Chapter

II A Brief Review of Alternative Measures for Assessing the Stance and Thrust of Fiscal Policy

Author(s):
Ahsan Mansur, Richard Haas, and Peter Heller
Published Date:
May 1986
Share
  • ShareShare
Show Summary Details

It has long been recognized that the actual budgetary position of a government may be a misleading indicator of the thrust of fiscal policy, because it is not clear whether changes in that position are the cause or the result of changes in economic activity. In particular, it is important when analyzing fiscal policy to distinguish between certain cyclical factors that have a transitory effect on the actual budget balance and the effects of changes in policy or structural changes in the economy that may have a more durable impact on the budget balance. Over the last two decades, this concern has led to the development of various alternative techniques for adjusting the fiscal accounts to yield a more accurate measure of the fiscal policy stance. A principal objective of this paper is to evaluate the quantitative measure that the Fund is using to appraise the economic thrust of fiscal policy. This section briefly surveys the various quantitative indicators of the stance and thrust of fiscal policy that have been developed by the various national and multinational agencies that routinely provide assessments of fiscal policy.

Review of Existing Techniques

There are two basic approaches used to deal with the inadequacies of the change in the actual observed budget balance as a measure of the expansionary or contractionary influence of fiscal policy. One approach calculates a measure of the total impulse or initial stimulus to aggregate demand arising from fiscal policy from whatever source, whether discretionary or otherwise, during a given period. This is the approach used in the Fund. Conceptually it treats any change in the actual budget which is not transitory, in a cyclical sense, as imparting a “fiscal impulse.” This measure takes account of the effects of both discretionary policy changes and those automatic fiscal stabilizers that arise from any differences of the income elasticities of revenue and expenditure from unity (as opposed to those arising from the effects of the cycle).

In this approach, one begins by making a cyclical adjustment—that is, adjusting for the fact that the economy is generally not operating at a “normal” level of resource utilization. Further adjustments may be made if it can be argued that the fiscal impulse calculated in this manner may not give rise to altered economic activity. A correction for the effect of inflation on government interest payments is an example of this type of correction. Other corrections can be made if there is believed to be an improper specification of some element—for example, unemployment transfers—in the budget.

The second approach focuses on classifying budgetary items as endogenous or exogenous. Here the idea is to measure the discretionary change in fiscal policy (i.e., exogenous changes in policy instruments), not the total thrust of policy. This differs from the first approach by not including any effect of automatic stabilizers in the fiscal impulse measure. The idea is to correct, as best one can, for all endogenous effects on the budget. In principle, this could include not only changes in income but also changes in the prices of goods and services, the interest rate, and the exchange rate. The argument is that these variables give rise to feedback effects that trigger changes in automatic expenditures and receipts that should not be included in a measure of discretionary policy. This approach is used by all of the measures discussed below except those used by the Fund and by the European Community (EC).

The first approach does not draw a distinction between a change in aggregate demand that results from a discretionary budget decision made today and one that results from past policies that generated automatic fiscal stabilizers. The second approach does. Consider, for example, two countries alike in all respects save that one has automatic fiscal stabilizers while the other has none. Each is striving to reach the same real output target in the face of a deflationary shock. The country with the automatic stabilizers will show a smaller discretionary (exogenous) change in its budget than the country without them. Yet, in some sense, fiscal policy has been equally expansionary in both countries.

The methods outlined below recognize that the budgetary position is partially a function of the income level and make some sort of adjustment for this fact. Some studies have also corrected for inflationary effects, but it is fair to say that no single measure is adjusted for all sources of bias. This is not to suggest that there is an ideal measure of fiscal impulse that should be calculated. Measurement of discretionary fiscal policy is a multidimensional problem, which does not lend itself well to description by a single number. Indeed, to give just one example, the empirical measure of the fiscal impulse that is most useful in evaluating the effects of short-term policy on real income may be different from the measure needed for other uses, such as evaluating the effect of budgetary changes on financial markets. It is important to bear this qualification in mind when appraising the different fiscal impulse measures and to recognize precisely what a given measure purports to measure. Furthermore, it must be borne in mind that the concept of the fiscal impulse focuses on the budget alone, not on its final effect. In particular, the more integrated the international economy, the more fiscal policy changes in one country will spill over into neighboring countries.

