Information about Europe Europa
Journal Issue
Share
Article

Switzerland: Selected Issues and Statistical Appendix

Author(s):
International Monetary Fund
Published Date:
June 1998
Share
  • ShareShare
Information about Europe Europa
Show Summary Details

II. Fiscal Policy Rules and the Business Cycle26

A. Introduction and Summary

43. While Switzerland has a long tradition of sound public finances, the contribution of fiscal policies to macroeconomic stability over the business cycle has often been criticized.27 In particular, this criticism has emphasized the weakness of automatic fiscal stabilizers and the, at times, procyclical stance of discretionary fiscal policy. This chapter has two aims: One, to gauge quantitatively the automatic and discretionary responses of general government finances to cyclical output movements during 1970-96. And two, to examine the main options for improving the stabilization role of fiscal policy over the business cycle. In the latter context, the chapter describes and assesses a proposed constitutional amendment mandating a balanced Confederation (central government) budget over the business cycle after 2001.

44. The empirical results reported in this chapter suggest that the size of the automatic cyclical responsiveness of general government finances—the change in the general government balance as a percent of GDP in response to a 1 percentage point increase in the output gap—in Switzerland is broadly comparable to those in other industrial economies. However, a considerable lag between income accrual and income tax collections (2-4 years) weakens automatic fiscal stabilizers. Moreover, Switzerland’s moderate inflation rates notwithstanding, the extraordinary length of the income tax collection lags also reduces the real value of tax revenue collections. As regards the stance of discretionary fiscal policy, it is found that fiscal policy has often been procyclical during the period 1970-96, undermining the operation of the already weak automatic fiscal stabilizers, in particular in years of large excesses or shortfalls of aggregate demand.

45. The empirical results suggest that three adaptations of present fiscal policy rules would prove beneficial to enhancing macroeconomic stability over the cycle: (i) a significant shortening of the long lags in the collection of income taxes; (ii) the adoption of fiscal rules that avoid pronounced procyclical discretionary fiscal policies, in particular during periods of significant excesses or shortfalls of aggregate demand; and (iii) enhanced coordination of fiscal policy stances among the different level of governments. Indeed, various proposals to reform Switzerland’s fiscal institutions in these directions have been put forward during the last 30 years. However, Switzerland’s deeply entrenched status-quo bias appears to have lessened in this area. In particular, several cantons including the large canton of Zürich are considering reducing significantly the present long collection lags for income taxes. At the Confederation level, a constitutional amendment has been proposed that would aim after the year 2001 at a balanced budget over the business cycle at the Confederation level, while providing room for countercyclical discretionary fiscal policy in the face of large aggregate demand disturbances. Prospects for enhanced fiscal policy coordination over the cycle among the different tiers of government, however, appear to be dim, despite an already existing constitutional obligation mandating such coordination.

46. The remainder of this chapter is organized as follows. Section B provides some background on fiscal developments and institutions. Section C outlines a framework to gauge the cyclical responsiveness of budget balances. Sections D reports empirical estimates on the size of automatic fiscal stabilizers and the discretionary fiscal response to cyclical output fluctuations, respectively. Finally, section E describes and assesses the proposed fiscal policy rule for balancing the Confederation over the business cycle.

B. Background

47. Switzerland has a long-standing tradition of sound public finances. This is vividly illustrated by comparisons of developments in fiscal deficits and public debt in Switzerland and in other industrial countries over the last twenty-five years (Figure II-1). While the Swiss general government balance moved into deficit in the 1970s—as in most other industrial countries—a broadly balanced fiscal position was restored by 1980. At the same time, the public debt-GDP ratio declined steadily from the mid-1970s to 1990, contrasting with the sharply rising levels of public debt in many other industrial countries over the same time period. The on-set in 1991 of a prolonged economic slowdown in Switzerland was accompanied by a sharp rise in the general government deficit and a reversal of the trend decline in the public debt-GDP ratio. However, since 1993, the fiscal position has again improved markedly, reflecting strenuous consolidation efforts in the face of a protracted slump in economic activity.

Figure II-1.Switzerland: International Comparison of General Government Finances, 1970-96

(In percent)

Source: WEO database; European Economy N63, 1997.

48. In addition to exhibiting generally low average levels of fiscal deficits and debt (as a percent of GDP), the variability of Switzerland’s general government balance has also been remarkably low over the last 25 years, notwithstanding a comparatively high variability of real output fluctuations (Figures II-2 and II-3). Indeed, in many other small open industrial economies, the variability of general government balances was a multiple of Switzerland’s during the period 1971-96. However, as discussed in more detail in the following sections, the stylized fact of comparatively low variability of Switzerland’s fiscal balance is in part a consequence of relatively weak automatic fiscal stabilizers and the procyclical behavior of discretionary fiscal policies.

Figure II-2.Switzerland: Variability of General Government Balances in Selected Industrial Countries, 1971-96 1/

Source: WEO database; and staff estimates.

1/ Time range for all countries is 1971-96 except for Norway, where the time range is restricted to 1976-96.

Figure II-3.Switzerland: General Government Finances and the Business Cycle, 1970-96

Source: WEO database; and staff estimates.

