The Weighted Budget Balance Approach to Fiscal Analysis: A Methodology and Some Case Studies
Author:
Jitendra G. Borpujari Teresa Ter-Minassian
Search for other papers by Jitendra G. Borpujari Teresa Ter-Minassian in
Current site
Google Scholar
Close

THE PURPOSE OF THIS PAPER is to present a relatively simple method for analyzing the effects of fiscal policy through the use of a set of weights for the various budget items. This method has been devised to fulfill two main aims: (1) to take into account the differences in the impact of various budget items on gross national product (GNP) and on the balance of payments, and (2) to provide a tool of fiscal analysis that requires relatively limited time and empirical information so that it may be used relatively easily by those responsible for determining the stance of fiscal policy.

Abstract

THE PURPOSE OF THIS PAPER is to present a relatively simple method for analyzing the effects of fiscal policy through the use of a set of weights for the various budget items. This method has been devised to fulfill two main aims: (1) to take into account the differences in the impact of various budget items on gross national product (GNP) and on the balance of payments, and (2) to provide a tool of fiscal analysis that requires relatively limited time and empirical information so that it may be used relatively easily by those responsible for determining the stance of fiscal policy.

THE PURPOSE OF THIS PAPER is to present a relatively simple method for analyzing the effects of fiscal policy through the use of a set of weights for the various budget items. This method has been devised to fulfill two main aims: (1) to take into account the differences in the impact of various budget items on gross national product (GNP) and on the balance of payments, and (2) to provide a tool of fiscal analysis that requires relatively limited time and empirical information so that it may be used relatively easily by those responsible for determining the stance of fiscal policy.

Section I explains the weighted budget balance (WBB) approach to fiscal analysis and compares it with two other approaches, namely, that based on the simple unweighted budget balance and that based on multiplier analysis. Section II shows how consistent sets of weights, indicating the first-round impact of the various budget items on GNP and on the basic balance of payments, can be derived from a generalized model based on some a priori notions of the working of industrialized economies. Section III presents three case studies to illustrate the application of the method. Finally, Section IV presents an evaluation of the advantages and limitations of the WBB approach in general and of the model presented in Section II in particular, and also indicates directions for further research.

I. The Weighted Budget Balance as a Tool of Fiscal Analysis

Several techniques have been designed to measure the impact of fiscal policy on macroeconomic variables.1 These techniques differ considerably in analytical complexity as well as in the amount of empirical information needed for their application.

The simplest approach consists of using the budget balance2 (or its year-to-year change) as an indicator of the expansionary or contractionary impact of the budget on the economy. This approach has the advantage of using concepts that are well known even outside the province of economics and about which empirical information is usually available. However, it also suffers from some serious limitations.

First, the simple budget balance approach implies that all budget items have, in absolute value, the same one-to-one impact on aggregate demand. However, it can be shown, even within the most simplified framework of a static Keynesian model of an economy with no foreign and monetary/financial sector, that taxes and expenditure have a different absolute impact on aggregate demand as long as the marginal propensity to consume is less than 1. In more complex analytical models that include a foreign sector and allow for the monetary repercussions of fiscal policy actions, it becomes more evident that the impact of various items of revenue and expenditure can vary widely and that the net impact of the budget as a whole can be quite different, both in sign and in size, from what the simple sum of expenditure (with positive signs) and revenues (with negative signs) would indicate.3

The approach presented in this paper has the advantage, over the use of the simple budget balance, of taking into account the differences in the effects of the various budgetary items by assigning to each item a weight indicating its first-round impact on GNP. If each budget item is multiplied by its corresponding weight, the sum of the products may be called the WBB.4 In general, the WBB will be different from its unweighted counterpart, that is, the simple budget balance, in size and sometimes also in sign. In particular, the weighted deficit will be larger if the average expenditure weight5 exceeds (in absolute value) the average revenue weight, and conversely.

A second limitation of the simple budget balance as a tool of fiscal analysis is the fact that it indicates the impact of the budget on total demand rather than on its domestic and foreign components separately. More specifically, use of the unweighted balance as a measure of the effects of a given budget ignores the facts that government expenditure abroad (whether in the form of direct imports, of wage payments, or of transfers to foreigners) does not add to domestic demand and therefore does not increase domestic employment or output6 and that government receipts from abroad (whether grants or loans) do not reduce private domestic resources. The leakage caused by the positive marginal propensity of the private sector to import is also ignored, although this leakage obviously reduces the impact of domestic transfers and taxes on internal demand.

