Abstract
The swedish ministry of finance has developed a simple technique of short-range fiscal analysis to aid policymakers in assessing the effects of fiscal policy on the national economy. This technique is significant in that it goes beyond changes in the budget balance to analyze the effects of changes in various budget items on the level of total demand.
The swedish ministry of finance has developed a simple technique of short-range fiscal analysis to aid policymakers in assessing the effects of fiscal policy on the national economy. This technique is significant in that it goes beyond changes in the budget balance to analyze the effects of changes in various budget items on the level of total demand.
Study of the Swedish technique of fiscal analysis is particularly interesting because the Swedish Government has long used countercyclical fiscal policy to achieve full employment. Swedish economists have been engaged since the 1930s in both formulating and criticizing government budgets in light of the effects that they were anticipated to exert on the goals of full employment and price stability.
This paper describes the approach to fiscal analysis that is currently applied by the Ministry of Finance to estimate the effect of fiscal action on the national economy. It also explores some of the implications of this approach and of the economic model upon which it is based. The concluding section presents a brief evaluation of the technique.
I. Evolution of Swedish Budget Philosophy
Deliberate countercyclical fiscal policy began in Sweden in 1933 when Ernst Wigforss, the Social Democratic Minister of Finance, committed himself to an expansionary budget in order to achieve full employment. His motivation apparently came from the Fabian discussion in England in the late 1920s on the use of expansionary fiscal policy, especially through public works projects, to combat unemployment.1
The Budget Reform of 1937, designed largely by Dag Hammarskjöld, formalized a particular kind of fiscal policy—the balancing of the current budget over the cycle2—by allowing it to run deficits in slack periods and offsetting surpluses in booms. This was considered an effective method of overall demand management. The aim was to smoothen fluctuations in the level of economic activity by balancing the current budget over the longer run, thereby shifting demand between periods and thus promoting a high level of employment.
This Budget Reform also highlighted the differences in financing the current and capital budgets. The current budget was to be financed chiefly by taxes, while the capital budget was to rely mainly on loans. The long-run balancing of the current budget focused attention on the profitability of government investment expenditure; to qualify as capital expenditure, government investment had to earn a rate in line with the market rate for government borrowing.
In the late 1940s, the Ministry of Finance changed the requirements for government investment expenditure, thereby increasing the scope for capital budget financing. To be in the capital budget, a project had to be profitable only in a national economic sense, that is, it should add to future net national income enough to cover the increased interest on government loans. The Ministry of Finance also indicated that a long-run deficit on the current budget was acceptable as long as the growth in the deficit paralleled the growth in national income. These reforms of the Ministry of Finance represented the views of many Swedish economists who thought that the long-run balancing of the current budget posed too strict a budget rule.3 In their opinion, running budget deficits in recessions, not matched by surpluses in booms, was an acceptable instrument for achieving full employment. Also, since they looked upon the size of the public debt as a purely technical issue, they did not focus their attention on questions of future tax burdens and intergenerational equity.
In the 1950s much political debate took place on the question of surpluses or deficits in the current budget and on the feasibility of cutting taxes. The division of the budget into current and capital accounts facilitated the manipulation of figures, with the result that the political maneuvering confused the issues and hampered the development of a rational fiscal policy.4
During the latter part of the 1950s discussion centered more on the total budget (current plus capital) rather than on the current budget alone. Since the total budget was often in deficit, while the current budget was usually in surplus, the political debate on budget surpluses and the desirability of cutting taxes became less important. Also, at this time it became clear that the desired levels of future investment (public plus private) would require increased government saving to keep long-run aggregate demand in equilibrium.5 This implied a long-run surplus on the current budget. Consequently, the Ministry of Finance considered a budget-balancing rule relating to the total budget, that is, a rule that revenues be equal to current plus capital expenditure. This implied that government investment expenditure would be financed by taxes and that government borrowing would eventually diminish to zero. As this approach to budget policy proved incompatible with the prevailing situation, a less dogmatic and more pragmatic approach was later adopted.
Thus, the traditional approach to fiscal analysis in Sweden employed the budget deficit as a summary measure of the effects of fiscal policy. The common thread running through the discussion and analysis was the employment of a simple, unweighted budget balance to indicate the nature of the effects of change in fiscal policy. This traditional approach contrasts with the recently developed technique of fiscal analysis by the Ministry of Finance, presented in the next section, which emphasizes the differential effects of changes in budget aggregates on the level of total demand rather than any simple measure, such as the budget balance.
