area Periphery: Inflation Prior to Euro Adoption

(In percent, year average)

Sources: Eurostat; IMF, World Economic Outlook; IMF, International Financial Statistics; and country authorities.

One important consideration is how the Maastricht inflation criterion will be interpreted. The Maastricht Treaty specifies that inflation in the assessment period must be no higher than in “the three best-performing states in terms of price stability,” plus a margin of 1½ percentage points. Subsequently, in EC (1998), “best performing” was interpreted as the average inflation rate in the three EU countries with the lowest inflation. The September 2003 World Economic Outlook (IMF, 2003d) projects that in 2006–08 the three lowest EU inflation rates will average 1.4 percent, compared with a euro-area average of 1.8 percent. Thus, to meet the inflation criterion on this interpretation would require that the CECs limit their inflation rates to no more than 2.9 percent. This interpretation, however, predates the existence of an EMU-wide monetary policy. Now that the ECB has a formal inflation objective—close to but below 2 percent—it would seem reasonable that this would be used as the standard for the “best-performing” countries in terms of inflation. In this interpretation, the inflation criterion would be close to but below 3½ percent.

A second key consideration will be the likely contribution of B-S effects to inflation, particularly if nominal exchange rates were more stable during ERM2 than they have been in recent years. The few, properly measured estimates of B-S effects in the CECs suggest that real CPI-based exchange rates should rise by some 1–2 percent a year (Kovacs, 2002). B-S effects of this size would not necessarily be incompatible with meeting the inflation criterion, provided that nominal appreciations of 1–2 percent a year were tolerated or the base for the inflation criterion were taken to be the ECB inflation objective. Also, countries have some degrees of flexibility in meeting the inflation criterion: it must be met only in the test date year (not over the full two-year period when the exchange rate stability criterion is assessed); countries have some discretion over the choice of the test date year; and the 12-month period over which the criterion is tested need not coincide with a calendar year or the assessment year for the fiscal deficit criterion.

Box 7.3.Accounting for Pension Reform—The Current Debate in Context

At the heart of this debate is the question of how to record governments’ unfunded social security obligations. The current definition and treatment of social security schemes is identical in all macroeconomic statistical manuals—for example, System of National Accounts 1993 (SNA) (EC and others, 1993); European System of National Accounts 1995 (ESA-95), (Eurostat, 1996); and Government Finance Statistics Manual 2001 (GFSM) (IMF, 2001). Pension schemes arranged by government for all or a large segment of the labor force are referred to as social security schemes. Social security schemes may be funded (have identifiable reserves from which to cover all or part of obligations) or unfunded. Even where separate funds are identified, they remain the property of the government and not of the beneficiaries of the schemes (SNA, Annex 4.13). Under existing treatment, all social security contributions and all public pension payments are recorded as current transfer payments by households and government, respectively, even when the contribution entitles the contributor to a pension in the future. In contrast, funded pension schemes other than social security generate a liability equivalent to the present value of pension obligations accrued to the current date. SNA justifies the different treatment of social security on the grounds that “institutional arrangements in respect of social security differ from country to country, and the amounts raised and paid out in social security contributions and benefits may be deliberately varied in order to achieve objectives of government policy that have no direct connection with the concept of social security” (SNA, paragraph 4.112). But recognizing that government-sponsored social security schemes have at least a conditional obligation, GFSM requires a memorandum item in the government’s balance sheet indicating the present value of social security benefits payable in the future that have been earned under existing laws and regulations.

This treatment of government pension liabilities has been challenged on the grounds that pension liabilities represent a genuine obligation (even if subject to change) on which the government cannot realistically avoid payment and that current treatment is intransparent as regards the public finance effects of fiscal policies.1 Advocates for changing the treatment note the lack of comparability with alternative pension arrangements and that “privatizing” all or part of a social security schemes does not change the true net worth of the government, even though the government’s balance sheet position deteriorates. They propose the recording of an imputed expense each period equal to the estimated increase in new pension liabilities. On impact, this change would raise government debt, and could worsen the fiscal balance if accruing liabilities (an expense under the alternative treatment) exceed cash payments of benefits (the expense under the existing treatment), as might be expected if the labor force is larger than the number of pension beneficiaries. Payment of benefits would be recorded as a financing transaction, rather than an expense, and would reduce government liabilities.

Pending any changes in the definition and treatment of public social security schemes, the fiscal impact of pension reform in the CECs must be interpreted against current guidelines. A ruling by Eurostat in March 2004 clarifies instances where, in line with ESA-95, a pension scheme is not regarded as a social security scheme and, therefore, revenues (including contributions and interest) to, and benefits paid from, the scheme are not regarded as government revenue or expenditure. This ruling covers defined-contributions-funded schemes, even if they are managed by the government. Moreover, the existence of a government guarantee on minimum pension benefits is not sufficient grounds for classifying a pension system as a social security scheme. As a result of this ruling, countries will need to exclude from their ESA-95 fiscal accounts contributions to second-pillar schemes.

1 Arguments for not recognizing a liability for public pensions include: (1) the liability is not a legal obligation; (2) one government cannot commit a future government to make expenditures; and (3) the value of the liability is subject to large estimation error.

The potential for an inconsistency between the inflation and exchange rate stability criterion, however, cannot be discounted. First, B-S projections may be underestimated. Changes from past patterns can occur because of cyclical variations in investment, faster catch-up as countries derive the benefits of acceding to the EU, and shrinking differences in the size of nontradables sectors.77 Second, real appreciations in each of the CECs except Slovenia have been substantially larger—indeed, by a factor of 2 to 3 during 1995–2003—than estimates of B-S effects would suggest. Whether this is due to underestimates of B-S effects, other structural influences, or disequilibrating movements is unknown, but it leaves open the possibility that real appreciations of a similar size will persist. Were exchange rate movements reduced in ERM2, inflation would have to be higher to accommodate real appreciations of a size similar to recent rates. Third, if markets assume that countries will rely on nominal appreciation to accommodate (possibly overestimated) B-S effects, a bias toward upward pressure on currencies could develop. Any overshooting of equilibrium values would increase risks of disruptive changes in market sentiments.

The experience of the noncore EMU members adds weight to the case for interpreting the inflation criterion against the ECB inflation objective rather than the three lowest inflation rates in the EU. Having managed to squeeze inflation below the Maastricht ceilings in 1998 (2000 for Greece), each of the non-core countries except Finland and Italy have seen inflation rates rise to a broadly stable 3–4 percent (Figure 7.6). With a single monetary policy for all countries, considerable disparities in fiscal positions, and general wage moderation, national inflation rates averaging 1½–2 percentage points above the core euro area presumably reflect, at least in part, structural influences stemming from the catching-up process. In contrast, the three lowest inflation rates in the EU are currently in Germany, Austria, and Belgium, where income catch-up is not occurring. This evidence argues for recognition in the interpretation of the Maastricht inflation criterion that, once in the currency union, catching-up countries are unlikely to be able to sustain inflation below 3–4 percent.

Figure 7.6.Noncore Euro-Area Countries: Inflation Indicators After Euro Adoption

(Period average change, in percent)

Source: OECD.

1 Definition of the reference value for the inflation criterion is discussed in Section VI.

Both the EC and ECB have recognized the structural factors underlying real appreciations. In its 2002 review, the EC stated that “Convergence of productivity and living standards … would create a tendency toward price level convergence, giving rise to differentials in inflation, with low price level countries tending to experience somewhat faster rates of price increase” (EC, 2002a). The ECB has stated “To the extent that convergence of prices is a natural consequence of the integration of markets, the resulting inflation differentials need not be viewed as posing problems for economic policy” (ECB, October 1999). The question arising from these observations is whether countries should delay adopting the euro until real convergence reduces inflation differentials attributable to structural and transition factors or whether such differentials are compatible with full integration in the monetary union.