A brief description of the measure used by the Fund in the World Economic Outlook exercise and of several other commonly used measures of the fiscal impulse follows, with particular attention paid to the theoretical advantages and disadvantages of each measure, as well as the practical problems of implementing them.4

GCEE-IMF Measure

The measure currently used by the Fund was originally developed by the German Council of Economic Experts (GCEE) and is described in detail by Dernberg (1975).5 It starts by establishing a base year in which actual and potential real income are judged to be the same. The “cyclically neutral budget” is derived from the actual budget by assuming that nominal tax revenues are unit elastic with respect to actual nominal income, and government expenditures are unit elastic with respect to potential output valued at current prices. Thus, government expenditure is termed cyclically neutral if it increases proportionately with increases in nominal potential output. A more-than-proportionate increase from whatever source (e.g., discretionary policies or the effects of inflation on expenditure) is defined as expansionary; a less-than-proportionate increase is defined as contractionary. A similar statement is true for changes in revenue with respect to changes in actual nominal output; a more-(less-) than-proportionate change is classified as contractionary (expansionary), regardless of the source of the change in revenue (e.g., a discretionary tax increase or a progressive tax structure). The cyclically neutral budget is calculated under the assumption of unitary elasticities of expenditure and revenue with respect to potential and actual output, respectively, not because the assumption is realistic but because defining the reference (i.e., cyclically adjusted) budget in this fashion has the effect of allocating the contribution of automatic stabilizers to the fiscal impulse.

Equation (1) shows the decomposition of the actual budget balance B (= T – G) in the Fund measure:

where

  • t0 = T0/Y0, the revenue ratio in the base period

  • g0 = G0/Y0, a base-year expenditure ratio

  • Y = actual output in nominal prices

  • Yp = potential output in nominal prices

  • T = government revenues

  • G = government expenditures

and FIS is a measure of the fiscal stance.

The subscript “0” refers to the base-year values of any variable. As defined, an actual deficit in excess of the cyclically neutral deficit is deemed expansionary, relative to the base-year fiscal stance, and the fiscal stance measure is positive. The base year is chosen to be a period when actual and potential output are assumed to be roughly equivalent. The budget deficit can be decomposed into three elements: the “base-year surplus” (the first term in equation (1)), the cyclical component (the second term), and the fiscal stance (FIS). Two elements, the base-year surplus and the cyclical component, can be merged to define the “cyclically neutral budget.” Equation (1) can be rewritten as

Taking the first difference of the fiscal stance measure, one obtains an absolute measure of the fiscal impulse6, FI

In relating the fiscal impulse to GDP, two alternative methodologies can be used: (a) calculating FI/Y = Δ FIS/Y directly, or (b) taking the first difference of the ratio of the fiscal stance to GDP—namely, A (FIS/Y). The second method is used in the Fund’s World Economic Outlook exercise and in the analyses that follow, since it permits one to evaluate whether the fiscal position has become more or less expansionary or contractionary in a given period. The ratio of FIS/Y in a period suggests how the fiscal policy stance has changed since the base period (when, by definition, fiscal policy is assumed to be neutral). The impulse in a given period reflects the change in fiscal stance, and it would appear reasonable for the stance in any year to be normalized by output in that year.

In effect, if the absolute level of the fiscal stance, or the extent to which the cyclical and actual fiscal balances diverge, remains unchanged as a share of GDP across two periods, it would not be reasonable to assert that an additional impulse has been imparted owing to fiscal policy.7 If the thrust of fiscal policy has become more expansionary (contractionary) relative to the previous year, the fiscal impulse measure will be positive (negative) in sign.