The technique of short-range fiscal analysis presented here has evolved from the work of several Swedish economists. In the late 1950s the Swedish Institute of Economic Research, under the direction of Bent Hansen, published surveys of the economic situation that included weighted public sector impact analysis. These impact studies were continued by Lindbeck and his associates, who assessed policies in the years 1963-68, and by Hansen’s survey of the period 1955-65 for the Organization for Economic Cooperation and Development (OECD).6 When the Planning Department was established within the Ministry of Finance in 1962, the assessment of public sector impact, in the context of national budget projections, shifted from the Institute to the Ministry. The Institute, however, still contributes major sections to both Preliminary and Revised National Budgets. The budget impact model that is described next, which first appeared in the 1969 issue of the Preliminary National Budget (PNB), was developed within the Ministry of Finance.
II. The Budget Impact Analysis
It is widely agreed that, ideally, an econometric model should be employed to determine quantitatively the effects of fiscal policy on the economy. However, econometric models are not readily available, and even when they are, in view of the present state of the art, policymakers often are understandably unwilling to base their decisions on the bare results of any one model. Besides, for discussion of policy in terms understandable to administrators, legislators, and others, a nontechnical, shorthand method of analysis is generally considered useful. The Swedish Ministry of Finance has developed such a technique of analysis. A description of this technique was first presented in the PNB of 1969 and has been included in all successive issues of both Preliminary and Revised National Budgets.
The motivation for developing this technique of fiscal analysis was the desire to improve upon the budget balance concept in determining the effects of fiscal policy on total demand. The Swedish authorities stress that changes in the budget balance do not accurately reflect the effects on the economy of changes in various budget items, stating:
A given change in the residual or in financial saving can thus be due to a large number of combinations of changes in the various items of the budget, each combination having a characteristic effect upon the national economy. One cannot therefore automatically assert that an over-balanced budget has a contractive effect or an under-balanced budget an expansionary effect on national demand. Still less can it be assumed that a balanced budget … leaves demand unaffected.7
Since changes in the budget balance cannot be taken to be indicators of the effects of fiscal policy, the Ministry of Finance proposes a technique for estimating fiscal stimulus that aims to measure the effects on total demand of changes in budget aggregates.
scope of the analysis
The analysis is applied to the budget for the consolidated public sector, which consists of the Central Government and local authorities as well as the social security system, the investment funds, and housing investments controlled by the Central Government.8 Since the Central Government exercises some control over these sectors of the economy, they represent the areas through which the Central Government may influence the level of economic activity.9 Therefore, the consolidated public sector indicates the Government’s potential scope for shaping financial policy in its widest sense.10
The Swedish approach measures the effects of fiscal policy on total demand and not on the level of gross national product (GNP). Total demand includes the demand for imports as well as the demand for domestic production, or GNP. Hence, a given increase in total demand will generally be larger than the accompanying rise in GNP. In fact, most studies on the effectiveness of fiscal policy use the level of GNP rather than that of total demand to measure the effects of fiscal policy changes.
The Swedish method of fiscal analysis also uses a particular definition of budget deficit: financial saving or the difference between total revenues and expenditure exclusive of net lending, other than government support to the housing sector.11 Financial saving may be described in national accounting terms as the excess of public sector saving over its own investment. In the budget accounts, a measure of overall budget balance—that is, the budget balance inclusive of net lending (relating to the Central Government only)—is also presented.12 This overall budget balance represents the financing requirement of the Central Government. This is similar to the definition of budget deficit used in some Western European countries and in the United States. However, the overall budget deficit has no direct role in the budget impact analysis in Sweden.
description of the analysis
As stated earlier, the Swedish approach to fiscal analysis is based on the proposition that different combinations of changes in government expenditure and revenue, although yielding the same change in the budget balance, affect total demand in different degrees. The first step consists of separating all budget changes into two main parts. The first part comprises changes in the central and local government consumption and investment expenditure, as well as changes in housing investments (loans) controlled by the Central Government. In the model, these changes in the composite public sector budget constitute direct increases in the demand for real resources emanating from the public sector, and they are assigned a multiplier effect of 1 in determining budget impact on total demand. The second part of budget changes consists of changes in the level of financial saving of the consolidated public sector: that is, changes in the budget balance exclusive of net lending, arising from changes in direct government expenditure and transfers, on the one hand, and in government tax receipts, on the other hand.
The Swedish model assumes that the level of financial saving forms a net contribution to the disposable income of the nonpublic sectors, and it further assumes that a given absolute change in financial saving will result in an equal but opposite change in the level of total demand. For example, if the Government increased tax revenues by 10 and did not increase expenditure, according to the model, the disposable income of the private sector would decrease by 10, leading to a decrease of 10 in total demand. Therefore, an increase in financial saving is assigned a multiplier of—1 in determining impact on total demand, implying that a balanced budget increase (revenue and expenditure increasing by the same amount) is assigned a multiplier of 1 in determining budget effects on total demand. Also, an increase in direct expenditure not accompanied by a corresponding rise in revenue is assigned a multiplier of 2, whereas an increase in tax revenue not accompanied by a corresponding increase in expenditure is assigned a multiplier of -1 in determining budget effects on total demand.
