Fiscal Rules

Evidence suggests that in the design and implementation of fiscal frameworks concern for cyclicality has been secondary, with the primary focus generally on long-term sustainability. When dealing with cyclicality issues, most frameworks appear to take for granted the policymaker’s intent to run a countercyclical policy and overlook the actual tendency to asymmetric procyclicality. These frameworks often rely on nominal deficit ceilings or posit fiscal targets over the medium term. This means that effectively they play little role in correcting the policymakers’ incentives in good times (Box 4.1).

Box 4.1.

Cyclical Asymmetry in Fiscal Policy and Fiscal Targets “Over the Cycle”

By overlooking the tendency of policy toward asymmetric cyclicality, fiscal frameworks based on targets defined as averages over the cycle may carry the seeds of their own demise.¹ Reference to average budgetary outcomes allows the conduct of relatively loose policies and the postponement of adjustment towards the end of the reference period (or the postponement of the end of the reference period itself, if this is not tightly specified). The likely result is that the fiscal framework will come under strain as soon as it requires a procyclical adjustment in bad times. The unchecked deficit bias in good times will ultimately undermine the credibility of the framework because its prescriptions will be seen as lacking economic rationale; it will be either de facto disregarded or repudiated outright.

Recent developments in the European Union provide a clear example of this type of problem. The difficulties of several member states in keeping their deficits within the limit set by the Maastricht Treaty arose after 2001, in an adverse macroeconomic environment. However, their origins are rooted in insufficiently ambitious policies over 1999–2000, when conditions were favorable (see figure).

France and Germany: Cyclically Adjusted and Nominal Balances, 1998-2003

(In percent of GDP)

Source: European Commission (2005).

• Download Figure
• Download figure as PowerPoint slide

View Full Size

France and Germany: Cyclically Adjusted and Nominal Balances, 1998-2003

(In percent of GDP)

Source: European Commission (2005).

• Download Figure
• Download figure as PowerPoint slide

France and Germany: Cyclically Adjusted and Nominal Balances, 1998-2003

(In percent of GDP)

Source: European Commission (2005).

• Download Figure
• Download figure as PowerPoint slide

In this case, the different status of the 3 percent of GDP ceiling set for the annual nominal deficit and of the medium-term objective of close to balance or in surplus may have been an augmenting factor. While the former is defined in the Maastricht Treaty and sanctions are foreseen in case of noncompliance, the latter is defined in the Stability and Growth Pact and is not backed by similar incentives. This asymmetry has allowed attention to focus on the nominal deficit ceiling at both the policy and the monitoring levels, thus reducing pressure against the adoption of procyclical policies in good times. Significantly, most of the reforms proposed by policymakers aim at relaxing the nominal deficit ceiling in bad times, while little attention is paid to how to induce countercyclical behavior in good times. This tendency is also reflected in the reform proposals recently agreed upon, as described in European Council (2005); see also Annett, Decressin, and Deppler (2005).

¹ This box draws on the discussions in Balassone (2005).

The experience with the Stability and Growth Pact (SGP) adopted by member countries of the European Economic and Monetary Union (EMU) illustrates the difficulties arising when tackling both sustainability and cyclicality issues in a “rules-oriented” framework. Despite the intent to avoid procyclical policies, the SGP’s initial focus, as enshrined in the Maastricht convergence criteria, has been on sustainability. The “hard-law” clauses are focused on sustainability (the 3 percent deficit ceiling), while the cyclicality issue is addressed by provisions more akin to “soft law” (the close-to-balance or in surplus position to be observed over the medium term). While in several countries this framework was successful in reconciling fiscal discipline and flexibility, in others it was not able to reduce procyclicality in good times. This mixed record likely reflects differences in the degree to which the SGP rules feed through to the national fiscal framework. A “medium-term objective of a budgetary position close to balance or in surplus” as such provides little guidance concerning yearly targets, and the nominal deficit ceiling is unlikely to be binding in periods of high growth. Therefore such a framework may not be effective where it is not backed by political commitment (e.g., in the form of a multiannual fiscal program announced at the beginning of a government’s term), and by transparency in both the definition and the execution of the budget to facilitate monitoring and promote accountability (Hallerberg, Strauch, and von Hagen, 2001 and 2004).

Similar problems can arise in less rules-oriented frameworks, which rely more on public scrutiny and democratic accountability to ensure the credibility of commitment to sound policies. For instance, the fiscal frameworks introduced in Australia, New Zealand, and the United Kingdom all feature a reference to medium-term objectives for the budget balance, where medium term is usually interpreted as referring to the length of the economic cycle. The intention is to allow policy to respond countercyclically to changing macroeconomic circumstances. But the definition of a medium-term objective does not effectively prevent slippages associated with asymmetric procyclicality. This is in part because of the temptation to take some leeway in the short term and count on correction at the end of the cycle.