The approach presented here provides separate measures of the effects of a given budget (or its changes) on GNP and on the basic balance of payments: two different sets of weights are assigned to the various budget items (or their changes)—one reflecting their first-round impact on GNP and the other their direct effects on the basic balance of payments. Indeed, the procedure can be generalized to evaluate the impact of the budget on any other macroeconomic variable, such as investment, aggregate consumption, or balance of trade. Obviously, the impact of the budget on total demand, as indicated by the simple budget balance, generally is different in size (and possibly in sign) from that on GNP as measured by the WBB. Whether the former is larger or smaller than the latter depends, among other things, on the relative sizes of the marginal propensities to import of the private and public sectors and on the composition of the budget. For instance, if the marginal import propensity is higher for the public sector than for the private sector, an increase in government expenditure, matched by an equal increase in taxes, will have no effect on the unweighted budget balance but will, of course, cause a net increase in imports that can be captured through a weighting procedure. Differences in the year-to-year changes of the two measures reflect shifts in the composition of the budget. For example, if the public sector has a larger marginal import propensity than does the private sector, an increase of government expenditure on goods and services that is larger than that of other budget items results in a smaller increase in the WBB than in the unweighted balance.

Despite its various shortcomings, the simple budget balance is still widely used as an indicator of the expansionary or contractionary character of a given budget, especially at the policy-making level. However, the recent years have witnessed an increasing effort by both academicians and those concerned with policy making to create and use more complex but more accurate tools of fiscal analysis. More and more, macroeconometric models are being constructed in which an attempt is made to specify and estimate the interrelationships between fiscal and other economic variables, such as the various components of aggregate demand, income shares, prices, and monetary aggregates. Such models are used to simulate the effects of various policy changes as well as the stabilizing (or destabilizing) responses of the budget to exogenous shocks in the economy. In fact, if appropriate lags 7 are introduced into the various equations, econometric models can also be used to trace the time profile of the effects of given fiscal policy actions, that is, to compute multipliers indicating the effects of a unit change in a fiscal instrument on a given endogenous variable within the time unit of the model, in several such periods of time, or in the long run.

Such a use of macro-models represents, in theory, the most satisfactory approach to measuring the impact of fiscal policy on the economy. However, it is well known that building a satisfactory fiscal policy model, accurately specifying the links between fiscal and other variables as well as the various lags that affect the working of fiscal policy, is indeed a complex undertaking that may not always be practicable. The approach presented in this paper is an intermediate one between the use of the simple budget balance and the simulation of a complex econometric model. On the one hand, unlike the simple budget balance approach, it involves weighting the various budgetary items in such a way as to reflect their different effects. On the other hand, the weights are derived from a simple static macro-model and indicate only the first-round impact of the items. This model represents a framework of analysis designed to take into account the major interrelationships between the budget and the economy; it can and should be modified to allow for particular features of any given economy as well as for the extent of the availability of data.

Formulas indicating the impact of the various budget items on GNP are derived by solving the model for GNP in terms of the various exogenous variables. The coefficient of each term in the resulting reduced form equations is the product of a multiplier that is the same for all terms and a multiplicand (weight) that represents the first-round effect of the exogenous variable in question. Given the static nature of a model of this kind, multiplier effects are ignored here and the analysis is limited to first-round effects8 for the following major reasons.

(1) If substantial lags exist in the transmission of the effects of fiscal policy changes, it is not appropriate to assume that the full-multiplier effects take place during the year of the change,9 while in general it is reasonable to assume that first-round effects do so.

(2) Disregarding lags in the specification of the behavioral functions of the model generally results in biased estimates of the parameters of the functions and therefore of the coefficients of the reduced form of the model (i.e., of the weights and of the multiplier). However, the estimate of the first-round effects is affected only by the bias in the weights, so that the error under the WBB approach is smaller than that under a full-multiplier approach.

Weights indicating the direct impact of the budget on the basic balance of payments are derived by solving the model for the basic balance itself, in terms of the various exogenous variables and the level of GNP. Therefore, these weights do not include the indirect repercussions on the balance of payments of the budget impact on GNP.