The Ministry of Finance explains this choice of multipliers as follows:
Here it has been summarily assumed that these two types of changes are equivalent with respect to their effects on activity in the national economy and hence can be combined by simple addition. The influence exerted via changes in the flows of income in the non-public sectors is admittedly curbed by leakages as saving, imports, etc., but also gives rise to chain reactions. Here it is presumed that in these cases one can insert a multiplier of 1. This means, in other words, that a change in the expenditure surplus [financial savings] … is assumed to be accompanied by an equally large change in the other sectors’ demand for goods and services.13
Interestingly enough, this approach to the estimation of budget effects does not distinguish between domestic and foreign expenditure by the Government. Since foreign expenditure will have little effect on domestic demand, the multiplier for domestic expenditure is actually higher than the total government expenditure multiplier given in the model. However, the Ministry of Finance claims that variations in government expenditure abroad will have no appreciable effect on the calculation of budget impact and therefore need not be considered further.14
This method of analysis may be depicted in the following manner: let ΔK represent the change in total demand arising from changes in the consolidated public sector budget and be equal to the sum of ΔK1, the change in total demand arising from direct government expenditure, and ΔK2, the change in total demand arising from changes in the level of financial saving. Then
and, since ΔK1 = ΔG, and ΔK2 = - ΔFS (change in financial saving),
Since financial saving is equal to the difference between direct government expenditure (G) and transfers (E), on the one hand, and tax revenue (T), on the other hand, we may write
By substitution in equation (2), we derive an equation relating changes in total demand to changes in budget aggregates:
Equation (2) presents the Swedish model as described in the PNB; the changes in both direct government expenditure and financial saving are ascribed a multiplier of 1 in determining budget effects on total demand. This approach appears similar to the traditional type of Keynesian analysis, where multiplier effects are ascribed to changes in budget aggregates and not to changes in the budget balance itself. The multiplier of 1 for changes in financial saving reflects the assumption that all nonfinancial transactions of the consolidated public sector (taxes, transfers, and expenditure) exert similar effects, apart from sign, on the level of private sector (nonpublic) demand: that is, the change in the expenditure surplus indicates the influence on demand arising from the other sectors as a result of government fiscal action.16
As just demonstrated, the Swedish method employs a government expenditure multiplier of 2, a tax multiplier of -1, and a transfer multiplier of 1. Although these multipliers seem similar to the multipliers in simple Keynesian models, the Ministry of Finance’s model has a broader scope because it focuses on changes in total demand rather than on changes in GNP. Therefore, the multipliers in equation (3) are not the exact equivalents of the usual Keynesian multipliers. Thus, it might be expected that the multipliers in the Swedish model would differ in size from the multiplier whose only function is to estimate the effects of changes in the budget aggregates on the level of GNP.
rationale of the multipliers
The effectiveness of the Swedish model as an instrument to measure the impact of fiscal policy changes on the level of total demand depends partly on the accuracy of the multipliers used. The explanation of the values that these multipliers assume has appeared in every issue of the PNB since 1969, as described earlier. The 1969 PNB compared the results of the Swedish model with those of a model used by Bent Hansen to estimate the effects of fiscal policy changes on the level of GNP in his study of the effects of fiscal policy in OECD countries for 1955-65.17 With a Keynesian-type model, Hansen estimated for the Swedish economy multipliers of 1.92 for government expenditure for goods and services, -0.92 for direct taxes, and -1.15 for indirect taxes, compared with the multipliers of 2 and -1 in the Swedish model.
However, the similarity in value between Hansen’s multipliers and those of the Swedish model is largely coincidental. The Ministry of Finance points out that the two sets of multipliers are not comparable because (1) Hansen’s model estimates the budget effects on the level of GNP—not on the level of total demand as is the intention in the Swedish model; and (2) the time duration of the multipliers differs in the two models. On the first point, the Ministry of Finance maintains that the marginal effects of budget changes on total demand may be even 40 to 50 per cent greater than on GNP, and therefore total demand multipliers should be larger than GNP multipliers based on the same parametric values. However, the total demand multipliers of the Swedish model are approximately equal to Hansen’s GNP multipliers because the multipliers in the Swedish model are truncated and are therefore lower in value than full effects multipliers based on similar assumptions.18 On the second point, Hansen’s model is designed to estimate the effects of fiscal policy in a comparative static setting—the total effects of budget changes on the level of GNP after all adjustments have occurred—and therefore yields full static multipliers with no time dimension. In contrast, the Swedish model attempts to determine the effects of budget changes on total demand that occur within a year and disregards the multiplier effects of budget changes occurring outside this one-year period.19 Therefore, the multipliers in the Swedish model would be expected to be smaller than ordinary total demand multipliers insofar as they indicate the effects on total demand realized only within the first year of a budget change. After taking these two countervailing considerations into account, it seems that the Ministry of Finance feels that the values of the multipliers selected by it are broadly consistent with those in Hansen’s study.