In developing economies, low market tolerance for high debt has in general produced an even greater emphasis on sustainability. However, as economic development has continued apace, there has been an increasing realization of the role fiscal policy can play in influencing economic fluctuations, with a corresponding increase in the attention paid to the cycle. A clear example of this is Chile, where a fiscal rule explicitly targeting a cyclically adjusted surplus on a yearly basis was introduced in 2001. The rule has been operated through both a downturn and the ensuing rebound. While the Chilean economic environment may be considered to be particularly favorable, given its low public debt and high policy credibility, the fiscal framework has been seen to be successful.

Several options are available to improve the balance between sustain-ability and cyclicality concerns in the design of fiscal frameworks. In frameworks based on “broad guidelines” and assuming sustainable budgetary positions, an explicit requirement that reaction of fiscal balances to changes in output gap be proportionate and symmetric could be adopted.² In rules-based fiscal frameworks, an option would be to introduce yearly targets in terms of cyclically adjusted fiscal balances. In this context, rules concerning the evolution of expenditures can be an important complementary instrument.

Regardless of the precise option, an assessment of the economic cycle and its impact on the budget is critical to the pursuit of countercyclical policies. In the absence of a reasonably reliable gauge of the state of the cycle and its implications for revenues and expenditures, policy cannot be implemented effectively. Indeed, in these circumstances, it is likely to end up being counterproductive, amplifying the cycle. This raises a number of analytical and operational issues, including the role of automatic stabilizers, which are explored in the following two sections.

Cyclically Adjusted Fiscal Balances

A reliable indicator of the cyclical position of the economy and of its impact on the budget is a precondition to promoting countercyclical fiscal policy in good times. It is essential for the design of policy in the first place, but also necessary to monitor outcomes and to hold policymakers accountable. Targeting cyclically adjusted fiscal balances can, in principle, assist in the design and monitoring of policy by focusing on the discretionary component of the budget.

The cyclically adjusted fiscal balance (CAB) is obtained by removing the cyclical component of the budget from the nominal fiscal balance. The cyclical component, in turn, depends on two factors: the size of the output gap; and the output elasticity of the budget, which is determined by the extent to which individual budgetary items react to fluctuations in output, as well as by the size of the budget (Box 4.2).

However, while CABs are regularly used by international organizations and national institutions, budgetary targets are seldom framed in cyclically adjusted terms. This reflects in part the relative complexity of the techniques used for the estimation of output gaps and budgetary elasticities. There are two main issues: (1) different methods for estimating CABs can yield different results; and (2) forecasts and outturns of CABs can be subject to, respectively, large errors and significant revisions, regardless of the specific method used. These issues are examined below.

Estimation Method and Accuracy of Estimates

Concerns over the accuracy with which CABs measure the impact of economic environment on the budget arise from a variety of sources: first, different methods of estimation of trend output and output gap produce different results; second, accurate estimation of budgetary elasticities is not always feasible given the informational requirements; and third, gauging the impact of the economic environment on the budget by means of the output gap alone may be insufficient.

The estimates of output gap are known to vary considerably depending on the method used. The results obtained with different methods display strong short-term comovements, although the range of level estimates is wide (see Orphanides and van Norden, 2002). A comparison of estimates prepared between 1999 and 2002 by major international organizations for the euro area reveals substantial dispersion (European Central Bank, 2005). Estimates published by the European Commission, the Organization for Economic Cooperation and Development, and the IMF—both real time and the latest available—span over a range often wider than 1 percent of GDP with averages of a similar order of magnitude. Given the structural shifts occurring in developing countries noted earlier, these problems are likely to be even more acute in those economies.

The accurate estimation of budgetary elasticities can be data intensive and require detailed institutional knowledge. In the absence of the requisite information, elasticities are often estimated econometrically using mac-roeconomic variables. A key drawback of this method is that its accuracy depends on adequately controlling for discretionary policy, while information on discretionary policy is often an output rather than an input to the computation of CAB. Moreover, econometric estimates may not distinguish between automatic and discretionary, albeit systematic, responses of budgetary items to cyclical conditions. Given these considerations, reliable estimates of budget elasticities may not be available for many countries.

Composition effects, which are often not captured in the CABs, can also matter. It is often the case that cyclical movements of various components of demand and national income are not synchronized. The resulting changes in the composition of output (for instance, the extent to which it reflects domestic or foreign demand) can have a substantial impact on the assessment of the cyclical component of the budget. A clear illustration is provided by developments during the 1990s in Europe where composition effects have been noticeable in a wide range of countries, with particularly marked effects in Belgium, Italy, and the Netherlands, reaching a peak of between ½ and 1 percent of GDP (Bouthevillain and others, 2001).

Factors other than output also affect the budget. While attention is usually restricted to GDP and employment movements, other factors—including interest rates, inflation, international commodity prices, exchange rates, and asset prices—can automatically have an impact on government budgets. These variables affect the budget directly—interest rates affect outlays; commodity and asset prices affect revenues—as well as indirectly through their impact on output and unemployment. In many emerging market and developing countries, the impact of these factors may indeed be dominant. Hence, any systematic evaluation of the impact of “environment” needs to take them into account (see Box 4.3).

Errors in Forecasts and Revisions of Outturns

In assessing how reliable and usable is a CAB as an indicator, one benchmark would be the nominal balance itself. The latter is readily avaliable and requires minimal elaboration. It is also the standard reference for policy design, implementation, and assessment. The simplicity of the nominal balance is a key attraction.