Calculation of the fiscal impulse is primarily intended as a first step in the analysis of fiscal policy. It is not an indicator of the full impact of fiscal policy in the short or medium term, nor does it measure the contribution of the government sector to the growth in GDP.8 At best, it provides a measure of the magnitude of the initial stimulus to aggregate demand arising from the net effects of fiscal policy in a given period. Whether a given stimulus actually has an expansionary or contractionary effect on real output or prices will depend upon the degree of capacity utilization, the effect on the private sector of how a deficit (or surplus) is financed, the stance of monetary policy, the structure of marginal expenditure and revenue, and other such factors. One should also note that the fiscal impulse measure is designed to determine the magnitude of the change in budgetary stance—that is, whether budgets are moving toward expansion or restriction, rather than what the effect of the budget is. Thus a deflationary budget which becomes less deflationary and an expansionary budget which becomes more expansionary will both yield a positive fiscal impulse.

A principal advantage of this approach is the simplicity of the calculation. One needs only the changes in actual and potential output, a set of base-year expenditure and tax-to-income ratios, and the change in the actual budget balance to calculate the fiscal impulse. Notwithstanding these very modest data demands—relative to other techniques—some implied costs should be noted. First, the elasticity of tax revenue with respect to output is, as an empirical matter, not equal to unity in most countries and is likely to vary with the rate of inflation, reflecting inter alia the effects of progressivity and administrative lags in collection. The same is true for government expenditure. By defining a baseline, normal case in this fashion, the Fund method implies that the effect of automatic stabilizers that arise owing to differences from unity in the revenue and expenditure elasticities with respect to nominal GDP should be included in the fiscal impulse measure.

Second, by calculating the fiscal impulse residually (i.e., by purging the actual budget of all cyclical effects and taking first differences of the remainder to obtain the fiscal impulse), the fiscal impulse will include not only the effect of changes in fiscal policy and the subsequent effect of automatic stabilizers but also the effect of structural changes in the economy.

Third, this method suffers (as do the Organization for Economic Cooperation and Development (OECD) and full employment balance (FEB) techniques) from the so-called balanced-budget-multiplier problem: that is, the measure implicitly assumes that equal increases in government spending and taxes exert no additional stimulus on aggregate demand, whereas most conventional models implicitly assume that a change in government spending has a larger (and more direct) effect on income than an equivalent tax change.

Fourth, the Fund’s method only adjusts the budget for deviations of output from its potential level—a problem also encountered in the other techniques. The effects of prices, interest rates—both real and nominal—and the exchange rate are ignored. To measure accurately the thrust of changes in fiscal policy, the fiscal impulse should be adjusted for the effects of these variables whenever they have significant effects that can be included in the calculations.

The question often posed is whether a change over time in the base-year tax and expenditure ratios should be taken into account in the calculation of the fiscal impulse. In principle, one would not want to change the base-year parameters. The discretionary measures that underlie the change in such ratios are reflected in the impulse measure, which is precisely what one would want to happen. Since the discretionary change would also be reflected in the fiscal stance measure in all succeeding years, there would be no subsequent effects on the measure of fiscal impulse (which is the first difference of the fiscal stance measure for successive years). This is also appropriate. If one changes the base-year ratios, the fiscal impulse measure will change, but only in the years in which the shifts occur. If the calculations are made for the whole period using revised base-year ratios, the measure of fiscal stance will change, but the fiscal impulses will be unaffected. In effect, a shift in the base year is analogous to a change in the base year of an index.

OECD Measure

Of the remaining four measures, the OECD measure is the closest to that used by the Fund. It also emphasizes the differences between actual and potential output and is subject to many of the drawbacks outlined above. There are, however, two major differences in practice. First, the elasticities of cyclically neutral expenditure and revenue with respect to real output are not constrained to be unity in the OECD method. Rather, disaggregated information derived from simulations with the Interlink model9 is used to calculate estimates of cyclically neutral revenue and expenditure which may yield implicit elasticities different from unity (though the differences are not likely to be large). Thus, the OECD’s fiscal impulse measure is exclusive of automatic stabilizer effects, though not of the fiscal drag arising from inflation. It attempts to capture the combined thrust of discretionary shifts in expenditure and revenue policy and fiscal drag. Second, the OECD method effectively uses ratios of expenditure and revenue to potential and actual output, respectively, in the previous period in constructing its measure, rather than base-period values. Another difference between the results of the OECD’s Economic Outlook and the Fund’s World Economic Outlook exercises for current and prospective years is the estimates and forecasts for the various fiscal and economic aggregates; differences here can be quite significant.10