While pointing out the difference between the Ministry of Finance and Hansen models, the 1969 PNB does emphasize the similarity between the underlying parametric values, stating that the marginal saving and import ratios estimated by Hansen are not appreciably different from those applying to the Swedish model.20 To verify this assertion, an attempt may be made to derive the truncated multipliers from a model that makes use of the structural parameters of Hansen’s study (0.2 = marginal saving ratio and 0.4 = marginal import ratio).21 The multipliers arrived at in this manner may then be compared with those in the Ministry of Finance’s model.
The full static multipliers, which relate changes in the budget aggregates to changes in total demand, may be derived from the simple income expenditure model that follows. These multipliers may be reduced by one fifth to yield the truncated multipliers in the Swedish model, as derived from the formula suggested in the 1969 PNB.22
The model is composed of equations (4) through (10):
where Y = GNP
C = consumption
I = investment
G = direct government expenditure
X = exports
M = imports
T = taxes net of transfers
α1 = marginal propensity to spend
β1 = marginal propensity to import
In the model, equations (4) and (5) are behavioral relations and equation (6) is the usual national income identity. Equations (7) through (10) indicate the variables that are assumed to be exogenous; these variables include not only government expenditure but also taxes. As will be seen later, this assumption is consistent with the distinction between automatic and discretionary budget changes drawn by the Ministry of Finance. The model reduces to
where A = α0 − β0 + I + G + X
Since the Swedish model focuses on the effects of budget change on the level of total demand, the usual Keynesian multipliers must be augmented by a factor that relates budget changes to changes in the level of imports, yielding the total demand multipliers that are derived as follows:
where K = Y + M = total demand.
Now, from our model
Therefore,
By inserting Hansen’s values for the structural parameters into equations (12) and (13), the values for the total demand multipliers can be computed, and, by taking 80 per cent of the results, the truncated total demand multipliers in the Swedish model can be derived. These calculations are as follows:
Thus, the adjustment suggested in the 1969 PNB yields truncated total demand multipliers of approximately 1.9 for changes in direct government purchases and of -1.5 for changes in taxes. This government expenditure multiplier is approximately equal to the multiplier of 2.0 assumed in the Swedish model, whereas the tax multiplier of -1.5 is 50 per cent greater than the value of -1.0 assumed in the Swedish model.
The multipliers of 1.9 and—1.5 were derived on the basis of structural parameters that the Ministry of Finance felt were consistent with their approach. The multipliers derived above, however, are not exactly consistent with Hansen’s marginal saving and import ratios. It is clear, however, that this lack of conformity with Hansen’s structural parameters and multipliers is in itself no ground for criticism of the Swedish approach. If in fact the values of the parameters are slightly altered so that the marginal propensity to spend is equal to 0.7, while the marginal import propensity remains at 0.4, a truncated total demand multiplier for government expenditure of 1.6 and one for tax revenues of -1.1 are derived—values somewhat closer to those of the multipliers in the PNB exercise.25
Starting from the other end, if the value of 2.5 is inserted for the direct government expenditure multiplier (that is, raise the first-round impact by 25 per cent to get the total demand impact because 125 per cent of 80 per cent is 100 per cent) and -1.25 is inserted for the tax multiplier into equations (12) and (13) to solve for the values of the parameters, the following values are obtained: α = 0.50 and β = -0.16. Moreover, the value of the tax multiplier is seen to be particularly sensitive to changes in the assumed values of the parameters. For example, a change in the propensity to spend from 0.8 to 0.7, while keeping the propensity to import at 0.4, gives rise to a change of 0.4 in the tax multiplier, that is, from -1.5 to -1.1. In view of this high degree of sensitivity in the multipliers, it could be maintained that the truncated multipliers in the Swedish model are consistent with structural parameters that are not far different from those likely to apply to the Swedish economy, or those estimated by Hansen in his OECD study.
automatic versus discretionary effects
The Ministry of Finance distinguishes between budget changes arising as a result of deliberate policies (measures) and those occurring because of variations in the level of economic activity (automatic effects). By the usual definition, discretionary budget effects consist of changes in revenue arising from a change in tax rates or in the legal tax base, while automatic effects are taken to occur as a result of growth in the tax base (income) with a given rate structure. In the simple case where it is assumed that the marginal and average tax rates are the same, then the automatic and discretionary changes in revenue may be defined as follows:
where ΔT = T1 - T0
t = the average tax rate that would have obtained in the absence of a change, given a particular level of income.