Nonetheless, despite the relative complexity of CABs, the magnitude of errors in forecasts for these need not be very different from those for nominal balances. While forecasts for the actual nominal fiscal balance depend on estimates of actual GDP, forecasts for CAB depend on estimates of trend GDP. In general terms, it is not possible to say a priori which of the two estimates is subject to greater error. The order of magnitude is likely to be the same under most circumstances because, by design, forecasts of trend GDP are based on forecasts of actual GDP.

In an analysis undertaken by IMF staff, a comparison of errors in forecasts of actual output and trend output does indicate similar orders of magnitude. Forecasts for trend output in a given year, as computed in that year, were compared with the outturn estimated in the following year for a sample of industrial and developing countries over the 1990s.³ Differences were generally substantial, but they were markedly lower for industrial countries than for developing countries (around 1 percent and 2.6 percent, respectively). However, for the same sample, errors in forecasts of actual output (based on the IMF’s World Economic Outlook database) had an absolute average of 0.95 percent in industrial countries and of 3.31 percent in developing economies (Table 4.1).⁴

Box 4.3.

Other Factors Affecting the Budget: Exchange Rates and Commodity Prices

The fiscal balance reflects the current implications of past policies, “environmental” conditions, and new legislation. In the computation of cyclically adjusted fiscal balances, the impact of policy is obtained residually, after estimating the effect of environmental conditions, which are identified with deviations of output from potential. However, the arbitrary restriction of the set of relevant environmental variables to the output gap can result in biased estimates of the impact of policy.

A case study of the Asian crisis. IMF (1998) presents a decomposition of year-on-year changes in the fiscal balance by underlying factor in four countries (Indonesia, Korea, the Philippines, and Thailand) during the Asian crisis of the late 1990s. The factors considered include three “environmental” variables (the exchange rate, oil prices, and growth) and “new” policy actions. In all countries, the deterioration of the economic environment is found to contribute substantially to the deterioration of fiscal balances. The most relevant factor is the massive depreciation in exchange rates. Oil prices significantly affect the fiscal balance of only Indonesia.

Source: IMF (1998); these data reflect information available during or in the immediate aftermath of the crises and do not incorporate subsequent revisions.

			Indonesia		Korea		Thailand		Philippines
			1997/98	1998/99	1997	1998	1996/97	1997/98	1997
Change in fiscal
	balance		−2.2	−9.2	−4.0	−4.0	−3.5	−0.4	−1.2
Change due to
		environment	−4.2	−11.1	−1.5	−0.3	−3.1	0.7	−1.5
	Exchange rate		−3.5	−6.4	−0.9	−0.2	−2.0		−0.9
	GDP growth		−0.5	−4.0	−0.6	−0.1	−0.9	0.6	−0.6
	Oil price		−0.2	−0.7
Change due to policy			2.7	1.7	−2.5	−2.6	−0.6	−0.6	1.6
Residual (unexplained)			−0.7	0.2		−1.1	0.1	−0.5	−1.3

Source: IMF (1998); these data reflect information available during or in the immediate aftermath of the crises and do not incorporate subsequent revisions.

View Table

			Indonesia		Korea		Thailand		Philippines
			1997/98	1998/99	1997	1998	1996/97	1997/98	1997
Change in fiscal
	balance		−2.2	−9.2	−4.0	−4.0	−3.5	−0.4	−1.2
Change due to
		environment	−4.2	−11.1	−1.5	−0.3	−3.1	0.7	−1.5
	Exchange rate		−3.5	−6.4	−0.9	−0.2	−2.0		−0.9
	GDP growth		−0.5	−4.0	−0.6	−0.1	−0.9	0.6	−0.6
	Oil price		−0.2	−0.7
Change due to policy			2.7	1.7	−2.5	−2.6	−0.6	−0.6	1.6
Residual (unexplained)			−0.7	0.2		−1.1	0.1	−0.5	−1.3

Source: IMF (1998); these data reflect information available during or in the immediate aftermath of the crises and do not incorporate subsequent revisions.

A significant part of observed changes in fiscal balances remains unexplained. This partly reflects the effect of past policies and of other environmental factors. However, it is also a consequence of the judgmental nature of the assessment of what is a “new” policy action. Such judgment can also affect the partition between environmental and policy effects. In the case of Indonesia, for instance, the large environmental effect of exchange rate movements reflects the increase in commodity subsidies as a consequence of the failure to fully adjust the local price of imported goods. This could equally be seen as the result of a deliberate policy decision to expand social spending.

Adjusting for the price of oil in Russia. IMF staff analysis of the direct impact of oil prices on energy taxes in Russia suggests that at current prices, a $1 increase in the oil price would raise revenue by 0.4 percent of GDP. While estimates of the impact of the economic cycle on Russia’s government budget are hampered by the lack of any normal economic cycle during most of the transition period so far, staff uses an oil-price-adjusted fiscal balance to have an indication of the cyclically adjusted fiscal position. The sensitivity of revenues to changes in oil prices is estimated by category of taxes.