1/ Average of output gaps in years t-2 and t-3 for even years; and average of output gaps in years t-3 and t-4 for odd years.

49. The operation of automatic fiscal stabilizers through changes in general government revenue is impeded by Switzerland’s idiosyncratic income tax collection system. At the Confederation level and iii most cantons, income tax collection is spread over several years. For example, the income tax due for the years 1997 and 1998 (the tax period) will be based on the average taxable income during the previous two years, i.e. 1995 and 1996.28 As income tax is payable in the year after it falls due, actual income tax payments in 1998 and 1999 will reflect average incomes accrued during the years 1995-96. Given the average length of typical business cycles in industrial countries during the post-war period, which was estimated by Zarnowitz (1992) at about 4-5 years,29 this considerable collection lag for income taxes will not only reduce the size of automatic fiscal stabilizers relative to the automatic fiscal response to cyclical output fluctuations but may act as a partial off-set to automatic fiscal stabilizers. For example, when the economy entered a recession at the beginning of the 1990s, income tax collections reflected the high incomes accrued during the boom years of the second half of the 1980s, thus further hemorrhaging private households’ real purchasing power when income growth was slowing due to the recession. More generally, plots of the series for the current cyclical output gap and the cyclical output gap relevant for income tax collections indicate a high degree of asynchronism between the series at critical business cycle junctures (Figure II-3).30

50. Switzerland’s high degree of fiscal federalism has led to a roughly equal division of spending powers and resources among the three levels of territorial governments, i.e. the Confederation, the cantons, and the communes (Figure II-4). In this setting, use of discretionary fiscal policy actions to off-set large macroeconomic disturbances would require a considerable degree of fiscal coordination between the different levels of government. Although the Constitution obliges all tiers of governments to take account of the state of the business cycle when establishing their budgets (Article 31quinquies), the high degree of fiscal decentralization and complex budgeting procedures do not provide an auspicious setting for ensuring fiscal co-ordination. Moreover, the tendency at the cantonal and communal levels to aim for balanced budgets has been reflected in broadly stable but also to some extent procyclical fiscal balance developments at the lower government levels (Figure II-4).

Figure II-4.Switzerland: Government Finances at Different Government Levels, 1970-96

Source: Swiss Federal Finance Administration.

51. A small literature has analyzed the cyclical behavior of fiscal policy in Switzerland. Wagner (1973) examined discretionary fiscal policy actions during the period 1955-70.31 This study concluded that the Confederation’s budget was generally countercyclical during periods of excessive aggregate demand, but also that the Confederation undertook procyclical budget cuts during recession periods. On the other hand, lower government levels behaved procyclically during periods of excessive aggregate demand, but these lower levels pursued expansionary countercyclical policies during recession periods. The net result of this un-coordinated mix of fiscal policies at different government levels was found to be somewhat procyclical. In an OECD study of fiscal policy in small OECD countries during 1970-83, Chouraqui and Montador (1983) found that fiscal policy in Switzerland was procyclical in the first half of the 1970s, but it moved to neutral or countercyclical stances thereafter.32 A study by the Federal Office for Economic Policy (1986) reached a similar conclusion—fiscal policy during the 1970s shifted from pro- to countercyclical postures.33 Ammann (1995) provides a comprehensive review of estimates of structural balances at different government level and draws particular attention to the potentially destabilizing role of income tax collection lags.34

C. Fiscal Policy Rules

52. The actual or observed budget balance (bt can be decomposed into two unobservable components, the structural budget balance (bst) and the cyclical budget balance (bct):

where the actual balance is expressed as a ratio to nominal GDP, while the structural and cyclical balance components are both expressed as ratios to nominal potential GDP.35 The cyclical budget component captures the automatic or built-in response of the budget to cyclical output fluctuations:

The parameter α measures the automatic response of the balance-GDP ratio to a 1 percentage point change in the cyclical output gap (GAPt), which, for expositional simplicity, is assumed to occur without lagged responses, due, for example, to tax collection lags. While the structural balance component abstracts, by construction, from the built-in budgetary responses to cyclical output fluctuations, discretionary fiscal policy actions could lead to systematic co-movements between the structural balance and the cyclical output gap:

where the parameter γ measures the discretionary response of the budget to cyclical output movements. For expositional simplicity, the discretionary fiscal policy response is also assumed to take place without lag, and Ψt captures the component of the structural balance that is unrelated to cyclical output fluctuations. Inserting equations (2) and (3) in (1) gives a generic fiscal policy rule linking budget balances and the business cycle:

53. This framework is useful to highlight various parameter restrictions imposed by four specific fiscal policy rules:36

Balanced budget rule: γ = -α and Ψt = 0.

Structural balanced budget rule: Ψt = 0 and γ = 0.

Countercyclical budget rule: γ > 0.

Maastricht budget deficit limit rule: Ψt ≥ -0.03 - (α+γ)GAPt.