ΔtΔY is considered to be a discretionary change ex post, and is normally omitted from the measure because its value is considered negligible.
The Swedish authorities define discretionary budget changes to include also those changes in tax revenue that result from a planned or anticipated increase in the tax base (income) with a given legal rate structure. Thus, by the Swedish definition, most of the change in tax revenue in a given year would be taken to result from the operation of discretionary budget measures. Only the increase in tax revenue that was due to an unplanned or unanticipated increase in the tax base (income) would be considered automatic. In other words, tΔY in equation (14) is broken into two parts: one part measures the discretionary changes in tax revenues resulting from a planned increase in income, while the other part indicates the automatic effect exerted on tax revenues from unplanned increases in income. That is,
where ΔYp = planned increase in income
ΔYup = the unplanned or unanticipated change in income, where again ΔtΔY is considered to be negligible and where t is assumed not to undergo a significant change as a result of the attainment of different levels of income.
The Ministry of Finance states that such a distinction between automatic and discretionary budget effects serves to highlight the importance of deliberate budget policy for the development of government revenue. It argues that a decision not to change tax rates when the tax base (income) is expanding should be accepted as a deliberate measure of policy just as much as a decision to change tax rates to increase revenues.26
In order to compute the discretionary change in revenue according to this definition, it is necessary first to measure the planned and unplanned changes in income (or in the tax bases if a more disaggregated approach is used). In the simple case where the marginal tax rate equals the average tax rate (apart from legislative changes), the unchanged rates may be applied to the computed changes in income to derive tΔYp and tΔYup. However, in a real-world situation, the appropriate marginal tax ratio would have to be worked out.
The Ministry of Finance argues that since it is very difficult to determine objectively the increases in the tax bases (which are unplanned in the sense used earlier), any direct measurement of discretionary budget effects is not really possible. However, it also reasons that insofar as actual growth in the economy deviates from potential growth there will additionally accrue or fail to accrue some amount of revenue to the public sector. By assuming such deviations in the economy’s growth to equal the unanticipated changes in income, the change in revenue could be computed. Ideally, this may be done by applying a proper marginal tax ratio to the difference between actual and potential growth. But, because of the great difficulty of constructing a hypothetical marginal tax ratio for each year, the Ministry of Finance proposes to use the actual average tax ratio instead. This implies that the marginal and average tax rates are assumed to be equal. The Ministry of Finance justifies this procedure also on the ground that any change in the actual tax ratio resulting from a difference between the marginal and average ratios may be considered to be the result of a deliberate policy choice, namely, the act of allowing the actual tax ratio to increase.27
The following procedure is adopted for measuring the budget effects: first, the change in the volume of direct demand from the Central Government, local governments, and housing construction supported by the public sector is calculated by taking the volume change in the expenditure of the Central Government and local governments for consumption and investment and of the “funded” expenditure for residential construction. Second, financial saving in the various subsectors is computed as explained earlier. (For the housing sector, investment at current prices is taken to represent financial saving.) Total financial saving in the public sector is then deflated by the gross domestic product (GDP) deflator for the year in question and changes in “real” financial saving are calculated. These are taken to represent the effects of indirect demand from the public sector. Finally, the effects of direct and indirect demand are added, and the sums are expressed as percentages of GDP at constant prices. In Chart 1 total budget effects as a percentage of GDP for the years 1961-72 are plotted in curve A.
Sweden: The Calculated Effect on the Economy by the Public Sector, 1961-72
Citation: IMF Staff Papers 1974, 002; 10.5089/9781451947434.024.A006
Source: Swedish Ministry of Finance, Preliminary National Budget, 1972, p. 190.A = The effect of the public sector on total demand.B = The effect of the public sector on total demand after certain automatic effects on public income have been eliminated.To identify the effects of discretionary changes, the Ministry of Finance provides an estimate of changes in hypothetical financial saving, corresponding to the development of tax revenue generated at potential output, which is derived by applying the actual tax ratio to the potential growth in output. Budget effects are then recalculated on the basis of changes in the new hypothetical level of financial saving, as illustrated in curve B of Chart 1.