Adjusting for the price of copper in Chile. Starting in 2001, Chile began targeting a 1 percent of GDP structural budget surplus. This is not enshrined in legislation; rather, it is a commitment of the current government. All the adjustments to get the structural balance affect the revenue side of the budget as no expenditure item is judged to have a significant automatic cyclical component. The adjustments are made against a benchmark when both the economy is operating at full potential and copper price is at its long-term (10 years) level. (Copper is Chile’s leading export; about 4 percent of total government revenues come directly from copper, but copper feeds through the entire economy.) Two separate panels of up to 14 experts each are assigned the task of computing potential output and long-term copper price, as well as the implication for the budget of deviations from such values. Each expert submits her or his estimate, the two extreme on each side are disregarded, and the simple average of the remaining 12 is used as a benchmark forecast.

The above suggests that forecasts of CABs in general need not be subject to larger errors than those concerning nominal balances. Indeed, for countries in the euro area, the average absolute difference between forecasts and outturns for CABs over 1999–2002—as computed by the European Commission—was in fact very close to the corresponding figure for nominal balances, 0.99 percent and 0.85 percent of GDP, respectively (Table 4.2).

Table 4.1.

Errors in Forecasts in Actual and Trend Output (t Versus t +1)¹

Sources: IMF, World Economic Outlook database; and IMF staff calculations.

¹In percent of GDP.

²These included Australia, Austria, Belgium, Canada, Denmark, Finland, France, Germany, Greece, Iceland, Ireland, Italy, Japan, the Netherlands, New Zealand, Norway, Portugal, Spain, Sweden, Switzerland, the United Kingdom, and the United States.

³These included Argentina, Bangladesh, Brazil, Cambodia, Chile, China, Côte d’lvoire, the Czech Republic, Egypt, Ghana, Hong Kong SAR, Hungary, India, Indonesia, Jordan, Kenya, Korea, Latvia, Lebanon, Lithuania, Malawi, Malaysia, Mexico, Nigeria, Pakistan, Peru, the Philippines, Poland, Russia, Singapore, the Slovak Republic, South Africa, Thailand, Turkey, Uruguay, and Republica Bolivariana de Venezuela.

Table 4.1.

Errors in Forecasts in Actual and Trend Output (t Versus t +1)¹

	Actual Output		Trend Output
		Standard		Standard
	Mean	deviation	Mean	deviation
Full sample	2.39	1.42	2.01	1.07
Industrial countries2	0.95	0.16	0.97	0.19
Developing economies3	3.31	1.04	2.64	0.87

Sources: IMF, World Economic Outlook database; and IMF staff calculations.

¹In percent of GDP.

²These included Australia, Austria, Belgium, Canada, Denmark, Finland, France, Germany, Greece, Iceland, Ireland, Italy, Japan, the Netherlands, New Zealand, Norway, Portugal, Spain, Sweden, Switzerland, the United Kingdom, and the United States.

³These included Argentina, Bangladesh, Brazil, Cambodia, Chile, China, Côte d’lvoire, the Czech Republic, Egypt, Ghana, Hong Kong SAR, Hungary, India, Indonesia, Jordan, Kenya, Korea, Latvia, Lebanon, Lithuania, Malawi, Malaysia, Mexico, Nigeria, Pakistan, Peru, the Philippines, Poland, Russia, Singapore, the Slovak Republic, South Africa, Thailand, Turkey, Uruguay, and Republica Bolivariana de Venezuela.

View Table

Table 4.1.

Errors in Forecasts in Actual and Trend Output (t Versus t +1)¹

	Actual Output		Trend Output
		Standard		Standard
	Mean	deviation	Mean	deviation
Full sample	2.39	1.42	2.01	1.07
Industrial countries2	0.95	0.16	0.97	0.19
Developing economies3	3.31	1.04	2.64	0.87

Sources: IMF, World Economic Outlook database; and IMF staff calculations.

¹In percent of GDP.

²These included Australia, Austria, Belgium, Canada, Denmark, Finland, France, Germany, Greece, Iceland, Ireland, Italy, Japan, the Netherlands, New Zealand, Norway, Portugal, Spain, Sweden, Switzerland, the United Kingdom, and the United States.

³These included Argentina, Bangladesh, Brazil, Cambodia, Chile, China, Côte d’lvoire, the Czech Republic, Egypt, Ghana, Hong Kong SAR, Hungary, India, Indonesia, Jordan, Kenya, Korea, Latvia, Lebanon, Lithuania, Malawi, Malaysia, Mexico, Nigeria, Pakistan, Peru, the Philippines, Poland, Russia, Singapore, the Slovak Republic, South Africa, Thailand, Turkey, Uruguay, and Republica Bolivariana de Venezuela.

The situation is different with regard to the stability of initial assessment of outturns. The computation of the nominal balance for a given period only requires data for that period. Subsequent computation can only lead to different results if data for that period are revised. In contrast, the computation of the corresponding CAB requires also output forecasts for subsequent time periods. This is so since, regardless of the method used, estimates of trend GDP are essentially weighted averages of realized data and forecasts of actual output for a number of periods ahead. This means that subsequent computation of CAB outturns for a given period can give different results simply because forecasts of actual GDP for subsequent periods are revised. This problem may be more significant in developing than in industrial countries, since the former are subject to greater and more frequent shocks and structural breaks.