The balanced budget rule imposes the requirement to fully off-set the operation of automatic fiscal stabilizers by procyclical discretionary fiscal policy actions. By contrast, a structural balanced budget rule allows for the full operation of automatic fiscal stabilizers. The countercyclical budget rule seeks to reinforce the operation of automatic fiscal stabilizers. Finally, the Maastricht budget deficit limit rule imposes a ceiling on the maximum (general government) deficit once adjustments are made for automatic fiscal stabilizers and cyclical discretionary policy actions.

54. Three additional considerations regarding fiscal policy rules appear to be noteworthy. First, the different levels of government may follow different rules, e.g. lower government levels may follow balanced budget rules, while the central government allows automatic fiscal stabilizers to operate. Thus, the behavior of general government finances will reflect a (weighted) mix of fiscal policy rules, and it can be of interest to examine the contributions of different government levels to the cyclical behavior of the general government balance, a consideration of considerable interest in the case of Switzerland. Second, some of the above fiscal policy rules can be combined to yield a more complex rule, e.g. a combination of the structural balanced budget rule and the countercyclical budget policy rule, as in the case of the constitutional balanced budget amendment presently under discussion in Switzerland (see Section E for details). And third, normative discussions of macroeconomic policies have often emphasized the distinct roles of automatic stabilizers and discretionary policy actions in a well-designed stabilization framework.37 Automatic stabilizers are considered to be best suited to counteract economic shocks that are small, frequent, and difficult to identify. By contrast, discretionary policy actions are considered most effective in the case of large shocks that are tied to events whose sources can be readily identified.38 If fiscal policy would be set according to these broad norms, the simple fiscal policy rules set out above could not work well as an empirical description of the behavior of fiscal policy over the business cycle. In particular, the value of the parameter γ would depend on the size of the shocks to the output gap, implying a “switching” fiscal policy rule, where the size of the budget balance responds discontinuously to changes in the state of economic activity.

55. Within the framework of equations (1) to (4), a two-stage approach to gauging the size of automatic and discretionary fiscal responses to cyclical output fluctuations can be developed. In the first stage, the automatic fiscal response parameter (a) can be derived from estimates of cyclical budget elasticities conventionally employed to construct structural budget balances (see Annex for details). In the second stage, the estimates of the automatic response parameters are used to derive estimates of the structural budget balance series, which can then be employed to estimate a version of equation (3) using OLS regressions to derive an estimate of the discretionary fiscal response.

56. The results from this two-stage estimation exercise are, however, subject to several caveats. As cyclical output gap and discretionary fiscal actions could occur at the same time, OLS regression could be subject to simultaneous equation bias and the use of an instrumental variable regression technique may be preferable. However, in addition to the use of statistical regression techniques to estimate the discretionary response to the cycle, alternative, less formal methods, can be utilized to gauge the relationship between discretionary fiscal actions and the cycle. Indeed, there are several reasons why a less formal approach to interpreting the data may also be useful in this context: First, measurement errors in the unobserved fiscal balance components, which could be particularly large in the case of Switzerland given the considerable uncertainties regarding estimates of the output gap, could distort the statistical results. Second, fiscal policy could change over time, leading to shifts in the parameter γ. Third, fiscal policy could be more responsive to “large” cyclical output gaps and/or react asymmetrically to positive and negative cyclical output gaps, thus invalidating the assumptions underlying conventional regression approaches.

D. Estimates of the Cyclical Responsiveness of Budget Balances

57. Cyclical budget elasticities capture the automatic response of a nominal revenue (or expenditure) component to a 1 percentage point change in the cyclical output gap.39 As shown in the annex, cyclical budget elasticities can be used to estimate automatic cyclical response parameters. Estimates of the automatic responses of general government finances to cyclical output fluctuations are summarized in Table II-1. The first part of the table provides parameter estimates for revenue elasticities and response parameters. The revenue elasticity estimates are drawn from work by the Federal Finance Administration (FFA), by Ammann (1995), and staff estimates. The overall elasticity of general government revenue is estimated at 1.1, close to the overall general government revenue elasticity estimates for other industrial countries.40 Using average revenue-GDP ratio estimates for the period 1975-95 and the conversion formulae in the annex, the overall revenue elasticity yields an estimate of the automatic revenue increase (including lagged effects) of 0.37 in response to a 1 percentage point increase in the cyclical output gap. However, in Switzerland, due to income tax collection lags, roughly one third of revenue collections occurs with a lag of 2-4 years. Thus,

Table II-1.Switzerland: Automatic Responses of General Government Finances to Cyclical Output Fluctuations
Elasticity Parameter Estimates 1/Response Parameter Estimates 2/
CurrentLagged 3/TotalCurrentLagged 3/Total
Total revenue0.950.161.110.250.120.37
Income tax0.001.201.200.000.120.12
Other direct taxes1.200.001.200.060.000.06
Social contributions0.700.000.700.070.000.07
Turnover tax/VAT1.300.001.300.040.000.04
Customs duties0.800.000.800.040.000.04
Other taxes1.000.001.000.010.000.01
Other revenue1.000.001.000.030.000.03
Total expenditure-0.070.00-0.07
Unemployment benefits 4/-0.070.00-0.07
Other expenditure0.000.000.00
Overall fiscal balance0.320.120.44
Confederation0.160.020.18
Cantons0.070.050.12
Communes0.020.050.07
Social Security Funds0.070.000.07
Sources: Federal finance Administration; Ammann (1995); and staff estimates.