If the planned increase in income shown in equation (15) is assumed to be the same as the potential growth in output as described by the Ministry of Finance, then the measure of budget impact in curve B describes discretionary budget changes, and is similar to the full employment budget surplus (FEBS) concept developed in the United States.28 Both the FEBS and the estimate of budget effects depicted in curve B compute the yield of the budget structure at some potential level of output. In the Swedish model, the normalized growth in revenue is used to determine changes in the hypothetical level of financial saving, which in turn are employed in conjunction with changes in direct government purchases to estimate the effects of fiscal policy change.29 This amounts to a weighting of the FEBS in the determination of budget effects.30 On the other hand, the FEBS, derived as the hypothetical budget outcome at full employment, is used in the United States as a summary (unweighted) indicator of the nature and direction of fiscal policy.
III. Evaluation and Conclusions
The Swedish technique of fiscal analysis may be said to represent an improvement over the use of the unweighted budget balance as a summary indicator of the effects of fiscal policy. Although both techniques estimate the expansionary impact on total demand, rather than on the level of GNP, the unweighted budget balance approach, unlike the Swedish technique, assumes that all changes in budget items (expenditure on goods and services, transfers, or revenues) will affect economic activity to the same degree. In the use of the unweighted budget deficit, therefore, a change in government expenditure and revenue that did not result in a change in the budget balance would indicate no change in the expansionary impact of the budget. The use of this simple deficit would thus overestimate or underestimate the expansionary impact of a budget change if the composition of budget items changed along with the levels of revenue and expenditure.
In contrast, the Ministry of Finance’s model is based on the proposition that changes in different budget items will exert varying effects on the level of demand and, thus, that changes in the budget deficit are not unequivocal indicators of the effects of fiscal policy. Specifically, the Swedish approach assumes that an equal change in government expenditure and revenue, leaving the budget deficit unchanged, will exert an expansionary impact on demand, within a one-year period, equal to the amount of the increase in expenditure: that is, the balanced budget “multiplier” within one year is unity.
The Ministry of Finance’s model assumes that a balanced budget change will increase demand through the increase in direct government purchases; by implication, no change would occur in the level of disposable income since the budget residual or level of financial saving is unaffected. This will be true, however, only if the level of GNP rises; and for this to happen, the marginal propensity to spend must be less than unity. Moreover, a balanced budget multiplier equal to unity, as in the Swedish model, requires further restrictive conditions not specified in the approach.31 Thus, the Ministry of Finance offers only a partial explanation for the operation of the balanced budget multiplier.
The Swedish model assumes that 80 per cent of the multiplier effects are realized within the first year of a budget change. Implicit in this assumption are notions concerning the nature of the lags relating to different items of the budget and the length of the respending rounds. For 80 per cent of the total effects of a budget change to be realized within a year, the lags or the length of a respending round must be short. However, it is not clear how it can be assumed that these are uniform for all the budget items.
To measure the effect of the total budget on demand, all changes in both revenues and expenditure must be assumed to be exogenous. This point has been brought out by Bent Hansen, among others, concerning the unweighted budget balance. Since the budget deficit is not, strictly defined, a government policy instrument, but rather an endogenous variable, its use as a summary measure of fiscal policy is suspect. However, the Ministry of Finance’s definition of automatic and discretionary budget changes implies that most of the changes in revenue and expenditure are the result of discretionary policy. Therefore, it can be argued that the Swedish approach encompasses a methodology in which almost all the changes in revenue and expenditure may be considered to be exogenous and that total budget effects may therefore be estimated.
However, since the automatic change in revenue is measured as the difference between the yields of the tax system at the level of potential output and at the level of actual income, by implication the yield at potential output is considered to be the result of discretionary policy. This implicit definition would seem inappropriate unless it was assumed that the budget estimates were always based on the level of potential output. Rather, the use of the estimated revenue at potential output instead of the anticipated revenue at the time of budget presentation makes the Swedish technique analogous to the FEBS approach; thus, curve B in Chart 1 is to be used for ordinal comparisons of the changes in the impact of policy rather than as a cardinal measure of the first-year impact of the total budget. For this purpose, however, a straightforward weighting of the FEBS would seem to be a more suitable technique.
The additional justification by the Swedish authorities for the use of the average tax ratio for computing the revenue at potential output does not seem tenable. The policy choice in regard to tax revenue has to be exercised through the tax structure. For any given level of income, the tax ratio is determined by the tax structure and cannot be specified ex ante. Indeed, if the change in the tax ratio is to be looked upon as the result of deliberate policy choice, as the Swedish Ministry of Finance contends, the anticipated (discretionary) tax revenue will have to be computed on the basis of the higher tax ratio and the nonrealization of the higher tax ratio will have to be attributed to the operation of the automatic element.