Revisions to initial assessment of outturns are generally smaller than errors in forecasts, but there is a high likelihood that they are significant at economic turning points. IMF staff analysis suggests that revisions to initial assessment of outturns are about half the size of errors in forecasts (as measured by the difference between the forecast and the initial assessment of trend output for any given year; Table 4.3). For industrial countries the average revision of trend output outturns after one year is about ½ percent. As budget elasticity averages around ½ percent for industrial countries, this would in turn imply an average revision of CAB outturns after one year of ¼ percent of GDP. However, this average hides a considerable diversity of experience. In the case of turning points, the revisions can be quite large. A vivid example is provided by revisions in the assessment of recent budgetary developments in some European countries: the 2001 nominal deficit for France and Germany was estimated in 2002 at 1½ percent and 2¾ percent of GDP, respectively. The corresponding CAB estimates were 1¾ percent and 2½ percent (European Commission, 2002). Nevertheless, a year later (in 2003) while estimates of nominal deficits were essentially unchanged, CAB estimates were revised to 2½ percent and 3 percent, respectively (European Commission, 2003). This mainly reflected sharply lower than expected growth in 2002, leading to a downward revision of the estimated trend output for 2001, and a reduction of the corresponding negative output gap.

Table 4.2.

Revisions in Cyclically Adjusted Balances and Nominal Balances in the Euro Area, 1999–2002 ¹

Sources: European Commission, Spring Forecasts; and IMF staff calculations.

¹In percent of GDP.

Table 4.2.

Revisions in Cyclically Adjusted Balances and Nominal Balances in the Euro Area, 1999–2002 ¹

	(t Versus t + 1)		(t+1 Versus t + 2)
		Standard		Standard
	Mean	deviation	Mean	deviation
Nominal fiscal balances	0.85	0.48	0.34	0.26
Cyclically adjusted fiscal balances	0.99	0.57	0.89	0.62

Sources: European Commission, Spring Forecasts; and IMF staff calculations.

¹In percent of GDP.

View Table

Table 4.2.

Revisions in Cyclically Adjusted Balances and Nominal Balances in the Euro Area, 1999–2002 ¹

	(t Versus t + 1)		(t+1 Versus t + 2)
		Standard		Standard
	Mean	deviation	Mean	deviation
Nominal fiscal balances	0.85	0.48	0.34	0.26
Cyclically adjusted fiscal balances	0.99	0.57	0.89	0.62

Sources: European Commission, Spring Forecasts; and IMF staff calculations.

¹In percent of GDP.

Table 4.3.

Errors in Forecasts and Revisions of Outturns for Trend Output¹

Source: IMF staff calculations.

¹In percent of GDP.

²See footnotes to Table 4.1.

Table 4.3.

Errors in Forecasts and Revisions of Outturns for Trend Output¹

	Errors in Forecasts		Revisions of Outturns		Revisions of Outturns
	t Versus t+1		t+1 Versus t+2		t+2 Versus t+3
		Standard		Standard		Standard
	Mean	deviation	Mean	deviation	Mean	deviation
Full sample	2.01	1.07	0.53	0.67	0.80	0.35
Industrial countries2	0.97	0.19	0.52	0.10	0.24	0.04
Emerging markets2	2.64	0.87	1.53	0.57	1.14	0.34

Source: IMF staff calculations.

¹In percent of GDP.

²See footnotes to Table 4.1.

View Table

Table 4.3.

Errors in Forecasts and Revisions of Outturns for Trend Output¹

	Errors in Forecasts		Revisions of Outturns		Revisions of Outturns
	t Versus t+1		t+1 Versus t+2		t+2 Versus t+3
		Standard		Standard		Standard
	Mean	deviation	Mean	deviation	Mean	deviation
Full sample	2.01	1.07	0.53	0.67	0.80	0.35
Industrial countries2	0.97	0.19	0.52	0.10	0.24	0.04
Emerging markets2	2.64	0.87	1.53	0.57	1.14	0.34

Source: IMF staff calculations.

¹In percent of GDP.

²See footnotes to Table 4.1.

Large revisions are less likely in changes in CABs. This is primarily because the revisions of forecasts of actual output generally affect estimates of trend output for contiguous years in a roughly similar manner, leaving the change in trend output between years relatively unaffected. This suggests that revisions to changes in CABs are also likely to be smaller than revisions in their levels. Indeed, a recent study based on simulation analysis suggests that, for the EU countries, a revision in CABs of up to ½ percent of GDP only causes a revision of the change in CABs of up to 0.1 percent of GDP (González-Mínguez, Hernández de Cos, and del Río, 2003). The data revisions for France and Germany discussed above provide a concrete illustration in this respect. The change in the cyclically adjusted deficit from 2000 to 2001—as seen in 2002—was 0.9 percent of GDP for Germany, with no change for France; in 2003 these estimates were unchanged for Germany and only slightly changed (to 0.1 percent) for France.