The change in the “discretionary,” or “cyclically corrected,” portion of the budget is the OECD’s measure of the fiscal impulse, Fl’, which is defined as

where

  • ΔGP = change in government expenditure arising from a change in policy

  • ΔYp = change in potential output level

  • ΔTp = change in tax receipts arising from a change in policy

  • γ = the expenditure elasticity with respect to Yp

  • ε = the tax elasticity with respect to Yp

In practice, this is calculated residually by subtracting the effect of built-in stabilizers from the actual budget11

where m denotes the marginal tax rate (net of changes in unemployment benefit expenditure) with respect to the divergence between actual and potential levels of output.

In its most recent Economic Outlook analyses, the OECD has also begun to take account of the effect of inflation on government interest expenditures.12 The argument for doing this, roughly put, is that a component of current nominal interest outlays may represent compensation for the declining real value of the nominal stock of outstanding bonds owing to inflation. Private wealth holders are assumed not to view this component of their interest income as income to be spent but merely as an accelerated repayment of principal. Thus, increased government debt-servicing costs brought on by higher inflation ought not to be included in the adjusted budget because they are neither a discretionary action on the part of the authorities nor a factor leading to higher real incomes. (See Section V.)

Inflation has other important effects on public sector real expenditure. These include increased entitlements triggered by price movements. There are also effects if government budgets are specified in nominal terms, as they are in most cases, and there are revenue effects if the tax system is progressive but not fully indexed.13 It may or may not be important to adjust for these in the calculation of the fiscal impulse, depending on whether the objective is to capture the total fiscal impulse arising from the budget or simply to measure that impulse derived from discretionary policy. In the former case, it is not desirable to make an adjustment; in the latter, it is.

The OECD approach requires more extensive data than the Fund method, since it requires estimates of government expenditure and revenue elasticities. One need not build a disaggregated macroeconomic model to obtain these, as the OECD does. They could, in principle, be derived from several reduced-form equations or from alternative estimation procedures (such as the Prest adjustment method of estimating tax elasticities).

Table 1 illustrates the different impulse statistics that may arise from the OECD and Fund methodologies. For some years and countries, the differences can be as large as 1 percent of GDP (e.g., France and Italy in 1984, or the United Kingdom in 1983). The differences may occur for several reasons: (i) differences in the estimates of the gap between potential and actual output (either owing to differences in the estimates of either actual or potential GDP, or in the estimates of both);14 (ii) differences in the estimates of the budget balance for the current and future years; and (iii) differences in the assumed elasticities of cyclically neutral revenue and expenditure to changes in real and potential output, respectively. It should also be noted that the OECD provides only estimates for general government, while the Fund provides estimates for both central and general government.