Exercises designed to estimate the first-round effects of budgetary changes,32 as well as those designed to measure their total multiplier effects,33 consider the impact on aggregate demand within the economy or GNP. The Swedish model, in contrast, attempts to indicate the impact on total demand (GNP plus imports). It has been demonstrated earlier34 that if truncated multipliers of + 2 for an increase in expenditure on goods and services and of −1 for an increase in financial saving are assumed, it would be implied that the marginal propensity to import is close to zero. Such complications arise because the Swedish approach involves the use of multipliers. If the measure of impact were to be confined to first-round effects, the failure to distinguish between domestic and foreign expenditure (and receipts) of the Government could be justified on the grounds that the proportion of foreign expenditure was unlikely to vary appreciably from year to year.
Also, the Swedish model operates on the basis of financial saving or deficit. Since this is equal to net borrowing minus net lending, expenditure by the public sector is taken net of lending, except for the Central Government’s support for the housing sector, with the underlying assumption that all other lending represents only financial intermediation by the Government. In the particular context of Sweden, this assumption may be appropriate.
In summary, the Swedish model described in this paper represents an interesting attempt at developing a simple and readily understandable technique of fiscal analysis intended to overcome the principal defect in the use of the unweighted budget deficit. It is also an improvement over the latter to the extent that it recognizes the operation of balanced budget changes in the expansion of demand and presents an estimate of budget impact for normalized revenue growth (Chart 1, curve B), a measure similar to the FEBS. However, the Swedish model does not differentiate between different types of nontransfer expenditure nor between different taxes. The only differential impact taken into account is that between expenditure on goods and services, on the one hand, and taxes and transfers, on the other hand. While the attempt to introduce the notions of lags into the simple analytical framework is a promising innovation, the lack of distinction between impact on GNP and on imports and the assumption of identical lags for all budget items seem to limit the quantitative significance of the results obtained.
APPENDIX: An Alternative Rationalization of the Multipliers in the Swedish Model
Professor Lars Matthiessen makes use of a model slightly more complex than the one described in this paper in order to determine the assumptions implicit in the Ministry of Finance’s approach to the estimation of budget effects.35 He illustrates one possible solution of his model by this equation:
where e = total demand
g = government expenditure
T = taxes in nominal value
G = government expenditure in nominal value
Py = GDP deflator
Matthiessen states that this reduced form equation is equivalent to that of the Swedish approach (equation (2) in this paper), where government expenditure and changes in financial saving are both given multiplier effects of 1 to determine changes in total demand. However, he shows that the structural parameters consistent with this solution to the model are as follows: α (marginal propensity to spend) = 0.5 and μ (marginal propensity to import) = 0. These results are approximately the same as those derived by the simpler model described in this paper.
In his interpretation of the nature of the multipliers employed in the Swedish approach, Matthiessen criticizes the Ministry of Finance for supposedly assuming that the full multiplier effects are realized within the same year as the budget change. It has been shown in this paper, however, that the Ministry of Finance assumes rather that only partial or truncated multiplier effects are realized within a year, and further that the Ministry of Finance assumes that a determination of only these truncated effects is sufficient to compare the effects of policy in different years.
Mr. Di Calogero, a graduate of Fordham University, was attached to the Fiscal Analysis Division of the Fiscal Affairs Department under the Young Professionals Program when this paper was prepared and is now in the Financial Relations Division of the Treasurer’s Department.
This paper is the third in the series of studies of the techniques of fiscal analysis applied in selected industrial countries that has been undertaken by the Fund’s Fiscal Affairs Department. The other two papers were prepared by Daryl A. Dixon and also appeared in Staff Papers—“Techniques of Fiscal Analysis in the Netherlands” (November 1972, pp. 615-46) and “The Full Employment Budget Surplus Concept as a Tool of Fiscal Analysis in the United States” (March 1973, pp. 203-26).
Otto Steiger, Studien zur Entstehung der Neuen Wirtschaftslehre in Schweden (Berlin, 1971), pp. 84-87.
Bent Hansen, Fiscal Policy in Seven Countries, 1955-1965: Belgium, France, Germany, Italy, Sweden, United Kingdom, United States, Organization for Economic Cooperation and Development (Paris, 1969), Chapter 7, p. 352.
Assar Lindbeck, “Theories and Problems in Swedish Economic Policy in the Post-War Period,” American Economic Review, Supplement, Vol. 58 (June 1968), pp. 33-34.
Ibid., p. 34.