The above discussion suggests that while the computation of CABs raises a number of issues, a variety of measures can be undertaken to make them a more robust reference for policy design and implementation. These include taking into account changes in the composition of output when estimating the output gap, as well as relying on estimates of elasticities derived from tax and expenditure laws. These estimates should be updated periodically and the impact of any new legislation should be assessed in a timely manner. The stability of policy assessment based on CAB estimates could be enhanced if attention focused more on the dynamics of output and budget balances rather than their levels, and particular attention is paid to the assessments around likely turning points. Further gains could be obtained by reference to medium-term estimates of trend output in order to make CAB estimates less influenced by the volatility of forecasts for output two or more periods ahead.

Moreover, for many countries even relatively unsophisticated CAB calculations would represent a significant step forward. Most of the improvements suggested above may be challenging in emerging market countries because of higher output volatility and inadequate data. However, even CAB calculations based on deviations from a simple measure of trend output, and on plausible values of the output elasticity of revenue and expenditure, can be more useful than nominal balances for assessing past or proposed fiscal measures. As noted above, reference to changes in trend output (and CABs) would reduce the consequences of an imprecise estimation of the level of potential output. Concerning budget elasticity, plausible benchmark values, taking into account stylized facts regarding revenue and expenditure elasticities, can be computed (Box 4.2). Controlling for the most significant changes in the composition of output (i.e., large shifts from domestic to foreign demand and vice versa) should also be feasible.⁵

In order to maximize the gains in terms of accountability expected from a sharper focus on the cyclicality of policy, care should be taken to reduce room for subjective interpretation of data. A transparent implementation of CAB targeting could be aided by delegating to an independent fiscal agency the macroeconomic assumptions underlying budgetary plans (in some countries trend output is presented as a policy objective), as well as the estimation of the nominal balance necessary to attain the desired CAB.

Targeting Expenditure

Targeting government expenditure can facilitate the implementation of countercyclical policy. Committing to a predetermined rate of growth of expenditure can curb the tendency to increase public spending in good times while leaving the automatic stabilizers on the revenue side free to operate.⁶ An expenditure rule of this type can be relatively easily disseminated to the public, and clearly monitored, provided that the control aggregates are clearly specified. Largely reflecting these factors, expenditure targeting—whether formally incorporated in a rule or not—has been playing a role in the fiscal framework of an increasing number of countries (Box 4.4). Nonetheless, a number of important issues arise in the implementation of an expenditure targeting framework. It should also be stressed that, even if implemented effectively, expenditure targeting as such does not preclude procyclical policies that may arise from the revenue side nor does it ensure budgetary sustainability.

A variety of issues can arise in the implementation of an expenditure targeting framework that will have a bearing on its effectiveness. These include the choice of expenditure aggregate to be targeted, with respect both to the items included, the institutional coverage, and the level of disaggregation of the ceilings. In addition, specific characteristics of the framework, including the time horizon, underlying macroeconomic assumptions, and valuation criteria can have an important impact.

In general, the expenditure aggregate to be targeted should be broad based. The use of comprehensive aggregates reduces the risk that expenditure ceilings are circumvented by increasing the outlays in the categories not included in the targeted aggregate or expanding the activity of bodies outside the coverage of the target. For instance, some expenditure items may be moved to the capital account to avoid restrictions applying exclusively to current outlays; transfers to public corporations may be channeled through capital injections; quasifiscal bodies may be set up to run outside the general government sector programs that are in fact a public responsibility; and local government spending may increase to compensate for restrictions on central government outlays. Concerning the latter, in countries where subnational governments have fiscal autonomy, expenditure targets for central government may have to be complemented by intergovernmental agreements. Finally, targeting gross expenditures, rather than net of nontax revenues, is preferable: a measure net of such revenues can be difficult to monitor, expose public spending programs to unnecessary volatility, and induce procyclicality.

It is preferable, however, for some specific items to be excluded. First, interest outlays need to be excluded since they are not a policy variable (at least in the short term). Second, spending on entitlements could also be excluded since it is generally difficult to change it in the short run; however, over the medium term, this category needs to be fully reflected in the aggregate target. Third, cyclically sensitive expenditure items (e.g., unemployment benefits) should be excluded to avoid impeding the operation of the automatic stabilizers. Fourth, since there may be a tendency to comply with the ceilings by compressing those items which are politically less costly in the short term, regardless of the long-term implication, it is often argued that investment spending should be excluded from the targeted aggregate. Interest spending and unemployment benefits may be relatively easy to identify. However, the distinction between current and capital spending can be more problematic and it should not be assumed that all public investment is productive (see, e.g., Balassone and Franco, 2000).

The level of disaggregation at which the ceilings are set has a bearing on the flexibility or rigidity of the framework. The key issue is the extent to which savings in some categories can be used to offset overruns in others. Lower than expected expenditure on some programs, whether they are the consequences of more favorable than expected cyclical developments or they result from productivity gains, should generally be saved. In the case of cyclical developments, this is necessary to let the automatic stabilizers work. However, excessively disaggregated expenditure ceilings may eliminate the needed degree of flexibility in the management of agencies that are assigned multiple tasks—as it is usually the case with the provision of public goods and services.