Percentage point change in nominal revenue in response to a 1 percentage point change in the cyclical output gap.

Percentage point change in the ratio of fiscal aggregate to GDP in response to a 1 percentage point change in the cyclical output gap. See the chapter’s appendix for a description of the calculation of parameter estimates.

Lagged response refers to the average of output gaps in t-2 and t-3 for even income tax collection years and to the average of output gaps in t-3 and t-4 for odd income tax collection years.

Response parameter estimates are based on data covering the time period 1992-96.

Sources: Federal finance Administration; Ammann (1995); and staff estimates.

Percentage point change in nominal revenue in response to a 1 percentage point change in the cyclical output gap.

Percentage point change in the ratio of fiscal aggregate to GDP in response to a 1 percentage point change in the cyclical output gap. See the chapter’s appendix for a description of the calculation of parameter estimates.

Lagged response refers to the average of output gaps in t-2 and t-3 for even income tax collection years and to the average of output gaps in t-3 and t-4 for odd income tax collection years.

Response parameter estimates are based on data covering the time period 1992-96.

58. the automatic contemporaneous increase of the general government revenue-GDP ratio to a 1 percentage point is estimated at only 0.25, while the remaining automatic responses occurs with considerable lag.

59. As regards the automatic cyclical responsiveness of spending, only unemployment benefits are assumed to respond to cyclical fluctuations. However, the coverage and generosity of the unemployment benefit insurance system in Switzerland have expanded considerably since the system was inaugurated in 1977, and the observed unemployment rate has undergone a clear structural break at the beginning of the 1990s (see also Chapter III on the Phillips curve). As a consequence, the reported estimate of the automatic responsiveness of unemployment benefits of -0.07 is based only on data covering the period 1992-96.41

60. Combining the automatic response estimates for general government revenue and expenditure yields an estimate of 0.44 (including lagged effects). This response parameter estimate is significantly smaller than the estimated total response parameters for most other industrial countries, in particular the member countries of the European Union (Table II-2). The size of automatic fiscal stabilizers in Switzerland appears to be constrained by the comparatively small size of the public sector, as measured by the ratio of general government revenue to GDP (Figure II-3). For example, the average general government revenue to GDP ratio in Switzerland during the period 1990-96 amounted to 35¼ percent of GDP, as compared with an average of 45 percent of GDP for the European Union. However, this apparent “scale disadvantage” of Switzerland largely reflects different institutional characteristics, particularly as regards the social security sector. In Switzerland, the revenues of the funded occupational pension system (“second pillar”) and the health care insurance funds are excluded from the data on general government, although these funds are subject to strict government mandates. The reason for the exclusion is that these funds are operated within the private sector. Total revenue of the two funds from employer and employee contributions during 1990-96 amounted to some 10 percent of GDP, thus bridging the gap between observed revenue-GDP ratios in the European Union and Switzerland. Assuming the excluded contributions amount to about 10 percent of GDP and further assuming a cyclical elasticity for contributions of 0.70, including these mandatory contributions would increase the total response parameter by 0.07, bringing the total response parameter (0.51) closer to the average value calculated for the European Union (0.59).

Table II-2.Switzerland: Automatic Responses of General Government Balances to Cyclical Output Fluctuations in Selected Industrial Countries
Response Parameter Estimates 1/
CurrentLagged 2/Total
Switzerland0.320.120.44
United States0.320.040.36
Japan0.360.010.37
Germany0.490.050.54
France0.520.050.57
Italy0.340.030.37
United Kingdom0.300.440.74
Canada0.580.040.62
Spain0.600.040.64
Netherlands0.630.080.71
Belgium0.510.010.52
Sweden0.960.121.08
Austria0.470.020.49
Denmark0.580.160.74
Finland0.560.030.59
Ireland0.480.020.50
Australia0.370.130.50
New Zealand0.530.120.65
Memorandum item:
European Union average 3/0.480.110.59
Source: Staff estimates.

Automatic or built-in percentage point change in the ratio of general government balance to GDP in response to a 1 percentage point increase in real GDP. The calculation of the parameter estimates is described in the chapter’s appendix.

The lag for all countries except Switzerland is one year. For Switzerland, the lag refers to the average of years t-2 and t-3 in even tax collection years and the averages of years t-3 and t-4 in odd tax collection years.

Weighted average based on GDP weights.

Source: Staff estimates.

Automatic or built-in percentage point change in the ratio of general government balance to GDP in response to a 1 percentage point increase in real GDP. The calculation of the parameter estimates is described in the chapter’s appendix.

The lag for all countries except Switzerland is one year. For Switzerland, the lag refers to the average of years t-2 and t-3 in even tax collection years and the averages of years t-3 and t-4 in odd tax collection years.

Weighted average based on GDP weights.