Hansen, Fiscal Policy in Seven Countries, 1955-1965 (cited in footnote 2), p. 352.
Ibid. Also, see Assar Lindbeck, Statsbudgetens verkningar på konjunkturutvecklingen, Statens Offentliga Utredningar, No. 48 (Stockholm, 1956), in which he breaks down budget categories to achieve some measurement of the budget impact.
Swedish Ministry of Finance, Preliminary National Budget, 1972, p. 186.
The investment funds consist of a percentage of the taxable profits of business that are deposited free of income tax in blocked accounts with the Sveriges Riksbank. The Central Government at appropriate times permits businesses to invest these funds to stimulate demand and thus to increase employment. For further details on these sectors, as well as the relationship between the Central Government and local governments, see Hansen, Fiscal Policy in Seven Countries, 1955-1965 (cited in footnote 2); Lindbeck, “Theories and Problems in Swedish Economic Policy in the Post-War Period” (cited in footnote 3); and Preliminary National Budget, 1972, pp. 175-86. For details on the use of investment funds in financial policy, see Gunnar Eliasson, Investment Funds in Operation, Occasional Paper, No. 2, published by the Swedish Institute of Economic Research (Stockholm, 1965).
For further discussion of instruments and goals in economic policy, see Bent Hansen, The Economic Theory of Fiscal Policy (London, 1967), Chapters I-II, pp. 3-40.
Although the local authorities are considered part of the public sector for the purpose of estimating the impact of budget policy, they do in fact exercise considerable freedom in their fiscal operations as they have their own taxation rights and some freedom of decisions relating to expenditure. Moreover, in recent years, the reactions of the local authorities to central government fiscal measures have not been easy to predict. Thus, the question arises as to what extent the local authorities may be considered to form a part of a uniformly directed public sector, and to what extent the impact of their fiscal operations may be considered part of a centrally formulated budget policy. The Swedish Ministry of Finance is no doubt aware of this problem, but apparently it is unable to split the local authorities’ sector into autonomous and controlled parts owing to lack of statistical data. Without such data, the Ministry may have decided that including the entire sector was a better choice than totally excluding it.
Preliminary National Budget, 1972, p. 182, Table VIII.
Ibid.
Ibid., pp. 188-89.
Ibid., p. 189.
This equation is similar to equation 5 in Joergen Lotz, “Techniques of Measuring the Effects of Fiscal Policy,” OECD Economic Outlook: Occasional Studies (Paris, July 1971), p. 9.
Preliminary National Budget, 1972, p. 188.
Hansen, Fiscal Policy in Seven Countries, 1955-1965 (cited in footnoted).
Swedish Ministry of Finance, Preliminary National Budget, 1969, p. 119. The Swedish authorities claim that this limitation is justified by their intention to compare the direction of fiscal policy year by year, but they state that some studies indicate that limiting the analysis to only those effects of fiscal policy that occur within one year might cause a downward adjustment of one fifth.
Preliminary National Budget, 1969.
Ibid.
Ibid. For another attempt to work out a possible basis for the PNB approach through the use of a more fully articulated model, see Lars Matthiessen, “Budgetary Effects as Determined by the Swedish Ministry of Finance,” Swedish Journal of Economics, Vol. 73 (June 1971), pp. 252-56. A brief summary of this approach appears in the Appendix to this paper.
Preliminary National Budget, 1969, p. 119.
Where
Where subscript a indicates abbreviated or truncated multiplier effects.
Preliminary National Budget, 1972, p. 187.
Ibid., p. 193.
See Daryl A. Dixon, “The Full Employment Budget Surplus Concept as a Tool of Fiscal Analysis in the United States,” Staff Papers, Vol. 20 (March 1973), pp. 203-26, for an explanation of the FEBS and an examination of some of its implications.
The Swedish model does not consider separately the effects of changes induced in government expenditure as a result of variations in the level of output. Such automatic changes would include, for example, variations in the level of unemployment compensation as output changed over the cycle.
Weighted not on the basis of first-round effects but on the basis of truncated multipliers.
Compare J.W. Nevile, Fiscal Policy in Australia: Theory and Practice (Melbourne, 1971).
See, for example, Assar Lindbeck, “Fiscal Policy as a Tool of Economic Stabilization—Comments to an OECD Report,” Kyklos, Vol. 23 (1970), pp. 7-32.
Hansen, Fiscal Policy in Seven Countries, 1955-1965 (cited in footnote 2), and Nevile, op. cit.
See the section, rationale of the multipliers, in this paper. It is possible, however, that the implicit model of the Swedish approach, or the one from which the multipliers have been taken, is more complicated and does not require such an assumption.
Matthiessen, op. cit.