Another approach to the derivation of weights for the budget items consists in relying on “intuition” or “informed judgment,” possibly using estimates of the impact on various fiscal policy instruments derived from different existing econometric models of the particular country. Most of the earlier exercises in weighting budget items have relied on this approach.10 It has some advantage when limitations of time or data are particularly stringent. However, it has the obvious shortcoming of not ensuring consistency of the weights for budget items. The approach presented here seems more satisfactory, since it ensures consistency of the weights and at the same time is flexible enough to be applied in a manner best suited to the extent of the information available on any particular economy.

II. An Illustrative Model for Deriving the Weighted Budget Balance

General characteristics of the model

This model fits, in many respects, into the usual Keynesian framework of aggregate demand analysis. It is assumed that the level of income is purely demand determined and that there is no supply constraint. Therefore, the model is more appropriate for analyzing fiscal policy in industrialized than in developing countries. Since the model is designed for the derivation of weights for budget items, an attempt is made to incorporate into it all the major components of the budget and their relationships with other economic variables, while keeping the size of the model rather small to facilitate its algebraic solution. In its basic version, the model contains seven behavioral equations and four identities (equations (1), (4), (6), and (11)). The behavioral functions explain three components of aggregate demand (consumption, investment, and imports) and the distribution of income by factor shares. The consumption function (equation (2)) follows the Cambridge tradition 11 in the assumption that the marginal propensity to consume varies among earners of different types of income. Three different types of income are distinguished: (1) profits of unincorporated enterprises and wages, (2) interest, dividends, and rents, and (3) transfers. Normally, one would expect the marginal propensity to consume to be the highest for receivers of transfers and the lowest for receivers of interest, dividends, and rents. As is well known, this assumption has not always found empirical support and indeed in some countries, for example, the United States, there appears to be evidence that consumption behavior is related mainly to the size distribution of income and the ratio of non-human wealth to income. A consumption function such as equation (2) should therefore be tested for applicability in any particular case study.12

As regards aggregate private investment, since the main interest here is to make explicit its relationships with fiscal variables, it is regressed on corporate profits net of corporate income tax, on the sum of net lending and net capital transfers by the government and on an index of the cost of capital—which could be a weighted average of current interest rates (equation (3)). Lending by the government is assumed to affect private investment, whereas borrowing by the government, from either the central bank or other sources, does not affect aggregate demand. Therefore, net lending receives a positive weight, while borrowing receives a weight of zero. It is, of course, open to question whether it is appropriate to assume that government borrowing from private sources does not curtail private investment. In certain countries, it may be possible to identify a “crowding out” effect of government borrowing on private investment and therefore to assign an appropriate negative weight to it.

In the foreign sector of the model, exports are treated as exogenous, while the import function (equation (5)) reflects the assumption of different marginal import propensities for private consumption, private investment, public consumption, public capital formation, and exports.13 The definition of the basic balance of payments (equation (11)) makes explicit the foreign components of the government budget, namely, taxes paid by nonresidents, wages paid abroad, current and capital transfers to and from abroad, and foreign lending net of repayments and borrowing net of redemptions. Private transfers and long-term capital flows are taken to be exogenous.

The model, of course, does not embody a complete theory of income distribution. Private income is defined as net national product at factor cost minus government income from property and entrepreneurship (equation (6)), and four shares in it are identified. These shares, which are assumed to be constant,14 are wages (equation (7)), corporate profits (equation (8)), profits of unincorporated enterprises (equation (9)), and other incomes, that is, interest receipts other than from the government and rents.15 Dividends are assumed to be a constant fraction of corporate profits net of corporate taxes (equation (10)).

Since the labor market, productivity, and money supply are not incorporated in the model, the price level is treated as exogenous. Indeed, in this version of the model, all the functions are specified in current prices, and therefore the WBB indicates the impact of the budget only on nominal GNP.

In fiscal analysis one may be interested in assessing the impact either of budget items (or their total changes) or of discretionary fiscal policy actions only. Taxes and expenditure should be treated as exogenous in the model that is used to derive weights for the budget items, since the effects of any given item (or its total change) on, say, GNP can be assessed only by comparison with the level (or change) of GNP in a situation where the budget item (or it change) was zero. Such a situation could be imagined only if the budget item were not related to the level of GNP.