61. However, the significant lags in income tax collection in Switzerland imply that only a portion (about 3/4) of the estimated response parameter represents effective automatic fiscal stabilizers, as current income tax collections are likely to reflect a phase of the business cycle that is asynchronous to current cyclical conditions. At the cantonal level, there are efforts underway to mitigate the problem of lagged income tax collections. In particular, the large canton of Zürich intends to base the assessment of income taxes on the current tax year from 1999 onwards. It is expected that other cantons, and perhaps the Confederation, will eventually follow the example of Zürich, thus strengthening considerably the operation of automatic fiscal stabilizers.

62. Moreover, shortening the long income tax collection lags would be worthwhile in order to mitigate the effects of inflation on the real value of tax revenue. For example, an average collection lag of three years and an annual inflation rate of 4 percent (the average Swiss inflation rate during 1971-95) reduces the real value of income tax collections—assuming a proportional income tax and no bracket creep—by about 11 percent, equivalent to about 1 percent of GDP given current income tax collections of some 10 percent of GDP.42

63. The plotted co-movements of the structural general government budget balance and the cyclical output gap provide a visual impression of fiscal stances over the business cycle (Figure II-5). The shaded areas in Figure II-5 indicate periods of procyclical fiscal stance, defined as years when output gap and structural balance moved into opposite directions. This (non-parametric) evidence is suggestive of extended periods of procyclical fiscal stance. In particular, the fiscal stance was procyclical in 17 out of 26 years during the time period 1971-96. Moreover, and more importantly, the fiscal stance appears to have often been procyclical close to business cycle peaks (early 1970s, 1976, and 1990) or troughs (1976, 1983).

Figure II-5.Switzerland: Fiscal Stance of General Government and the Business Cycle, 1970-96

Source: WEO database; and staff estimates.

1/ Shaded areas indicate periods of procyclical fiscal stance, defined as periods with opposite movements of output gap and structural balance.

64. As already indicated, attempts to obtain parametric estimates of the systematic relationship between discretionary fiscal actions and the cyclical output gap are subject to severe limitations. Nevertheless, the statistical results of regressing the first difference of the structural general government balance on the first difference of the current and lagged cyclical output gap is suggestive of a procyclical stance, although the statistical evidence is not significant at conventional significance levels (Table II-3).43 Moreover, running the same regressions for the different levels of government indicates that procyclical fiscal behavior is concentrated at the levels of the cantons and the communes and that it occurs with a one-year lag.

Table II-3.Switzerland: Discretionary Responses of General Government Finances to Cyclical Output Fluctuations 1/
Discretionary Response Parameter Estimates 2/
CurrentLagged 3/Total
Revenue-0.11(0.10)-0.19(0.09)*-0.30
Expenditure-0.03(0.10)-0.22(0.10)*-0.25
Balance-0.08(0.09)0.03(0.09)-0.05
Confederation-0.05(0.05)0.05(0.05)0.00
Cantons0.00(0.03)-0.01(0.03)-0.01
Communes-0.02(0.03)-0.04(0.03)-0.06
Social Security Funds-0.01(0.03)0.03(0.03)0.02
Source: Staff estimates.

Based on OLS regressions of first differences of structural fiscal aggregates on the first difference of the current and lagged cyclical output gap. Time range of all regressions is 1972-96.

Discretionary fiscal response to a 1 percentage point change in output gap. Numbers in parentheses are standard errors corrected for heteroscedasticity.

One-year lag.

Significant at the 5 percent significance level.

Source: Staff estimates.

Based on OLS regressions of first differences of structural fiscal aggregates on the first difference of the current and lagged cyclical output gap. Time range of all regressions is 1972-96.

Discretionary fiscal response to a 1 percentage point change in output gap. Numbers in parentheses are standard errors corrected for heteroscedasticity.

One-year lag.

Significant at the 5 percent significance level.

E. A New Fiscal Policy Rule for the Confederation

65. The proposed constitutional amendment (Schuldenbremse) to ensure budget balance at the Confederation level beyond the year 2001 envisages a combination of the structural balanced budget rule and the countercyclical budget policy rule.44 Also, the specific policy rule proposed in the consultation report (1995) on the constitutional amendment ties deviation from a balanced budget to the real GDP growth rate (μt). In particular, the proposed fiscal policy rule for the Confederation could take the form:45

The specific fiscal policy rule (5) imposes a strict balanced budget when real GDP growth is in a range between 0.5 and 1.8 percent. The rule allows the combined deficit effects of the automatic and discretionary responses to equal the change in the real output gap for real growth below 0.5 percent. Finally, it aims for surpluses of one half of the change in the cyclical output gap for real growth above 1.8 percent. Thus, outside the growth range between 0.5 and 1.8 percent, this rule allows for the operation of automatic fiscal stabilizers as well as countercyclical discretionary responses. For example, if economic growth in a given year amounts to minus 1 percent, the rule would allow the Confederation to run a deficit of 1.5 percent of GDP. In this particular example, and assuming potential output growth would be 1.5 percent, the output gap would widen by 2.5 percent and the operation of automatic fiscal stabilizers (using a contemporaneous response coefficient of 0.16) would increase the Confederation’s deficit by about 0.4 percent of GDP, and the fiscal policy rule would allow for a countercyclical discretionary response equivalent to 1.1 percent of GDP.