However the inflation criterion is interpreted, the CECs will need concerted efforts either to reduce inflation from present rates that exceed a reasonable estimation of B-S effects or to hold inflation at recent low levels. Even apart from the Maastricht criterion, countries will need to have inflation rates that ensure that competitiveness will remain adequate once exchange rates are irrevocably fixed. For Hungary, the Slovak Republic, and Slovenia, this means lowering inflation by some 1.2 to 5 percentage points relative to 2003 levels. In the Czech Republic and Poland, where current inflation is well below this notional target or any conceivable interpretation of the Maastricht criterion, significant output gaps—according to staff estimates, as large as 1.8 and 3.2 percent of potential GDP respectively—are playing a role. As recoveries proceed, strong efforts will be needed to prevent an inflation rebound.

The experience of the noncore euro-area members several years before euro adoption provides some guidance for the CECs. The countries with high pre-EMU inflation benefited from three environment changes—fiscal adjustments discussed in the previous section, wage moderation, and exchange rate stabilization. In the EU as a whole, nominal unit labor costs rose by 4.8 percent a year during 1990–93, but by only 1 percent during 1994–97. Moderation in wage increases, rather than higher productivity growth, was the primary cause: average increases in nominal compensation per employee slowed from 6.4 percent to 3.3 percent between the two periods. This moderation can be explained by several factors: inflation itself had fallen; inflation expectations became increasingly consistent with announced inflation objectives (EC, 1998); and wage-setting procedures were changed—automatic wage indexation was eliminated in Italy, and multi-year wage agreements were introduced in Spain, Ireland, Finland, and Greece. At the same time, the sharp depreciations following the ERM crisis subsided, and exchange rates stabilized. Not only did this have a direct effect on inflation through imports, but as the scope for using the exchange rate to boost competitiveness narrowed, wages became the key determinant of competitiveness: wage negotiations therefore increasingly took into account productivity, profitability, and labor market conditions (Figure 7.7).

Figure 7.7.Noncore Euro-Area Countries: Inflation Indicators Before Euro Adoption

(Period average change, in percent)

Source: OECD.

1 Definition of the reference value for the inflation criterion is discussed in Section VI.

Under last-minute pressure to meet the Maastricht criteria, some countries resorted to administrative measures, although these played a relatively small role compared with that of more substantive policies. Greece, for example, cut indirect taxes during October 1998–December 1999—0.9 percentage points of the cumulative 2.5 percentage point reduction in inflation between 1999 and 2000 is attributed to these measures (ECB, 2000). These cuts also helped reduce the inflation rate used for the calculation of a compensation clause in the 1998–99 wage agreement. Beyond this, Greece relied on informal agreements with commercial and industrial enterprises and service providers to reduce the retail prices of items accounting for up to one-third of the CPI basket. Moreover, in 1998, in response to strong capital inflows that threatened to derail inflation targets, Greece introduced temporary credit ceilings.

Among the CECs, the mix of measures to control inflation will differ considerably from country to country. In contrast to the noncore euro-area countries, none of the CECs except Slovenia has much scope for benefiting from stabilization of the exchange rate (Figure 7.8): although exchange rates have been volatile, they have on the whole followed broadly strengthening trends over the past few years. Slovenia, however, which has seen a steady depreciation of the tolar in a crawling-peg framework, could realize substantial gains for inflation as the value of the tolar is stabilized. For the four high-deficit countries, fiscal consolidation in support of a firm monetary policy will be crucial. Wage restraint will also be important. For some countries—notably Poland and the Slovak Republic—high unemployment rates will be a strong moderating influence on wages. Others may need to rely on structured incomes policies. In Slovenia, where fiscal restraint will play a small role, eliminating the vestiges of a pervasive system of indexation and shifting monetary policy from a largely accommodating role to more active resistance to inflation will be essential.

Figure 7.8.CECs and Noncore Euro-Area Countries: Real and Nominal Bilateral Exchange Rates Prior to Euro Adoption1

Source: IMF, International Financial Statistics.

1 Against the deutsche mark; t is the year before euro adoption. For Greece, 2000; for the others, 1998.

2 CPI-based real exchange rate.

Choosing Parities

Together with strong and consistent macroeconomic policies, the central parity is the glue in the overall framework for meeting the Maastricht criteria. If chosen properly to reflect underlying influences on the real equilibrium value of the currency, it will be a credible terminal conversion rate and will help preempt speculative tests on the market rate. It also will be central to the strategy for reaching or maintaining a low inflation rate. And to the extent that it becomes the terminal conversion rate in an environment of some stickiness in wages and prices, it can have an effect on competitiveness and growth well beyond the ERM2 period. As such, central parities will be contentious to set—involving issues of competitiveness not only against the euro area but also among the CECs.

Yet important as the central parity is, it will not be a panacea for all potential difficulties in ERM2. If macroeconomic policies are inconsistent with the basic goals of the Maastricht criteria, problems will result no matter how carefully and strategically the parity is chosen. This section addresses the key considerations in choosing parities, and the role of this choice in supporting otherwise sound policies.

The discussion of methods to identify equilibrium exchange rates in the CECs in Section V rendered several conclusions: (1) As of end-2002, no signs of flagrant over- or undervaluation were evident; but (2) measures of real exchange rates have varied quite widely and at any one time even suggest a rather wide range of competitive positions among the CECs; and (3) in general, the indicators provide less guidance on the absolute size of any under- or over-valuation than on movements toward or away from equilibria. These points imply that estimates of equilibrium exchange rates will be rather imprecise, providing guidance at best on a range of rates that would be manageable.

Strategies for Parity Setting

With objective indicators providing no more than a range of parities that would be consistent with underlying equilibrium real exchange rates, other considerations will also have to play a role. These should take into account current market pressures on the rate, the strategy with respect to inflation during ERM2, and recent developments in growth, investment, and export market penetration. The market rate should provide guidance, although the very notion of considering measures of the equilibrium reflects the fact that the market is subject to transitory influences that should not be built into the parity. The operational question is only where within a range of equilibrium estimates the parity might best be targeted.

A key consideration is that risks in ERM2 are asymmetric. Assuming a relatively narrow definition of the exchange rate stability criterion, pressures pushing a currency below parity will require a strong defense through interest rate hikes and possibly even intervention. As the scope for these is constrained by the country’s tolerance for high interest rates and by the availability of foreign exchange reserves, downward pressure could snowball into a situation that requires a devaluation. In contrast, resistance to upward pressure—interest rate cuts and sterilized purchases of foreign exchange—is subject to less binding constraints (the risk of overheating and the cost of sterilization). Also underpinning these considerations is the likely asymmetry in the exchange rate stability criterion: no downward adjustment of the parity may be made over the two-year assessment period, but revaluations are not ruled out. This asymmetry means that setting a parity too high is riskier than setting it too low.

The choice of the parity will also influence the strategy for keeping inflation low or reducing it, although the implications are not clear-cut. On the one hand, if a parity below current market rates is seen as a credible terminal rate, it would provide scope for holding interest rates above the euro-area level, allowing a measure of monetary policy independence without pushing the market exchange rate to uncompetitive levels. The viability of such a strategy is borne out by the experience of Greece: establishing the parity substantially below the prevailing market rate permitted interest rates to be held above euro-area levels for much of the ERM/ERM2 period (Box 7.4). On the other hand, two differences between the environment that Greece faced and that which the CECs are likely to face may put the Greek strategy out of reach. First, the CECs have the precedent of two countries (Greece and Ireland) revaluing shortly before final conversion. Markets may now anticipate a late revaluation, reducing a country’s ability to hold interest rates above euro-area levels without large capital inflows. Second, the recent history of real appreciations in the CECs, together with the large benefits for growth expected from euro adoption, may militate against the credibility of low parities as terminal conversion rates. Thus, a low parity may well be realizable as a terminal conversion rate only with substantial increases in prices and wages to bring the real value of the currency as it approaches the terminal rate into line with market views of a sustainable equilibrium. In these circumstances, the convergence to the low parity would actually exacerbate wage and price inflation. On balance, the CECs that enter ERM2 with remaining disinflation objectives will probably need to err on the side of slightly stronger parities within their equilibrium ranges than those that enter with inflation better under control.