Table 1.Seven Major Industrial Countries: Differences Between Fiscal Impulse Measures of OECD and IMF and Between Their Assumed Measures of the Fiscal Balance and Output Gap, 1981–84(As percentages of GDP)
Country1981198219831984
Canada
Fiscal impulse measure
OECD−1.90.10.7
IMF−1.00.40.40.1
Fiscal balance: general government
OECD−1.1−5.3−5.7−5.1
IMF−1.1−5.3−5.9−5.0
Output gap
OECD2.310.510.07.9
IMF1.17.67.26.3
Difference between OECD and IMF output gaps1.22.92.81.6
United States
Fiscal impulse measure
OECD−0.91.30.50.7
IMF−0.21.10.70.6
Fiscal balance: general government
OECD−0.9−3.8−3.8−3.7
IMF−0.9−3.8−4.2−3.8
Output gap
OECD4.59.28.56.4
IMF3.07.46.94.9
Difference between OECD and IMF output gaps1.51.81.61.5
Japan
Fiscal impulse measure
OECD−0.6−0.1−1.1−1.1
IMF−0.3−0.6−0.4−0.2
Fiscal balance: general government
OECD−4.0−4.1−3.4−2.5
IMF−4.0−3.6−3.3−3.1
Output gap
OECD1.72.53.33.4
IMF−0.50.50.8−0.4
Difference between OECD and IMF output gaps2.22.02.53.0
France
Fiscal impulse measure
OECD1.0−0.20.2−0.7
IMF0.70.1−0.5−1.3
Fiscal balance: general government
OECD−1.9−2.6−3.4−3.8
IMF−1.8−2.6−3.1−2.9
Output gap
OECD3.03.45.58.1
IMF2.73.45.24.5
Difference between OECD and IMF output gaps0.30.33.6
Germany, Federal Republic of
Fiscal impulse measure
OECD−0.3−1.8−1.4−1.2
IMF−0.9−2.4−0.9−0.6
Fiscal balance: general government
OECD−3.9−3.5−3.1−2.1
IMF−3.9−3.5−2.9−1.9
Output gap
OECD4.27.38.17.8
IMF1.95.05.94.1
Difference between OECD and IMF output gaps2.32.32.23.7
Italy
Fiscal impulse measure
OECD2.4−1.1−1.5−0.7
IMF1.9−1.5−2.10.6
Fiscal balance: general government
OECD−11.7−11.9−11.9−12.5
IMF−11.7−11.9−11.9−12.5
Italy
Output gap
OECD6.89.713.914.4
IMF1.75.39.18.2
Difference between OECD and IMF output gaps5.14.44.86.2
United Kingdom
Fiscal impulse measure
OECD−3.1−1.80.5−0.2
IMF−2.7−0.72.0−0.5
Fiscal balance: general government
OECD−2.8−2.0−2.7−2.3
IMF−2.8−2.1−3.6−2.8
Output gap
OECD7.38.47.77.6
IMF7.57.67.17.0
Difference between OECD and IMF output gaps−0.20.80.60.6
Sources: OECD (1983 c), Vol. 34; data provided by the OECD; and Fund staff estimates as of February 29, 1984.
Sources: OECD (1983 c), Vol. 34; data provided by the OECD; and Fund staff estimates as of February 29, 1984.

The extent to which differences in the output gap and budget balance can emerge is also indicated in Table 1. For example, for 1984, the Fund’s estimates suggest a narrowing of the output gap in Italy, whereas the OECD estimates suggest a widening of the gap. With roughly the same estimated fiscal balances, these divergences in potential output would lead to an increase in the cyclically neutral balance in the Fund estimates and thus a higher measure of fiscal stance and a more expansionary impulse. For France, there are also divergent estimates of the output gap, but this is offset by Fund estimates of a deficit lower than that estimated by the OECD, 2.9 percent versus 3.8 percent of GDP. Differences in the assumed tax elasticity can also play an important role in explaining differences between the OECD and IMF results. The different effects of an output gap of 1 percent on the assumed cyclically adjusted general government budget balances of the OECD and the IMF are indicated in Table 2 for 1981–83.

Table 2.OECD and IMF Models: Effect of a 1 Percentage Point Increase in Output Gap on Cyclically Adjusted Budget Balance of General Government(As percentages of potential GDP)
OECD
AverageIMF
Country19831981–831983
Canada0.30.40.35
United States0.40.40.30
Japan0.20.20.25
France0.60.60.39
Germany, Federal Republic of0.40.40.42
Italy0.20.20.34
United Kingdom0.70.60.36

Thus, for the United Kingdom, the more contractionary stance indicated in 1982 by the OECD model may reflect the greater increase in its estimate of the output gap, relative to that of the IMF, as well as the stronger response of the budget balance to the emergence of a gap.

Full-Employment Balance Measure

Unlike the two approaches discussed above, the full-employment-balance (FEB) measure, now more commonly referred to as the “high-employment” surplus or deficit, evaluates both expenditure and revenue at an assumed high-employment level of output and yields a measure of the budgetary position at this output level. It thus focuses more on the level of the cyclically corrected deficit than on the change in the deficit.15 This is done by a “grossing-up” process described in detail in de Leeuw and others (1980). Once tax and expenditure bases are calculated, estimated elasticities are applied to yield a high-employment budget balance. This approach has also been extended by de Leeuw and Holloway (1982) to include the effects of inflation on the budget. The FEB approach has been adopted by the U.S. Department of Commerce. The FEB is explicitly designed to measure discretionary policy. The feedback effects of actual income on the observed budget are eliminated because the FEB is not a function of the actual level of output. At any point in time, the FEB is solely a function of exogenous variables.16 Following Blinder and Solow, the change in the FEB can be written as

where

  • Y* = high-employment level of output

  • τ = exogenous tax instruments

  • G = government expenditure

and where dT/dτ is evaluated at Y = Y*. This is the formulation of the FEB that is the closest in concept to the fiscal impulse.