66. Budgetary planning and execution of the Confederation budget in year t under the proposed fiscal policy rule would be based on the following dates and actions (see consultation report (1995, p. 37):

DateAction
April t-1Federal Council proposes a target value for the Confederation balance as a percent of GDP based on the projection of real GDP growth for year t.
June t-1Parliament approves a target value for balance (which need not coincide with the Federal Council’s proposal).
October t-1Federal Council puts forward a budget for year t consistent with the target value passed by Parliament.
December t-1Parliament approves the budget, with the option, however, to deviate from the target value.
tExecution of budget.
August t+1In case outcome deficit exceeds the target value, the Federal Council orders budget cuts effective in year t+2.

As regards the compensation for the deviation from the target in year t, the Federal Council would have some flexibility in determining the size of budget savings depending on the state of the business cycle.

67. Figure II-6 reports the results of a counterfactual simulation of the path of the Confederation’s budget deficit and debt under the proposed new fiscal policy rule during 1976-96. This counterfactual simulation assumed that budget planning and the specification of the deficit target value in year t-1 were based on the GDP growth rate that actually materialized in year t (perfect foresight assumption). For the chosen time period, the deficit and debt path under the proposed fiscal policy rule represents a clear improvement on the actual outcomes. In particular, the Confederation’s fiscal policy would have been restrictive during the boom period in the second half of the 1980s and the level of the Confederation’s debt would have been stabilized around a level of about 14 percent of GDP. At the same time, however, the experience of the first half of the 1990s also suggests that the proposed rule would have implied a procyclical fiscal stance during this period. This reflects the conditioning of budgetary policies under the proposed rule on GDP growth rates instead of the size of the output gap. The rule requires policy makers to revert to a balanced budget once GDP growth reverts back to the “neutral range” between 0.5 and 1.8 percent, even though a prolonged period of low GDP growth could have yielded an expanding output gap.

Figure II-6.Switzerland: Simulation of Confederation Finances under Proposed New Fiscal Rule, 1976-96

(In percent of GDP)

Source: WEO database; and staff estimates.

68. Besides the issue of a likely procyclical behavior of fiscal policy during periods of persistently slow or rapid GDP growth, the proposed fiscal policy rule is subject to two additional caveats. First, the assumption of perfect foresight for GDP growth is clearly unrealistic, as illustrated by a comparison of actual GDP growth and projected GDP growth rates underlying the Confederation budgets during 1985-96 (Figure II-7). While projected real GDP growth has been consistently in the “neutral range” of 1-2 percent, actual GDP growth has fluctuated widely around the projections.46 Using forecasted instead of actual GDP growth for calculating the path of the Confederation’s deficit under the proposed fiscal policy rule shows that budget proposals would have closely followed a balanced budget rule. As the proposed rule would not have stifled the operation of automatic fiscal stabilizers in the face of lower-than-anticipated GDP growth, the Confederation’s fiscal policy would in practice follow a structural balance rule. Indeed, in view of the fact that annual budgets are usually planned and approved at least one year in advance, it appears to be difficult to formulate an operative fiscal policy rule that effectively counters large cyclical disturbances that may occur well after the budget’s approval. Second, the proposed fiscal rule could imply large and abrupt changes in spending under present fiscal institutions, which would likely imply inefficient spending patterns. In particular, given the long lags in income tax collections and the practice to base income tax assessment on two-year averages of incomes, the Confederation’s income tax collections are usually high in even years and low in odd years.

Figure II-7.Switzerland: GDP Forecasts and Fiscal Rules, 1985-96

Source: WEO database; and staff estimates.

69. In summary, the proposed constitutional Schuldenbremse amendment is likely to improve on present fiscal policy behavior at the Confederation level. However, given the limited accuracy of GDP growth forecasts for Switzerland, the rule would likely correspond to a version of a balanced structural budget rule. Thus, it is unlikely that the rule would allow for an effective countercyclical fiscal policy in the face of large shocks. Moreover, in periods of a persistently slow or rapid GDP growth, the rule could lead to procyclical fiscal stances. Finally, implementation of the rule would need to be preceded by reforms that assure a more steady flow of tax collections, in particular in view of the present idiosyncratic income tax collection system.

APPENDIX I: Estimation of Automatic Cyclical Response Parameters

70. This appendix outlines an approach for quantifying the automatic response of budget balances, expressed as a percent of GDP, to a 1 percentage point change in the cyclical output gap.47 The actual or observed budget balance (Bt) is assumed to be the sum of two unobservable components, the structural balance (BSt) and the cyclical balance (BCt):

71. The structural balance is defined as the difference between structural nominal revenue (RSt) and structural nominal expenditure (GSt):

while the cyclical balance is defined as the difference between cyclical nominal revenue (RCt) and cyclical nominal expenditure (GCt):