A third consideration for setting CEC parities is the presumption that the equilibrium exchange rate is rising over time in line with B-S effects of 1–2 percent a year. This suggests that the parity should be set above the current market rate by a margin sufficient to absorb at least the lower end of this equilibrium appreciation. Conceptually, this proposal has much appeal. But from a practical point of view, taking into account the asymmetric risks of ERM2 and high degree of uncertainty about the equilibrium exchange rate either at the time of setting the parity or two to three years later, this would not seem prudent. The scope for revaluation of the parity before final conversion should rather be used to accommodate any appreciation in the equilibrium rate subsequent to the setting of the parity. Moreover, arbitrage conditions mean that setting a credible terminal parity rate above the market rate would require that domestic short-term interest rates be below euro-area levels, undermining any semblance of monetary restraint.

Box 7.4.Greece: Setting and Adjusting the ERM Central Parity

In March 1998, Greece agreed with the European institutions to enter ERM. Preceding the decision, the drachma had been under severe downward pressure, fueled by worsening competitiveness and current account developments and contagion from the Asian crisis. The effects of the pressure on the drachma had been fenced off by high interest rates for some four to five months (see figure).

Greece: Exchange Rate and Interest Rate

Source: Bloomberg Financial LP.

The parity for the drachma was set 12.3 percent below the prevailing market rate. In support of the parity, the Greek authorities had agreed with the European institutions on a package of fiscal and structural measures. The parity was viewed as reestablishing adequate competitiveness. The market rate moved immediately to about 3 percent above parity, and short-term interest rates fell sharply. Contagion from the Russia crisis in August 1998 put pressure on Greek financial markets—triggering extensive currency intervention. Subsequently, however, renewed market confidence pushed the market exchange rate to some 9 percent above parity. From there it rather smoothly converged toward the parity during the subsequent two and half years. The gap between the market rate and the central parity was matched by the differential between Greek and euro-area interest rates.

Within ERM/ERM2, reducing inflation was the key objective of monetary policy, and the scope for holding interest rates substantially above those in the euro area during much of the ERM/ERM2 period was central to meeting this objective.

As interest rates converged to euro-area levels in the run-up to EMU entry, administrative measures and a revaluation were used to ensure that the inflation target was met. During 1999, the scope for interest rate policy to curb credit growth was undercut by the increasingly credible terminal point of interest rate convergence. In this setting, temporary credit controls were imposed. In addition, with inflation having picked up again in late 1999, the revaluation of the central parity in January 2000 by 3½ percent facilitated further disinflation.

Getting to the Right Parity

Ideally, parities will be set at or close to existing market rates. Although consideration of parities different from existing market rates should not be ruled out, potential conflicts with the exchange rate stability criterion are likely to be minimized when the market rate starts out close to the parity. The best approach to ERM2, therefore, is to establish a sound policy mix aiming for the prudent fiscal balance and monetary conditions that are as easy as possible while being compatible with an appropriate inflation target. This is likely to push the market rate toward the lower end of the equilibrium range, consistent with a viable parity.

But market conditions in the run-up to ERM2 entry may not be so tranquil. Added to the normal volatility of CECs’ financial markets will be the potential for strong speculative pressures as the market takes positions on expectations of where the central parity will be set. In these circumstances, countries will need to form a view prior to ERM2 entry on the appropriate parity and, within their scope for maneuver, attempt to guide the market rate toward that level. For the four inflation targeters, this means placing a significant weight—for the more pure inflation targeters, a higher weight than now maintained—on the nominal exchange rate in interest rate decisions. Obviously, this would have to be balanced against the inflation objective—a consideration that argues strongly for countries to wait to enter ERM2 until their inflation rates are close to a level acceptable under the Maastricht criterion. With clear signals through interest rate policy on a country’s preferred parity, market forces can be harnessed to help guide the rate toward that level.

Monetary Policy Frameworks

In ERM2, monetary policy will face multiple objectives: controlling inflation, keeping the exchange rate competitive, and meeting the exchange rate stability criterion. Moreover, the choice of a monetary policy framework will not be unconstrained. ERM2 requires that a central parity against the euro be maintained, while unilateral “euroization,” free floats, pegs to currencies other than the euro, and crawling pegs are ruled out.78 Thus each of the CECs—except Hungary, which already maintains an ERM2-type arrangement—must change its present framework at least to the extent of operating with an announced parity.

This section considers the main options for monetary frameworks during ERM2. It starts with general considerations that should guide decisions on where countries should aim on the spectrum of flexible to fixed exchange rate frameworks. Then, various possibilities are assessed with respect to transparency; the stringency of demands on other macroeconomic policies; and the likelihood that the framework, given disciplined fiscal policy, could guide inflation and the exchange rate to meet the Maastricht criteria without excessive real appreciation.

Choices for Exchange Rate Regimes—Optimality of the “Corner Solutions”

Over the past decade, exchange rate crises in emerging market and advanced countries point to the vulnerabilities implicit in soft peg arrangements when capital accounts are open. Underpinning this view is the impossible trinity—that fixed exchange rates, capital mobility, and a monetary policy dedicated to domestic goals cannot coexist for long without generating disequilibria that risk fomenting a speculative attack. In the absence of impediments to capital flows, undoing this “trilemma” implies forgoing either monetary independence (adopting a hard peg regime) or exchange rate stability (adopting a floating rate). In either of these so-called “corner solutions,” markets are given unambiguous signals about how exchange rates will respond to exogenous or policy-induced shocks and about the risks inherent in taking open foreign exchange positions.

Each “corner solution” has its own advantages. A hard peg provides a nominal anchor, eliminates exchange rate volatility, and avoids speculative exchange rate bubbles. A floating rate provides monetary policy independence, can be an automatic shock absorber, and discourages unhedged currency exposure. Moving away from these polar regimes toward the middle ground would seem to imply trading off some of these benefits for those of the other extreme.79 Such intermediate solutions, however, typically do not offer a safe harbor.

The key weakness of fixed-but-adjustable pegs is the perception that defending the peg is not the highest policy priority of a government. For example, the authorities may be reluctant to wage an all-out defense against downward exchange rate pressures because of the substantial damage that high interest rates could do to the banking system, the budget, and the real economy. Without full credibility that such side effects would be ignored indefinitely, a fixed exchange rate is highly vulnerable to a speculative attack (Obstfeld and Rogoff, 1995). Of course, similar concerns about the effect of exchange rate changes on the budget, bank balance sheets, real wages, and inflation can make governments reluctant to allow exchange rates to change. But markets are quick to realize when fundamental imbalances will not be resolved without an exchange rate adjustment, and persistent market pressure is likely to make resolution through an exchange rate adjustment inevitable.

Recent empirical evidence supports the view that intermediate exchange rate regimes (soft pegs and tightly managed floats) are more prone to currency crises than currency boards or free floats. Defining currency crises as severe exchange market pressures associated with sharp movements in either exchange rates or interest rates, Bubula and Otker-Robe (2003) find that during 1990–2001 the frequency of crisis was substantially higher under intermediate regimes than under polar regimes. In fact, the incidence of crises under intermediate regimes was about three times higher than under hard pegs across all countries, and close to five times higher for developed and emerging market countries. Intermediate regimes were also more susceptible to crisis than floating regimes across all countries. For developed and emerging market countries, the incidence of crises appears to be about the same in hard peg and floating regimes. Equally, no significant differences in the incidence of crises in different types of intermediate regimes is found.