Some of the drawbacks of the FEB approach are obvious and common to the GCEE and OECD measures. For example, both measures are based in the first instance on some measure of the (unobserved) potential or full-employment income level. The balanced-budget problem also remains, though some researchers have corrected for it.

There is, however, one drawback that is unique to the FEB approach. This concerns the potentially misleading signals generated by evaluating tax policy at a level other than the observed level of output. Suppose that under conditions of less than full employment, the authorities lower personal taxes but raise corporate taxes so that the average effective total tax rate decreases at the current level of income (where corporate profits are low) but increases at the full-employment level. Any FEB technique would term such a policy, which is obviously expansionary at the observed level of income, contractionary.

Finally, although the computational burdens of using the FEB need not be as great as they are for the U.S. Department of Commerce, it seems that even if one were prepared to accept less precision and refinement, the calculations involved in generating both expenditure and revenue values for a hypothetical level of income would be substantial.

Weighted Standardized Surplus Measure

The weighted standardized surplus (WSS) measure is included because it is a good example of an indicator of fiscal policy aimed at gauging discretionary action by the authorities. It is not, however, a measure used regularly by any national or international agency. This method was first proposed by Blinder and Solow (1973) and was implemented by Blinder and Goldfeld (1976) with U.S. data. Similar studies focusing on foreign economies have been done by the staff of the Board of Governors of the Federal Reserve System. This method does not involve the calculation of any measure of potential output and does not treat any endogenous variable as exogenous in the calculations. The method allows for the important element of time in fiscal policy and also evaluates policy at actual levels of output. It does require a well articulated, accurate structural econometric model with unbiased parameters. Simulation techniques are employed to decompose the budget into autonomous (exogenous) and induced (endogenous) components. The fiscal impulse is defined as the change in the exogenous component of the budget. The complexity and high cost of this technique would be distinct drawbacks, even if one were to use a small model, as Hansen and Snyder (1969) did in an early OECD study.

European Community Measure

The European Community measure (see Commission of the European Communities (1982)) considers the components of changes in the budget balance, allowing for cyclical variations or adjustments for the level of economic activity. In this summary measure, the actual year-to-year changes in the budget balance are attributed to a number of factors, which are then used in the EC’s policy analysis. The actual change in the budget balance ΔB is expressed as

where

  • ΔA = effect on the deficit of changes in the level of economic activity

  • ΔINP = effect of changes in net interest payments from the government to the domestic and foreign sectors

  • ΔR = the component of budget change which is conceptually similar to the fiscal impulse used in the Fund methodology

Conceptually, the European Community method is similar to those of the Fund and OECD, in that it allows for variations in the level of economic activity. In the EC approach, however, all variables are expressed as first differences; changes are calculated by comparing the current year’s figure with the previous year’s; and interest payments are treated separately, without allowing for changes owing to inflation. Interest payments are netted out in order to give a clearer picture of “discretionary policy” changes.17 The Commission of the European Communities (1982) report is not very explicit about exactly how the ΔA variable is derived, but conceptually it is defined as the difference between the actual budget balance and what the budget balance would have been if economic activity in a given year had remained at the same level as in the previous year. This method also uses potential and actual GDP, and their differences to derive ΔA; however, unlike the Fund method, the deviation of actual from potential GDP is not necessarily viewed as “cyclical,” and allows for random supply disturbances which may also cause deviations of actual output from potential output. The tax ratios used are similar to those in the Fund method, since a single marginal tax rate is used (which equals the average tax rate) and social transfers are assumed to vary with the level of unemployment.

    Other Resources Citing This Publication