72. To account for the size of the economy, the actual nominal balance is usually expressed as a ratio to nominal output (YNt) while the structural nominal balance is expressed as a ratio to nominal potential output (YNPt). As the structural and cyclical balance components need to add up to the actual balance component, the use of different scaling conventions for the actual and structural balances raises the issue of how to scale the level of cyclical balance. In the following, the cyclical balance is expressed as a ratio to nominal potential output. Thus, the decomposition of the actual balance into structural and cyclical components, all expressed as ratios to output, is given by:

where GAPt is the cyclical output gap, defined as GAPt=(YNt-YNPt)/YNPt.48 The term (Bt/YNt)GAPt represents a “nuisance term” introduced by the use of different scaling conventions for the actual and structural balances. This nuisance term will be small if the gap and/or the actual balance ratio are small, and the approximation given in equation (A.4) is therefore likely to work well under most circumstances. Using the same scaling conventions for the decomposition of revenue and expenditure gives:

73. By contrast to the decomposition of the balance in equation (A.4), the nuisance terms on the right-hand sides of equations (A.5) and (A.6) are likely to be large if the cyclical output gap deviates from zero, and as a consequence, to isolate cyclical revenue and expenditure ratios, it is paramount to take account of the nuisance terms when interest focusses on isolating the cyclical components of revenue and expenditure.

74. There are two possible approaches to determine the decompositions (A.5) and (A.6). The first and traditional approach determines the structural components on revenue and expenditure and derives the cyclical components as residuals. In particular, the OECD uses estimates of the cyclical elasticities of revenue and expenditure to calculate the structural revenue and expenditure components:

where ϵR, δR, ϵG, and δG denote current and lagged cyclical elasticity estimates for revenue and expenditure, respectively.49 From equations (A.7) and (A.8), the cyclical revenue and expenditure components can be directly calculated as a function of the current and lagged cyclical output gap:

where (A.7)′ and (A.8)′ are based on the approximation (1+GAPt)≈GAPt for GAPt small.50

75. The built-in flexibility parameter approach determines the cyclical revenue and expenditure ratios from:

76. In equation (A.9), the cyclical built-in flexibility parameters for revenue, αR and βR, describe the percentage point change in the cyclical revenue ratio for a 1 percentage point change in the current and lagged cyclical output gap, respectively. Similarly, the cyclical built-in flexibility parameters for expenditure, αG and βG, describe the cyclical responsiveness of expenditure. From equations (A.9) and (A.10), the cyclical balance ratio may be compactly written as

where α=αRG and β=βRG. Thus, the sum of the cyclical response parameters α and β in equation (A.11) provides a concise answer to the question: What is the total automatic response of the observed budget balance ratio to a 1 percentage point change in the cyclical output gap?

77. Comparing the equation pairs (A.7)′ and (A.9) and (A.8)′ and (A.10), respectively, shows that elasticity and built-in flexibility parameters are linked by the linear approximations:

where μR and μG denote averages of (Rt/YPNt) and (Gt/YPNt) over time, respectively. Thus, the approximations (A.12) allow to convert cyclical elasticity parameter estimates into cyclical built-in flexibility parameter estimates and vice versa.

78. While the application of these equations to the revenue side of the budget is relatively straightforward, the calculation of automatic cyclical response parameters for the expenditure side requires additional considerations. Following convention, only unemployment insurance benefits are assumed to respond automatically to cyclical output fluctuations. Unemployment insurance benefits (UIBt) are conventionally cyclically adjusted by multiplying actual unemployment benefits by the ratio between the cyclically-adjusted unemployment rate (UNt) to the actual unemployment rate (Ut)

79. The cyclical rate of unemployment (Ut-UNt) is assumed to be linked to current and lagged cyclical output gap by an Okun relationship

80. Using (A.14) to isolate an expression for the natural rate of unemployment and inserting the expression in (A.13), dividing by YNPt, subtracting the result from (UIBt/YNt), and re-arranging gives:

81. Expanding the right-hand side of (A.15) around GAPt=GAPt-1=0, (UIBt/Yt) and UtU, and ignoring terms of order higher than one, results in equation:

where the cyclical response parameters αUIB and βUIB capture the contribution of unemployment benefits to automatic stabilizers. Finally, the cyclical response parameters αUIB and βUIB are determined as:

where μUIB and μU denote average unemployment benefit-to-GDP ratios and the average unemployment rate, respectively. If only unemployment benefits are assumed to respond automatically to cyclical output gaps movements, the cyclical response parameters for expenditure are given by:

and the approximations (A.12) can be used to recover the overall cyclical elasticities of expenditure.