Fashioning monetary policy frameworks in ERM2 that are consistent with these lessons from the 1990s will be challenging. Free floating will not be an option, and only countries with exemplary policies—policies, in fact, that would from the start qualify a country for euro-area membership—should consider a hard peg to the euro. The challenge, therefore, will be to work within the constraints of ERM2 and the exchange rate stability criterion while ensuring that a transparent framework that encourages appropriate hedging and is credible is put in place. The remainder of this section considers several such frameworks.

Inflation Targeting with Wide Bands

In recent years, the Czech Republic, Hungary, and Poland have operated direct inflation targeting. Success in meeting inflation has been mixed: while the Czech Republic and Poland have reduced inflation to low levels, Hungary, which has operated simultaneously an exchange rate band, has been less successful in doing so. At the same time, exchange rates have varied largely within ±15 percent of the average exchange rate for 2000–03 (Figure 7.9) with only moderately supportive fiscal, wage, and structural policies, and, except in Hungary, without a central parity. This experience suggests that with some modification—in particular setting a central parity against the euro—a variant of inflation targeting would be feasible in ERM2.

Figure 7.9.CECs: Deviation from the Average Exchange Rate

(In percent)

Sources: Bloomberg Financial LP; and country authorities.

1 Deviation from the average of the euro per Polish zloty from January 2001 to April 2004.

2 Deviation from the average of the euro per Slovenian tolar from January 2001 to April 2004.

3 Central parity set by the National Bank of Hungary.

4 Deviation from the average of the euro per Slovakian koruna from January 2001 to April 2004.

5 Deviation from the average of the euro per Czech koruna from January 2001 to April 2004.

6 Average exchange rate from May 4, 2002 to May 5, 2004.

The experiences of Finland and Spain, which formally followed inflation targeting in ERM, were broadly successful, but point to the importance of supportive policies. Both countries adopted inflation targeting several years before their assessment periods—Finland in 1993 and Spain in 1994.80 In the run-up to euro adoption, conditions became quite stable, and exchange rate fluctuations were contained within a narrow (±2¼) band for the two years prior to euro adoption (Figure 7.10). The peseta benefited from sizable short-term interest rate differentials vis-à-vis most partner countries. In both countries, accompanying policies were supportive: fiscal deficits were reduced steadily during the inflation targeting period, multi-year wage agreements resulted in wage moderation, and a sizable devaluation within one year of the introduction of inflation targeting in Spain, and immediately prior to its introduction in Finland, left starting exchange rates at levels that were unlikely to be subject to downward pressures. Spain and Finland intervened in both directions, heavily at times, to fend off pressures on the exchange rate during ERM participation, and Spain even during the last two years of ERM (Figure 7.11).81 These pressures largely abated, however, once the conversion rate was announced. These experiences suggest that the monetary frameworks of these inflation targeters may not have been substantially different from those of other ERM participants in that active use of interest rates and intervention helped maintain exchange rate stability.

Figure 7.10.Finland and Spain: Interest Rate Differential and General Government Balance

(Interest rate differential against the comparable euro rate; in basis points)

Sources: Eurostat; IMF, International Financial Statistics; and IMF staff estimates.

1 Three-month money market rate.

2 Ten-year government bond yield.

Figure 7.11.Finland and Spain: Exchange Rates and Change in Reserves

Sources: IMF, RETS database; IMF, International Financial Statistics; and IMF staff calculations.

1 Horizontal black lines correspond to central parity, and horizontal dotted lines correspond with ERM bands. On August 2, 1993 the ERM bands were widened from ± 2.25% to ± 15%.

Inflation targeting in ERM2 would need to be a hybrid of inflation targeting and exchange rate targeting. Within the ERM2 band, monetary policy would target inflation; as and if the exchange rate approached the limits of the band, however, primacy would shift to keeping the exchange rate within the band. Countries would need to be able and willing to defend the band through interest rate changes and possibly foreign exchange market intervention. Even with perfect clarity on this approach, however, inflation targeting with wide bands would, in certain situations, leave markets to guess on policy responses. For example, if the exchange rate approached the strong (appreciated) end of the band, markets would become sensitive to the question of whether policy interest rates would be cut or the parity would be realigned, particularly if the inflation outlook were inconsistent with interest rate reductions. On the weak end of the band, where conflicts between the inflation and exchange rate objectives would be less likely, markets could still question the scope for intervention and the tolerance for high interest rates to support the currency (Box 7.5 and Figure 7.12).

Figure 7.12.Reserves in the Euro Area, CECs, and Baltic Countries1

Source: IMF, World Economic Outlook.

1 Euro-area countries include only Italy, Ireland, Portugal, Spain, and Sweden. The CECs include the Czech Republic, Hungary, Poland, Slovenia, and the Slovak Republic. The Baltics include Latvia, Lithuania, and Estonia. Data for euro-area countries are for 1996–98. Data for CECs and Baltics are for 2000–03.

A hybrid inflation targeting framework supported by strong fiscal and incomes policies should be compatible with the Maastricht exchange rate and inflation criteria, but risks of excessive nominal and real appreciations must be recognized. Specifically, nominal appreciations—whether as part of a deliberate policy to reduce inflation or due to strong capital inflows—could render the parity unrealistic as the final conversion rate: a sizable revaluation of the parity in the run-up to conversion would then become unavoidable. The experience of Poland and the Czech Republic points to the close relationship between changes in nominal and real exchange rates in an inflation targeting framework (Figure 7.13). Thus, while the framework would perhaps hold the greatest appeal to countries with an unfinished disinflation agenda, it is precisely in these circumstances that the framework risks producing an overly strong conversion rate. Attempting to keep the exchange rate very close to parity even on the upside would likely lead to inconsistencies between the exchange rate and inflation targets.

Figure 7.13.Czech Republic and Poland: Real and Nominal Effective Exchange Rates1

(January 1998 = 100)

Source: IMF, Information Notice System (INS).

1 An increase implies an appreciation.

Hungary’s experience with combining inflation targeting and a wide exchange rate band illustrates the tensions that may arise from inconsistent policies and targets. In mid-2001, after disinflation had stalled under a narrow exchange rate band regime, the Magyar Nemzeti Bank (MNB) with the agreement of the government, widened its exchange rate band to ±15 percent and adopted inflation targeting. All foreign exchange restrictions were lifted, and Parliament approved the new Central Bank Act defining the primary objective of the MNB to be achieving and maintaining price stability. The MNB initially set an objective of reducing inflation to EMU-compatible levels by 2004–05. The first year of inflation targeting saw a resumption of disinflation and only modest upward pressure on the exchange rate. But by late-2002, the exchange rate had risen to the top of the band. At that time, considerable stimulus from an expanding fiscal deficit and a large public sector wage increase fueled market expectations that the inflation target would be priortized over defense of the band. This resulted in a currency attack in January 2003 and a sharp subsequent increase in exchange rate volatility.82

Box 7.5.Foreign Exchange Intervention: Desirability, Effectiveness, and Feasibility

A meaningful commitment to exchange rate stability during ERM2 may entail a role for intervention. A critical question is whether the CECs themselves are in a position to carry out such intervention successfully in the face of even modest exchange market pressures. During ERM, 85 percent of intervention was done intramarginally: that ECB intervention is automatic only at the wide ERM bands suggests that most intervention was done by the countries themselves. Intervention to fend off appreciation would have to be either quickly reversed or sterilized to avoid potential conflict with the inflation criterion.