In assessing the impact of discretionary fiscal policy actions, however, a distinction should be made between automatic and discretionary components of budgetary changes: only the latter should be treated as exogenous and, therefore, should receive weights. Discretionary changes in revenue items are usually identified with changes in tax rates and/or in the definition of the base. Government expenditure is considered to be a policy variable and therefore is treated as exogenous in most macroeconometric models. However, there are also examples of functions that explain particular items of expenditure, such as unemployment compensation or social security benefits;16 in those examples, only changes in benefit rates are considered discretionary. Sometimes it has been assumed that the government decides the level of real expenditure and lets nominal expenditure be determined by the price level, so that only changes in real expenditure are considered discretionary.

In what follows, two variants of the basic model are presented; they differ only in the treatment of revenue items, since expenditure is taken to be exogenous in both. The first variant (referred to as Model I), in which taxes are treated as exogenous, will be used to derive weights for total revenue items and for their changes. The second variant (Model II), in which taxes are treated as endogenous (equations (12) to (15)), will be used to compute the first-round effects of discretionary tax changes, which are defined here as changes in the rates multiplied by the original level of the base.

Model I

Y = C+I+G+X M(1)
C = c0+c1(W+PuTdw)+c2(Q+D+In8Tdp)+c3Sh(2)
I=i0+i1(PcTc)+i2(NL+Sk)i3CC(3)
G=Gc+Gw+Gi(4)
M=m0+m1C+m2I+m3Gc+m4Gi+m5X(5)
YDEPTi+SfGIP=W+Pc+Q+Pu(6)
W=a(YDEPTi+SfGIP)(7)
Pc=β(YDEPTi+SfGIP)(8)
Pu=γ(YDEPTi+SfGIP)(9)
D=δ(PcTc)(10)
BP=(XM)+TfGwf+(TrfSw)+(TrkfSkf)+(BfNLf)+Sp+K(11)

Model II

Equations (1)–(11) are the same as in Model I

Tdw=tow+tw(W+Pu)(12)
Tdp=t0p+tp(Q+D+Ing)(13)
Tc=t0c+tcPc(14)
Ti=t0i+tiY(15)

Notation Used17

article image
article image

Exogenous variables 18

X, Tdw, Ing, Tdp, Sh, Tc, NL, Sk, CC, Gc, Gw, Gi, DEP, Ti, Sf, GIP, Tf, Gwf, Trf, sw, Trkf, Skf, Bf, NLf, SP, K, tw, tp, tc, ti

Endogenous variables

Y, C, G, M, W, Pu, Q, D, Pc, BP

Weights for the budget items

Table 1 presents the weights that indicate the first-round impact on GNP and the direct impact on the basic balance of payments of the various budget items and total changes in them.19

Table 1.

Weights for Budget Items and Their Total Changes

article image
Source: Model I.

σ = c1(α+γ) +c2[1–α–γ–β(1–δ)].

If the direct import content of government expenditure has been subtracted from total imports, it would have a weight of zero with respect to gross national product and of — 1 with respect to the balance of payments.

Among the weights indicating the GNP impact, the largest is that for government wages and salaries paid domestically, because there are no leakages in the first round from this kind of expenditure, either into savings or into imports. The ranking of the weights for the other expenditure items depends on the relative values of the various parameters: the marginal propensities to import consumption and investment goods for the private sector (m1 and m2) and those for the public sector (m3 and m4);the marginal propensities to consume out of wage and similar income (c1), income from capital (c2), and transfers (c3); the factor shares in national income (α, β, and γ); the dividend-payout ratio (δ); and the marginal propensities to invest out of corporate profits (i1) and out of capital transfers and loans from the government (i2).

In particular, the weights for public expenditure on consumption and investment goods reflect the corresponding import leakages (m3 and m4).

The weights for other current expenditure reflect the leakages into both imports and savings, namely, m1and (1 — c3) for current transfers to households, m1 and (1 — c2) for interest payments to the private sector, and m1 and a weighted average of the marginal propensities to save of the various recipients of factor incomes for current transfers to firms. Finally, the weights for capital expenditure reflect the leakages induced by the presence of an import component in private investment (m2) and by the marginal propensity of the private sector to hoard capital transfers or loans from the government (1 — i2).