Setting multiyear ceilings has potential advantages. A multiyear framework makes the government’s fiscal goals explicit and can help build consensus around them. Both fiscal prudence and stabilization cannot be properly pursued outside a medium-term fiscal framework. However, the issue arises as to how to deal with the need for flexibility during the period over which the targets are fixed. In this respect, a key element is whether expenditure targets are set on a rolling basis or in advance for a given period. The former provides wider margins for flexibility as the revision of plans is allowed when macroeconomic developments deviate from the underlying assumptions. However, there is also a risk that such flexibility is misused. Ceilings set on a rolling basis may also come under pressure when there is a change in government. In this respect, the practice to fix expenditure targets at the beginning of the legislature and for its entire span, together with limited contingency provisions to revise the targets, may represent a more transparent solution and one conducive to a higher degree of accountability.

The macroeconomic assumptions underlying an expenditure target determine its effectiveness. In particular, the adoption of a “realistic” or a “cautious” scenario can have an important bearing. The advantage of a cautious scenario is the likelihood of a favorable surprise ex post, as cyclically sensitive spending (to the extent that it is included in the aggregate) is likely to turn out lower than expected (and revenues higher). Accordingly, each year there will likely be some room to finance new programs or to deal with unexpected spending overruns. However, the adoption of an excessively cautious scenario can obfuscate the true fiscal goals of the government, defeating the purpose of the multiyear fiscal framework. Moreover, with a cautious scenario automatic stabilizers would not operate correctly. A cautious scenario may also induce the spending ministries to second guess the real scope for additional fiscal relaxation, and to engage in procyclical policy when the outturn is favorable. A more transparent arrangement to allow margins for new discretionary spending would be to accompany the adoption of a realistic scenario with the introduction of a contingency reserve.

Considerations concerning both ease of monitoring and the implications for stabilization policy weigh in favor of nominal targets as compared to real ones. Nominal ceilings are simpler to implement and to monitor than real ones as the assessment of compliance does not require the computation of the relevant deflator, or introduce distortions when a general deflator (such as the GDP deflator) is used. Moreover, nominal ceilings can reinforce the response of automatic stabilizers to demand shocks as higher than expected inflation would automatically lead to lower real government spending. Similarly, nominal ceilings would help absorb permanent supply shocks. However, nominal ceilings can have drawbacks: from an administrative point of view, the cutbacks in real spending required when inflation is higher than expected do not contribute to a smooth execution of the budget; and reductions in real spending may not be the appropriate policy response to surprise inflation caused by temporary supply shocks.

Cash- and accrual-based targets can play a complementary role. Accrual measures are generally superior as a means for budgetary planning. However, within-year monitoring of fiscal developments can be impeded by the delay with which accrual figures are usually available. Moreover, since accrual and cash measures are affected differently by attempts at creative accounting to circumvent the targets, there may be synergies to be exploited by cross-checking the consistency of accounting records based on the two criteria (Balassone, Franco, and Zotteri, 2006).

Leaving aside the specifics of expenditure ceilings, it is important to ensure that the procyclical bias is not transferred to the revenue side of the budget, and that there is a long-term anchor to fiscal policy. It is of course the case that procyclicality can arise from the revenue side. During boom periods for instance, governments might be tempted to cut taxes or increase tax expenditures, even while observing expenditure ceilings. This essentially occurred in the case of a number of member countries of the European Union (EU) over 1999–2001, which later led to difficulties in complying with the EU fiscal framework. This suggests that the expenditure ceilings cannot be set in isolation from provisions regarding revenue policy.

More generally, expenditure targeting as such does not correct a structural tendency toward excessive deficits. A constant rate of growth of expenditure can be consistent with a gradual deterioration of the fiscal balance if revenues do not keep pace with expenditures. An anchor in terms of budget balance is therefore essential. In a situation in which the budget balance is deemed to be close to a sustainable level, based on an estimate of trend output growth and of the corresponding revenues, the expenditure ceiling could reasonably be set at a level consistent with keeping the balance unchanged. Where it is not, the expenditure target can be readily set to help attain sustainability. Nonetheless, as with all fiscal rules and targets, a crucial factor determining the effectiveness of expenditure ceilings is the extent to which they are backed by political commitment.

Procyclicality and Market Constraints

With regard to market constraints identified in Chapter 3, a variety of measures could play a complementary role in reducing the procyclicality of policy. There are three interrelated areas: first, measures that can help stabilize investor sentiment during downturns; second, measures to modify the structure of debt to reduce procyclical pressures; and third, instruments to help governments deal with the impact of sharp changes in economic and market environment.