The effectiveness of sterilized intervention to fend off volatility from routine noise or more serious contagion is widely viewed as limited. Two main streams in the literature support this conclusion. First, direct empirical evidence on effectiveness of sterilized intervention is mixed. Literature from the 1970s and 1980s tended to conclude that sterilized intervention was effective, at best, in the short term (Edison, 1993). More recent empirical work on emerging markets finds that even when intervention is effective in targeting the exchange rate and reducing volatility, the impact varies with the time horizon but is generally short-lived (Domac and Mendoza, 2003; and Guimaraes and Karacadag, 2004). Only studies based on industrial economies suggest a significant effect both on the level and the change of exchange rates (Sarno and Taylor, 2001). Second, theory suggests that with no impediments to capital movements, there is little scope for sterilized intervention when exchange rates are pegged or subject to a strict rule. However, literature measuring the degree of capital mobility (the “offset coefficient” approach pioneered by Argy and Kouri, 1974, and Kouri and Porter, 1974) does not provide a definitive conclusion on the effectiveness of sterilized intervention (Kreinin and Officer, 1978; Horne, 1983; and Laskar, 1983, provide surveys). Estimates for emerging markets suggest some scope for sterilized intervention, at least over one quarter (Schadler and others, 1993).

The ability to defend the exchange rate from depreciating pressures depends on the size of the foreign exchange reserves. Measures of foreign exchange reserves relative to the sources of domestic pressure (ratios of reserves to broad money and of reserves to imports) or the size of the economy (reserves relative to GDP) suggest that the power of the CECs to defend against depreciation pressures is comparable to or higher than that of the intervening ERM countries and the Baltics (see Figure 7.12). However, the CECs’ reserves are small compared with the size of world financial markets, as well as with the intervention by some ERM member countries during the 1992–93 ERM crisis. For example, during that episode the Bank of France lost Fr 80 billion, Spain’s reserves fell by $5–20 billion, and the Italian lira left the ERM despite intervention of DM 64 billion (Buiter, Corsetti, and Pesenti, 1998a).

Exchange Rate Targeting

Exchange rate targeting is an option perhaps more in keeping with the spirit of ERM2; implemented with a sufficiently long target horizon, it could also provide adequate protection against speculative attacks. In an exchange rate targeting framework, interest rate and fiscal policies would aim to stabilize the exchange rate over, say, a two-year time horizon at a central parity, although no implicit or explicit bands interior to the ±15 percent ERM2 bands would be set. Thus, any market pressure might initially be allowed to move the exchange rate—even quite significantly within the ±15 percent band—but subsequent fiscal and monetary policy adjustments would aim to guide it gradually back toward parity. Direct intervention in the foreign exchange market would at most be minimal in order to avoid any signals of implicit bands. Inflation would be a key secondary objective, but increasingly subordinated to the exchange rate target if and as the market rate moved away from parity.

The objective of this arrangement would be to maximize the magnet effect of a well-chosen central parity. Such an effect should be strong when the market sees policies as fully consistent with the parity and expects an early conversion date. Ex ante commitment to flexibility in the exchange rate on both sides of parity would be critical to establishing the credibility of potential exchange rate variations—a key component of communications to markets on the scope for exchange rate fluctuations and the risks of taking unhedged positions. Paradoxically, barring exogenous disturbances, these conditions could produce exchange rate behavior in line with the narrow interpretation of exchange rate stability.

Properly implemented, this arrangement could combine a high degree of transparency of the monetary policy objective with a strong likelihood that the Maastricht exchange rate and inflation criteria would be met. Obviously, its success would be contingent on adequately supportive fiscal, incomes, and other structural policies. As in an inflation targeting frame-work, conflicts between inflation and exchange rate stability objectives could arise. Exchange rate targeting, however, would consistently prioritize the objective of moving the exchange rate toward parity, albeit not in a short-term time frame.

Narrow Bands

Countries might opt for an explicit narrow band (for example, ±2¼ percent) around parity for several reasons. First, ERM participants may be uncertain about how the EC and ECB will interpret exchange rate stability and, therefore, wish to aim for an outcome that will almost certainly be accepted as consistent with stability. Second, a country may expect that it could enhance the credibility of its parity and thereby minimize upward pressure on its final conversion rate by constraining the market rate within the narrow band. Third, countries may be highly confident about the strength of supporting policies and choose a narrow band arrangement because it is administratively easier, but substantively similar to a hard peg.

Whatever the case for a narrow band arrangement, risks are substantial. It is true that consistent supporting policies aimed at reducing inflation and relieving pressure on domestic interest rates, together with a credible parity, should help secure greater exchange rate stability than the CECs have experienced with inflation targeting in recent years. In fact, theory suggests that if the peg is fully credible, the exchange rate would stabilize even without active intervention (Krugman, 1991a).83 But to assume that the exogenous influences that have contributed to exchange rate volatility in recent years will be tamed in the future has less basis. Moreover, in the middle ground between broad exchange rate flexibility and a hard peg, when markets are left room to guess how policies will respond to stress on the exchange rate, risk premia are subject to considerable volatility, which in turn can propagate surges in inward or outward capital flows. Indeed, some argue that even with the confidence-instilling effects of a central parity, adopting a narrow band in an emerging market subject to several sources of capital account volatility is to invite speculators to test the central bank’s determination and ability to enforce the margins (Begg and others, 2003). In effect, the credibility and transparency of the framework under stress would have the well-known problems of soft pegs.

The potential for inconsistency between the exchange rate and the inflation criterion would be significant in a narrow band arrangement and could jeopardize meeting the Maastricht inflation criterion. Any gain in the credibility of the parity relative to that in a wider-margin arrangement would come at the cost of monetary policy independence. By reducing the scope for a gap between the central parity and market exchange rate, narrow bands would all but eliminate the possibility of establishing domestic interest rates above those in the euro area. Thus, a persistent challenge to the strong edge of the band that required a defense beyond the power of sterilized intervention could prove incompatible with the interest rate policy needed to reach the inflation target. On the weak side, of course, the resources available for intervention and the tolerance for high interest rates could well prove inadequate to a credible defense of the lower band.

The experience in ERM points to the demanding conditions that must be met to render implicit or explicit narrow margins successful. Austria, Belgium, and the Netherlands weathered long periods—for the last two, as many as 20 years—in ERM.84 During the two-year reference period used to assess compliance with the Maastricht criteria (March 1, 1996–February 27, 1998), these countries’ currencies remained close to their unchanged parities without a need for exceptional measures of support (Figure 7.14).85 While monetary policy aimed at keeping a close link between the home currency and the deutsche mark, only the Netherlands and Germany had a formal agreement—limiting fluctuation between the guilder and the deutsche mark to ±2.25 percent of parity. Fiscal consolidation along with complete subordination of domestic to German monetary policy were critical in achieving exchange rate stability. These countries were undoubtedly also helped by the high degree of synchronization of business cycles, trade integration, and the symmetry of shocks.

Figure 7.14.Selected Core Euro-Area Countries: Macroeconomic Developments Prior to Euro Adoption

Sources: Eurostat; IMF, International Financial Statistics; and IMF staff estimates.

1 Differential vis-à-vis the German short-term rate. Money market rate for Austria and call money rates for Germany, Belgium, and the Netherlands.