26Prepared by Albert Jaeger.
27See, for example, Paul Bernd Spahn, “Switzerland,” in: Fiscal Federalism in Theory and Practice, edited by Teresa Ter-Minassian, pp. 324-41, International Monetary Fund (Washington, D.C.: 1997).
28Five cantons (Basel-City, Geneva, Jura, Neuchâtel, and Solothurn) use tax periods of one year.
29This is an estimate of the average duration of business cycles in ten industrial countries including Switzerland during the period 1948-83. Duration is measured from cyclical trough to trough. See Victor Zarnowitz, “The Regularity of Business Cycles,” in: Business Cycles: Theory, History. Indicators, and Forecasting, by Victor Zarnowitz, pp. 232-64 (Chicago: 1992).
30The cyclical output gap estimates are based on the production function methodology outlined in last year’s paper on Selected Issues (SM/97/23).
31See A. Wagner, “Die Auswirkungen der orrentncnen Haushalte auf den Konjunkturverlauf in der Schweiz,” Schweizerische Zeitschrift für Volkswirtschaft und Statistik, pp. 17-30, 1973.
32See Jean-Claude Chouraqui and Bruce Montador, “Fiscal Policy in the Small OECD Countries Since the Early Seventies,” Schweizerische Zeitschrift für Volkswirtschaft und Statistik, pp. 259-83, 1985.
33See Federal Office for Economic Policy, Zur Messung finanzpolitischer Impulse und struktureller Haushaltsdefizite, Study No. 10 (Bern: 1986).
34See Yves Ammann, Le budget de plein emploi--un rèexamen, Study No. 20, Federal Office for Economic Policy (Bern: 1995).
35For more background on the formal derivations of equations see the annex to this chapter. As shown in the annex, equation (1) represents only an approximation due to the different GDP scaling conventions on the left and right hand side of equation (1).
36See the recent Board Paper on “Fiscal Policy Rules” (SM/97/175) for a discussion of different types of fiscal rules.
37See, for example, Olivier J. Blanchard and Mark W. Watson, “Are Business Cycles All Alike?,” in: The American Business Cycle: Continuity and Change, edited by Robert J. Gordon, Studies in Business Cycles Volume 25, pp. 123-79 (Chicago: 1986).
38Descriptions of Switzerland’s monetary policy framework often emphasize that the Swiss National Bank’s (SNB) monetary policy rule is consistent with these broad norms. Under “normal circumstances,” the SNB aims at meeting a medium-term target for monetary base growth based on an implicit inflation target of about 1 percent, which is held to induce automatic countercyclical movements of interest rates. However, in case of large shocks, in particular of large exchange rate disturbances, monetary policy reserves the option of discretionary deviations from the medium-term money target rule. See, for example, Georg Rich, “Die schweizerische Teuerung: Lehren für die Nationalbank,” Quartalsheft No. 1, March 1992, pp. 73-88 (Swiss National Bank: 1992).
39See Jean-Claude Chouraqui, Robert P. Hagemann, and Nicola Sartor, “Indicators of Fiscal Policy: A Reexamination,” OECD Economics and Statistics Department Working Paper No. 78 (Paris, April 1990) for a discussion of estimates of budget elasticities in most industrial countries.
40See, for example, the revenue elasticity estimates for the major industrial countries reported in Annex I of the October 1993 World Economic Outlook, Table 23, which average to about 1.0.
41This estimate is based on the assumptions of an Okun’s law coefficient of 3 linking fluctuations of cyclical unemployment and cyclical output—a 3 percentage points increase in the cyclical output gap decreases cyclical unemployment by 1 percentage point; an estimate of the natural rate of unemployment of about 3 percent; and an average unemployment benefits-GDP ratio of about 1 percent during 1992-96. The formulae underlying the calculation of the cyclical response coefficients for unemployment benefits given these data inputs are provided in the Annex.
42For a discussion of the impact of inflation on the real value of tax revenue given tax collection lags see Vito Tanzi, “Inflation, Lags in Collection, and the Real Value of Tax Revenue,” Staff Papers, International Monetary Fund, Vol. 24, pp. 154-67 (Washington, D.C.: March 1977). It is noteworthy that Tanzi’s empirical analysis only considers tax collection lags of up to 2 years, while the Swiss tax system allows for lags of up to 4 years.
43Employing alternative regression strategies, for example running the regression in levels, instead of first difference form, yields more conspicuous evidence but still statistically insignificant evidence of a procyclical fiscal stance.
44See consultation report Vernehmlassungsbericht zur Schuldenbremse (Bern: October 1995).
45This is the so-called “balance-based rule” of the consultation report. The report also discusses an alternative “expenditure-based rule fiscal rule,” which, however, would not impose explicit restrictions on the budget balance but would only restrict the expenditure-GDP ratio at the Confederation level to remain within specified boundaries.
46Private sector and other official forecasts also persistently underprojected real GDP growth during the second half of the 1980s and persistently overprojected real GDP growth during the first half of the 1990s.
47This annex extends Annex I, pp. 99-103, of the October 1993 World Economic Outlook.
48Equation (A.4) is derived by dividing equation (A.1) by YNPt and noting that (Bt/YNPt) can be written as (Bt/YNt)(YNt/YNPt)=(Bt/YNt)(1+GAPt)
49Equations (A.7) and (A.8) adopt two expository simplifications. First, the calculation of the structural balance may use disaggregated revenue and expenditure. Second, the operation of automatic fiscal stabilizers is restricted to a lag of one period.
50Equations (A.7)′ and (A.8)′ illustrate why ignoring the bias terms in equations (A.5) and (A.6) can lead to misleading inferences. For example, assuming that cyclical revenue is unit-elastic with respect to the cyclical output gap (ϵR=1, δR=0) and inserting (A.7)′ in (A.5), it follows that (Rt/YNt)=(RSt/YPNt), i.e. the bias term exactly off-sets the cyclical revenue ratio, independently of the size of the cyclical output gap. It would be misleading to conclude that the cyclical component of revenue is zero and that revenue therefore do not contribute to automatic fiscal stabilizers.

Other Resources Citing This Publication