A key element with regard to stabilizing investor sentiment relates to credibility of policies. Weak fiscal discipline in good times and an opaque policy framework can undermine credibility, and thereby aggravate financing difficulties during the downturn. Conversely, budgetary policies that are sustainable and transparent are likely to sustain confidence during the downturn (Corsetti, Guimarães, and Roubini, 2004). In this context, the relevance of credible medium-term budgetary frameworks, in conjunction with the types of expenditure rules noted earlier, is evident. Adapting institutions to deal with the common pool problem could also help in stabilizing market confidence. Measures that can increase social consensus on budgetary priorities may help in obtaining support for policies to deal with procyclicality, especially during the upswing, and stabilize investor confidence during the downswing.

The restoration of confidence may still require procyclical policies in the downturn to reinforce commitment to address the underlying lack of sustainability. Where fiscal correction is required, timing and quality of adjustment is crucial to reduce the degree of procyclicality. The signaling role is effective if policy measures are initiated early, and can help stabilize and support market confidence. This can lead to a significant reduction in risk premia and sovereign spreads, and maintain access during the downturn.

With regard to debt structure, debt management can be oriented to reduce the pressure to implement contractionary policies in the downturn. As credibility is attained and investor confidence stabilizes, a strategy geared to limiting the issuance of debt with short maturities, and in foreign currency, can play a particularly important role in this regard. As a number of recent cases have indicated (Turkey, for instance), given a sound policy framework, it is possible to shift gradually to longer-term debt denominated in domestic currency even as the economy is entering a downturn. This can then provide a temporary relief from immediate financing pressures since access to external finance is less critical.

There have also been a number of proposals for instruments playing a larger insurance role than the current credit instruments. These instruments could link more closely the requirements and the ability of a country to service its debt obligations. For instance, a country with substantial external debt could have a part of its liabilities converted to GDP-indexed bonds. These bonds would be structured so that the country pays less interest when growth is low to reduce the need for rolling over debt or adjusting spending during recession (Borensztein and others, 2004). The process would be symmetrical: if the economy performs well, interest payments are increased. In this way, the government’s ability to use the surge in revenues due to the upswing in activity is curtailed. Thus, while these instruments cannot remedy unsustainable policies, they can help reduce procyclicality.

GDP-indexed bonds are not used extensively.⁷ This suggests that financial market participants see constraints in their development and trading. While markets tend to be cautious in adopting new instruments (e.g., the adoption of collective action clauses), in this case, a number of specific issues arise. These relate to the initial fixed cost entailed in launching the instruments, adequate liquidity, the extent of standardization, the measurement of GDP and its trend growth, and uncertainty about the payoffs. None of this is unusual, nor is it unusual that a socially desirable instrument may not be forthcoming.

In such cases, international financial institutions can play a catalytic role. They could help, for instance, in the standardization of the instruments. They could also assist by encouraging the independence of statistical agencies and providing technical assistance to improve the quality of national income statistics, and assuring markets about their reliability. These institutions could also play a greater role in the issuance of bonds denominated in domestic currency and with longer maturities, in part by helping countries identify and deal with policy distortions that provide disincentives to issue debt with these characteristics.

Conclusions

This chapter has explored the role that cyclically adjusted fiscal balances and expenditure targets can play in ameliorating procyclicality of fiscal policies. In addition, for emerging market countries in particular, it considered measures to stabilize investor confidence and maintain market access to avoid procyclicality during economic downturns. The discussion points to the following conclusions:

• The availability of a reliable indicator to gauge the degree of procy-clicality of policy is a key precondition for any attempt at correcting the procyclical bias. While cyclically adjusted fiscal balances represent a natural candidate, international experience with fiscal frameworks confirms a certain degree of diffidence toward them. Moreover, while frameworks for fiscal responsibility have become increasingly alert to cyclical issues, CABs have generally not had a significant formal role in the design and implementation of fiscal rules.

• While there are a number of difficulties regarding the computation of CABs, a variety of measures can be undertaken to address them and CABs can play a useful role as a reference for policy design and implementation. The accuracy of CAB estimates can be improved by taking into account changes in the composition of output, and by increasing reliance on estimates of elasticities derived from tax and expenditure laws. Policy assessment based on CABs can be improved by focusing on changes in output and budget balances rather than their levels, and a reference to medium-term estimates of trend output can make CAB estimates less susceptible to volatility of forecasts for output.

• Nonetheless, CABs need to be used judiciously. This is especially so in emerging market and developing economies where structural breaks and output volatility complicate the task of estimating the trend output. Given the factors entailed in the computation of the measure, it is particularly important to have transparency in the estimation procedures. In any case, giving consideration to a more formal role to CABs in fiscal frameworks should not reduce the importance attached to the monitoring of nominal balances and debt dynamics.

• A multifaceted approach to dealing with the issue of procyclicality may be necessary. Since direct CAB targeting is likely to remain a challenge in many countries, targeting government expenditure can support the implementation of countercyclical policy in the context of deficit or debt targets. In the choice of expenditure aggregates, a balance needs to be struck between comprehensiveness and ease of monitoring.

• For emerging market economies in particular, a variety of market-related measures could play a complementary role in reducing policy procyclicality. Measures that can help external investor sentiment during downturns are likely to be helpful. At the same time, the structure of debt can be modified to reduce procyclical elements, and financial instruments can be developed to reduce the budgetary impact of sharp changes in the economic and market environment.