2 General government.

Hard Pegs

As an alternative to narrow bands, a hard peg—that is, a rigid commitment to a single exchange rate coupled with fully endogenous determination of interest rates and monetary aggregates—will warrant serious consideration. Its attraction will stem from several sources. First, successfully implemented, it would ensure compliance with the exchange rate criterion. This would be an especially attractive feature for countries that are concerned about how the EC and ECB will interpret the exchange rate stability criterion. Second, transparency vis-à-vis the market would be maximized: properly designed, it would provide the market with a clear statement of how the monetary authorities would react to any pressures, leaving virtually nothing for the market to second-guess. This transparency would be especially attractive to countries that fear upward speculative pressure on their currency. Third, this enhanced transparency would come at a negligible cost in terms of the rigor of policies required and loss of monetary autonomy as compared to a narrow band arrangement. Fourth, a decisive break with the past may provide a boost to the political process of securing public support for strong fiscal and wage policies. Of course, a hard peg also reduces the risk of a loss of competitiveness—transitory or long-lasting—that could result from shocks or a sustained interest rate differential in a wide band arrangement.

While transparency would be maximized under a hard peg, policy demands would be rigorous. To ensure viability of a hard peg, it would be essential that fiscal policy and inflation had been brought under control—preferably to levels consistent with a strong entry into the euro area, but at least to levels that would ensure meeting the Maastricht criteria within two years. Thus realization of the benefits of euro adoption—elimination of exchange rate risk premia, lower interest rates, and absence of vulnerability to exchange rate changes—would accrue almost fully with the adoption of the peg.

The long experience of the Baltic countries with currency boards points to the feasibility of a hard peg even in the face of internal structural transformation and an asymmetric shock as large as the Russian default in 1998. These small open economies have maintained currency boards (Estonia since 1992, Lithuania since 1994) or hard pegs (Latvia since 1994) for more than a decade. With initial conditions in terms of fiscal positions and structural characteristics little different from those in several of the CECs, the discipline of their monetary arrangements is often held up as an important factor behind the reining in of budget deficits, progress with privatization, and a high degree of wage and price flexibility (Gulde, Kähkönen, and Keller, 2000). The response of interest rates and monetary aggregates to the Russia shock was sharp but contained to a period of about one year (Figure 7.15). These countries experienced downturns more severe than those in the CECs, yet subsequent rebounds were far stronger due to their flexible economic structures and macroeconomic policies (Box 7.6).

Figure 7.15.Baltic Countries: Macroeconomic Developments

Sources: Eurostat; IMF, International Financial Statistics; IMF, World Economic Outlook; and IMF staff estimates.

1 Spread against the German short-term rate. Call money rate for Germany and money market rates for Estonia, Latvia, and Lithuania.

2 General government.

In sum, the types of monetary policy frameworks available to the CECs during ERM2 entail significantly different mixes of transparency, rigor of accompanying policies, and suitability for delivering compliance with the Maastricht criteria for the exchange rate and inflation. While all will require strong and consistent fiscal policy and rigorous wage discipline, the more constraining is the exchange rate arrangement, the greater will be the demands on other policies. In contrast, the transparency of the framework and its response to stress on the exchange rate—a critical determinant of vulnerability to speculative attacks—is greatest in a hard peg and weakest in the case of narrow margin arrangements. Embedded in a scheme whereby the Maastricht criteria on all but the exchange rate and interest rate are met before ERM2, a hard peg would deliver the Maastricht criteria with the greatest certainty, although with such supportive policies the more flexible arrangements should prove equally successful.

Appendix 7.1. Estimation of the Impact on Growth of Fiscal and Monetary Policy

GDP growth is estimated using a panel of annual time-series data (1980–2001) for 12 EMU countries—modeled as a function of the fiscal stance, monetary policy (short-term interest rate), the real exchange rate, the growth of partner countries’ imports and the long-term bond yield. The model allows for lagged effects of policies and exogenous factors, as well as the inclusion of a lagged dependent variable as explanatory variable:

Δyi,t = α + δΔyi,t–1 + X′itβ + ui,t,

with ui = μi,t + vi,t, i = 1, … 11, t = 1, … 21,

with y denoting GDP and X the vector of explanatory variables (including lags): the change in the cyclically adjusted primary general government deficit; the short-term policy interest rate, deflated with (backward) CPI inflation;86 the relative change in the real effective exchange rate (CPI-based); weighted growth of imports in the country’s trading partners; and the real interest rate on benchmark government bonds.

The following features are noted about the explanatory variables. First, long-term interest rates are included to capture credibility effects, including from potential “Maastricht effects” (discussed in Box 7.2). In reality, the long-term interest rate is endogenous, determined by short-term interest rates and forward-looking perceptions of macroeconomic policies and inflation. In the simple model here, where the forward-looking perceptions are not captured by any of the other explanatory variables, they are meant to be captured by the long-term interest rate.87 Second, weighted partner-country import demand is included as the most accurate proxy for external demand conditions. Third, real short-term interest rates and the real exchange rate are included as two separate explanatory variables, rather than combined into one monetary policy index.

Turning to econometric issues, there are assumed to be country-specific effects (the μi), which might be correlated with the regressors. Such correlation requires the use of instruments and/or transformations of the equations.88

The regressors (or instruments) need to be strictly exogenous (with respect to vit, the disturbance term) for the estimators to be consistent and unbiased. However, the inclusion of the lagged dependent variable on the right-hand side, in combination with the individual effects, implies correlation of an explanatory variable with the disturbance term. This makes both the ordinary least squares (OLS) and fixed-effect (“within transformation”) estimators biased and inconsistent. The generalized matrix of moments (GMM) procedure applied here deals with this correlation by exploiting additional instruments stemming from the orthogonality conditions that exist between lagged dependent variables and the disturbances (Arellano and Bond, 1991).

Another source of potential endogeneity could be partner-country import growth. With European countries trading intensely with each other, partner-country import growth could be correlated with GDP growth. It is difficult to construct a good instrument that is correlated with partner-country import growth but not with ui,t. In the absence of a better instrument, the average OECD output gap and its lags are used as instruments. Another way to deal with the potential endogeneity of partner-country import growth is to remove this variable altogether while including a time-specific effect in the panel setup.89 That such a regression produces quite similar results with regard to the main coefficients indicates that the potential endogeneity of partner-country growth does not bias the estimations significantly.

Box 7.6.The Success of Hard Pegs in the Baltics

The decade-long experiences with currency boards in Estonia, Latvia, and Lithuania suggest that hard pegs can be suitable for economies undergoing structural change and can be sustained even through severe crises. Hard pegs in the Baltics provided a tool for gaining monetary policy credibility and for achieving macroeconomic stabilization. The introduction of accompanying institutional and structural reforms helped create flexible market economies.

The fixed exchange regimes were not an obstacle to growth and low inflation. The Baltics enjoyed rates of growth averaging over 5 percent during 1995–2003. This is broadly consistent with the findings of Ghosh, Gulde, and Wolfe (1998) that countries with currency boards had higher output growth than countries with other exchange rate regimes. Such solid growth could reflect rebound effects (since many currency boards are set up after a crisis), or self-selection (governments willing to accept the rigor of a currency board are likely to be reformist) (Gulde, Kähkönen, and Kelles, 2000). Alongside solid growth, each of the Baltics successfully disinflated (achieving inflation rates below 3 percent by end-2002), while interest rates spreads against euro-area financial instruments also converged sharply as risks premia diminished.

The Russia shock in 1998 constituted a severe test for the hard peg regimes. The adverse direct and region-wide effects of the crisis triggered a sharp contraction in activity in the Baltics during 1998–1999. While exports to EU countries continued to grow, exports to Russia plunged. The pegs, however, were sustained, and the cyclical downturns—which were more severe than in the CECs during this episode—turned out to be short-lived.

Throughout the hard peg regimes, fiscal discipline has been strong. In 1997—just prior to the Russia crisis—Estonia and Latvia had general government surpluses, while Lithuania had a deficit of less than 2 percent of GDP. Although the Russia crisis caused sharp increases in deficits in 1998–99, prompt correction occurred as output returned to potential levels. Indeed, a remarkable trimming of primary expenditures relative to GDP—by 5½ percentage points during 1999–2002—allowed a sizable reduction in the tax burden (Kopits and Székely, 2004).

Despite large and persistent current account deficits, the overall balance of payments has been in surplus (with minor exceptions) since the Russia crisis. As a result, reserves have accumulated to back the creation of domestic currency. The favorable investment environment helped attract FDI, which covered on average 62 percent of the current account deficit during 1995–2003.

Various statistical tests are applied. First, a Hausman specification test is used to verify whether in the reported equations the set of regressors/instruments is strictly exogenous. Second, another Hausman specification test is used to test whether the set of individual effects are uncorrelated with the set of regressors/instruments (in which case OLS and fixed-effect estimators would be unbiased and consistent). Third, tests for the absence of serial correlation in the error terms—essential for consistent estimators and achieved by instrumenting the lagged dependent variable with further lags of the same variable—are based on AR(1) and AR(2) tests.90 Finally, the reported standard errors are robust to heteroscedasticity (across countries and time).

Results

Regression I

Table 7.8 presents the results of five panel regressions. Regression I suggests that the fiscal impulse has a significant but modest impact on growth, with a multiplier of around 0.2. Growth in partner countries’ imports is highly significant, with a coefficient of 0.26, underlying the importance of this factor. Given the high variance of this variable—its standard deviation is 4, compared with 1.4 for the fiscal impulse—its contribution to explaining variation in GDP growth rates is larger than that of fiscal policy. The real exchange rate is found to have a significant impact. Its coefficient would indicate that a 1 percent appreciation leads to a reduction in GDP growth of 0.04 percentage points. The real short-term interest rate has negative effects, but they are not (statistically) significant. The real long-term interest rate is also not significant. The absence of statistically significant impact of interest rates is in contrast to theory, although consistent with other empirical studies.

Table 7.8.Explaining GDP Growth in EMU Countries, 1980–2001: Regressions I–V1
IIIIIIIV2V2
Growth (t − 1)0.31(3.1)0.28(2.9)0.26(3.1)0.26(2.5)0.23(2.4)
Fiscal impulse (t)0.18(1.9)0.39(3.5)0.523(5.4)0.35(4.1)0.44(5.2)
(t− 1)−0.02(−0.21)−0.02(−0.2)−0.02(−0.2)−0.01(−0.1)0.02(0.2)
Partner countries’ imports (t)0.26(5.2)0.25(5.5)0.25(5.1)
(t − 1)0.01(0.2)0.05(1.1)0.07(1.4)
Real exchange rate (t)−0.04(−1.8)−0.05(−2.3)−0.05(−2.3)−0.02−0.8)−0.02(−1.2)
(t − 1)0.01(0.8)0.02(1.8)0.02(1.9)0.05(2.8)0.04(2.7)
Real short-term interest rate (t)−0.01(−0.1)0.01(0.1)−0.01(−0.1)0.03(0.4)0.00(0.0)
(t − 1)−0.08(−0.9)−0.06(−0.7)−0.07(−0.7)0.00(0.05)−0.04(−0.5)
Real long-term interest rate (LIR) (t)0.09(1.3)0.04(0.4)0.00(−0.0)0.07(0.7)0.06(0.6)
(t − 1)−0.02(−0.2)−0.07(−0.6)−0.02(−0.2)−0.30(−2.2)−0.21(−1.8)
Differential fiscal impulse effect
1 Different coefficient in “problem group”30.01(0.2)−0.01(−0.2)
(t − 1)−0.15(−1.5)−0.20(−1.7)
2 Additional coefficient for 1990s−0.19(−1.3)−0.23(−1.4)−0.06(−0.5)−0.07(−0.6)
3 1990s, Finland−1.17(−4.3)−0.80(−2.9)−1.40(−9.6)−1.06(−6.2)
1990s, Greece−0.48(−4.2)−0.59(−4.1)−0.58(−3.6)−0.65(−3.5)
Differential LIR effect
1 Additional coefficient in 1990s−0.13(−1.0)−0.07(−0.6)−0.04(−0.3)−0.50(−3.1)−0.49(−2.8)
(t − 1)0.05(0.2)0.03(0.2)0.00(−0.0)0.85(3.5)0.80(3.7)
Constant−0.08(−1.0)−0.11(−1.5)−0.12(−1.6)−0.06(−0.1)0.05(0.1)
Sigma2.152.092.052.021.99
Number of observations193193193193193
Serial correlationNoNoNoNoNo
AR(1) test: N(0,1)−2.16*−2.16*−2.15*−2.18*−2.20*
AR(2) test: N(0,1)−1.197−0.470.240.501.03

One-step generalized matrix of moments (GMM) with first-difference transformation, including individual effects.

Includes time-specific effect.

Two groups are distinguished; the fiscal effects are allowed to be different in the second group (Belgium, Finland, Portugal, and Spain).

One-step generalized matrix of moments (GMM) with first-difference transformation, including individual effects.

Includes time-specific effect.

Two groups are distinguished; the fiscal effects are allowed to be different in the second group (Belgium, Finland, Portugal, and Spain).

A Hausman specification test was used to test the hypotheses HA—“The set of regressors/instruments is strictly exogenous”91—by comparing the “within” estimator (only consistent if HA is true) with the first-difference estimator. The calculated X2 (26 degrees of freedom) of 2.29 means that HA cannot be rejected. Another Hausman specification test was used to test HB: “The set of individual effects are uncorrelated with the set of regressors/instruments.” Since HA was accepted, this test had to be based on a “usual” (Hausman and Taylor, 1981) test, using the difference between the within estimator and the GLS estimator.92 The calculated X2 (26 degrees of freedom) of 0.39 means that HB can be accepted, indicating that there is no need to make additional adjustments to the estimation procedure to remove such correlation.

Regression II

Regression II aims to capture “Maastricht effects” in the 1990s to verify their potential impact via fiscal multipliers or other channels. The results of von Hagen and others (see Box 5.3 in Section V) were replicated in a regression that allows for a differential impact of the fiscal impulse in the 1990s.93 The result, roughly in line with theirs, would suggest that the impact of the fiscal impulse was significantly lower in the 1990s—close to zero (results are not presented in the table). Further investigation revealed however that this differential impact in the 1990s is concentrated in only two countries: Finland and Greece. Once variables are included that model a differential impact of the fiscal impulse in the 1990s in these two countries, the general variable that models differential impact of the fiscal impulse in the 1990s is not significant anymore. The inclusion of these country-specific variables, which are strongly significant, cause the general coefficient on the fiscal impulse to increase to 0.4 (see Table 7.8).

The coefficients on other variables are virtually unchanged, compared with regression I, and none of the interest rate variables that were insignificant in regressions I and II is significant now, which suggests that the specification is robust. Partner-country import demand remains highly significant, with a coefficient of 0.25. The question remains why the impact on growth of the fiscal impulse is measured to be so different in the 1990s in Finland and Greece. In Finland, the collapse of the Soviet Union in the early 1990s and the take-off of the information and communication technology sector later in the decade have disrupted traditional economic relationships.94 For Greece, the reasons are less clear.

With a view to capturing Maastricht effects via other channels, this equation also includes a variable that captures a differential effect on growth of the long-term interest rate in the 1990s (in addition to the long-term interest rate itself). The impact of the long-term interest rate on growth in the 1990s is potentially stronger than in the 1980s, due to financial market liberalization. In the 1980s, financial markets and long-term interest rates in several European countries were still controlled. In such circumstances, one would not expect long-term interest rates to capture confidence and credibility effects. However, neither the long-term interest rate nor the differential long-term interest rate effect in 1990s appears to be significant.

Regression III

Regression III incorporates a further adjustment. Individual country regressions (discussed in more detail below) suggested significantly higher fiscal multipliers for many countries, with the multiplier between 0.4 and 1 for the large EMU countries (Germany, France, and Italy). To take account of the fact that the results are affected by a few countries, regression III allows for the fiscal multiplier to be different in a group of countries consisting of Belgium, Finland, Spain, and Portugal. Then, the results for the “normal” group gives a multiplier of 0.52. As before, the coefficients on partner countries’ import growth and the real exchange rate remain roughly unchanged, and none of the interest rate variables that were insignificant in regressions I and II is significant now. Again, the question is why the fiscal multipliers appear to be so low in these countries. As discussed, Finland went through drastic structural changes in the 1990s that do appear not to be picked up correctly in the explanatory variables. For Portugal, only a limited number of observations was available. For Belgium and Spain, it is less clear.

Regressions IV and V

Regressions IV and V replicate regressions II and III but replace the partner countries’ import growth variable with a time-specific effect to fully remove any endogeneity resulting from the partner countries’ import growth variable. These regressions do not include individual effects, in order to limit the number of parameters. Comparison of the results of IV with II, and those of V with III, indicates that the coefficient for the fiscal impulse is quite similar, supporting the robustness of the results of regressions I–III. This exercise validates the results of the Hausman tests that suggests that the set of regressors/instruments is exogenous, even though the coefficients of the less significant variables (the real exchange rate and the interest rates) are somewhat affected.

Country-Specific Regressions

An attempt was made to estimate this model on individual countries. With a low number of observations (21), the number of regressors was kept to a minimum by removing variables with insignificant effects from the standard general setup (in stages). The results should be interpreted with care, given the low number of observations. However, some interesting observations can be made. Virtually all measured results conformed with theoretical priors. In most countries, statistically significant effects were found for both the fiscal stance and partner countries’ imports. The other effects were significant in fewer cases. Theoretically, fiscal multipliers are smaller, and the impact of partner countries’ import demand stronger, in more open countries. The results presented in Figure 7.16 roughly conform to these priors.

Figure 7.16.Euro-Area Countries and CECs: Determinants of GDP Growth and Openness

(Estimates of fiscal multipliers and the elasticity of GDP growth with respect to partner countries’ import demand)1

Sources: OECD, Analytical Database, June 2002; and IMF staff estimates.

1 Estimates for the CECs are derived by extrapolating the relationships between openness and the variables on the x-axes found for the euro-area countries.

2 From estimation on individual euro-area countries (1980–2001).

3 Exports plus imports of goods and services, as percent of GDP (all in current prices, average 1980–2001).

Conclusions

The results suggest that, looking at 1980–2001, GDP growth in EMU countries has been determined mainly by changes in partner countries’ import growth and the fiscal stance. The fiscal multipliers are sizable; not as large as typically assumed in structural macro models, but larger than found in some other recent studies. Fiscal multipliers are in general not found to have become smaller in the 1990s. In the preferred specifications (regressions II and III), the fiscal multipliers are found to be 0.39 and 0.52, respectively. Tentative results on individual country estimation suggests that, overall, fiscal multipliers are smaller in more open countries. Because of the high variation of partner countries’ import growth over time, this variable contributes more to the explanation of variation in GDP growth rates than fiscal policy. This finding indicates the importance of including this variable, and the caution required in interpreting studies that do not include it. An additional effect, with the expected sign, is found for the real exchange rate. No statistically significant effects could be found for real short-term and real long-term interest rates.

IMF (2003d) provides a comprehensive survey of the issues related to public debt sustainability in emerging markets.

This assumes that a government credibly committed to servicing its debt in all circumstances should not borrow more than it would be able to service if low revenue outcomes were to persist for several years (IMF, 2003d).

The EC calculated “minimum cyclical safety margins” in 1998 for then-ERM countries and recalculated them in 2000 and 2002 (EC, 2002b).

A growing literature suggests that fiscal contractions can have expansionary effects when initial public debt and interest rate premium are high (Box 7.2). Public debt ratios and interest rates in the CECs, however, fall below thresholds for such positive effects.

In other studies where an effect is found, it is small. For example, in simulations using the IMF MULTIMOD model, Hunt and Laxton (2003) find that a 6 percentage point increase in the policy interest rate has the same effect on output as a 1 percent of GDP reduction in the fiscal deficit for industrial countries.

During 1994–99 for Greece.

Effective interest rates are calculated as interest paid relative to the stock of debt, including both domestic and foreign-currency-denominated debt.

For example, notwithstanding large inflows from the EU, Kopits and Székely (2004) estimate that, beyond 2006, EU accession will impart a negative direct net budgetary effect of 3–4¾ percent of GDP per year for the CECs. They suggest, however, that positive indirect effects may offset part of this fiscal cost.

In Spain, from 1992, each region agreed with the central government its maximum permitted deficit and debt. These agreements were revised to ensure consistency with Spain’s Convergence Programs and, in 1998, its first Stability Report.

See Ter-Minassian (1997) for the importance of this point.

Mackenzie and others (2003) discuss the effects of fiscal reform on national saving and the fiscal stance.

The size of B-S effects is directly related to shares of nontradables in the consumption basket—estimated at 34 percent in the CECs (Cipriani, 2000) and 57 percent in the euro area (Aitken, 1999). As this difference shrinks, B-S effects in the CECs will rise.

Monetary policy frameworks subsume an exchange rate regime. For brevity, therefore, “monetary policy framework” refers to the joint choice of monetary and exchange rate operating mechanisms.

Frankel (2003) notes that for most problems economists believe in the optimality of interior solutions.

Finland adopted inflation targeting a few months after the markka was allowed to float; Spain adopted inflation targeting while in ERM, and thereafter explicitly had two nominal targets.

For example, in January 1997, Finland defended the markka with large sterilized intervention in response to large speculative inflows.

Box 5.1 has a detailed description of Hungary’s experience.

However, Krugman (1991b) clarifies that a key feature of a target zone is that it allows a fairly wide range of variation for the exchange rate around the reference rate. Williamson (1998), for example, calls for a range of ±10 percent.

Hochreiter and Tavlas (2004) analyze the experience of Austria during this period.

At the same time, short-term interest rates differentials against those EU countries with the lowest rates remained insignificant, and inflation was brought under control (ECB, 1998).

Results using forward inflation and excluding a constructed cyclical component of the interest rate using a Taylor rule were not materially different.

From a statistical standpoint, long-term rates might be collinear with short-term rates. But this is an empirical matter, to be tested. In this case, no collinearity problems were observed.

The orthogonal deviations transformation used here removes such correlation. If the country effects were to be “hopelessly” correlated with all regressors, the “within” estimator should be used, since it would remain unbiased and consistent even then (Baltagi, 2001).

Similar to the inclusion of time-specific effects in panel estimations on different industries in one country to account for aggregate demand shocks.

If the disturbances are not serially correlated, there should be evidence of significant negative first-order serial correlation in differenced residuals and no evidence of second-order serial correlation in differenced residuals.

The Hausman specification test compares an estimator that is efficient (or more efficient) under the null hypothesis but inconsistent under the alternative with an estimator that is inconsistent (and less efficient) under both hypotheses (Baltagi, 2001).

See Baltagi (2001) for details. The “within” estimator is “consistent rain or shine,” but if HB should be rejected, the generalized least squares (GLS) estimator is not consistent.

Created by multiplying the fiscal variable with a dummy for the 1990s.

The partner countries’ import growth variable does not appear to pick up correctly the shock caused by the loss of trade with the Soviet Union in the early 1990s.

    Other Resources Citing This Publication