Taking next the revenue items, under the normal assumption that the propensity to consume out of wage income (c1) is larger than that out of other incomes, one would expect the personal tax on wage income to have a larger (negative) impact on GNP than the tax on other incomes and larger yet than the corporation income tax, provided that the dividend-payout ratio (δ) is less than 1. In this model, indirect taxes20 are assumed to be shifted backward onto factor incomes, rather than onto prices.21 Therefore, they affect consumption through a weighted average of the various marginal propensities to consume and investment through corporate profits.

Current and capital transfers between the government and the rest of the world obviously have zero impact on GNP and one-to-one impact on the basic balance of payments. Government expenditure on goods and services affects the balance of payments directly through its import component (m3Gc and m4Gi).22 Revenue items affect the private sector’s disposable income, and therefore the balance of payments, through the import content of private consumption and investment.

Weights for discretionary changes, derived by solving Model II for GNP and the balance of payments, are presented in Table 2.23 They provide an indication of the comparative effectiveness of the various fiscal policy instruments in terms of their GNP impact. These weights are the same as those derived from Model I except for some that are smaller because of the dampening effect of the built-in stabilizers represented by positive marginal effective tax rates (tw, tp, tc and Ti). The weights from the two models would be equal, of course, if the marginal effective tax rates were zero, that is, if taxes were treated as exogenous.

Table 2.

Weights for Discretionary Budget Changes

article image
Source: Model II.

σ* = C1(α + γ)(1–tw) + C2(1–tp[1–α–γ–β+δβ(1–tc)]

Tf is here taken to be exogenous. In actual country models, it could be explained by using the information available as to the types of tax paid by nonresidents; the weight then might differ from 1.

III. Case Studies24

Italy

This case study analyzes the effects of fiscal developments for the Central Government as well as the public sector as a whole, defined to include the Central Government, local authorities, and the social security system.

The model used for deriving the weights for the various budget items for Italy fits broadly into the framework of analysis laid out in the previous sections. However, some simplifications were made necessary by limitations in the availability of data. In particular, it was not possible to estimate separate marginal propensities to consume for the different types of income, since no published information exists on the distribution of the personal income tax by income shares in Italy. For reasons of multicollinearity, it was not possible to obtain separate significant estimates of the marginal propensities to import for the government sector. For lack of better information, it was assumed that the import component is the same for public consumption and investment as for private consumption.

The empirical estimates of the parameters were derived from the following ordinary least-squares regressions (equations (la) to (4a)) for the sample period 1955–71. The results of these regressions are fairly satisfactory from the econometric point of view: 25

C=781.1+(1.9) 0.78YD(63)(1a)R¯2=0.997DW=1.93
I=106.8(0.1)+0.56(11.8)(PTcSkNL)(2a)R¯2=0.902DW=1.37
M=1.045(10.1)+0.14(6.2)(C+Gc+GI)+0.51(6.0)I(3a)R¯20.994DW=1018
Sc=0.12(8.41)+0.92(76.9)(PcTc)(4a)R¯2=0999DW=1.42

where the notation is the same as that used in Section II. In addition, YD = disposable income;26 P = profits of corporate and unincorporated enterprises; Sc = undistributed profits of corporations;Td = personal income taxes; Trh = current transfers to the government from the private sector; R2 = multiple correlation coefficient corrected for degrees of freedom; and D-W=Durbin-Watson statistic. The numbers in parentheses under the coefficients are the t-ratios.

The weights indicating the first-round impact on GNP and the balance of payments are estimated by using the parameters of equations (1a) to (4a) and are presented in Table 3.

Table 3.

Italy: Weights for Budget Items

article image

Includes social security contributions.

Includes other nonprofit institutions.

This item is zero for general government transactions.

In deriving the WBB for the Central Government, current transfers to other public authorities were given a weight of 0.8 (as an average of the weights for current expenditure) and capital transfers a weight of 0.86 (the same as for general government capital expenditure), on the assumption that capital transfers to other public authorities are used by them to finance their capital expenditure.

Includes equity participation in public enterprises.

Indirect taxes receive the same weight (–0.68) as direct taxes on households because the effects of indirect taxes on prices are not taken into account in the model and no separate marginal propensities to consume are estimated for the various types of income. On the other hand, corporate taxes receive a substantially lower weight (–0.33) than other taxes, reflecting the assumption that they are borne by corporations and their main impact, therefore, is on investment. Among expenditure items, wages and salaries receive the largest weight (1.0), followed by current purchases and gross capital formation (0.86), which receive equal weights because the marginal import propensity is assumed to be the same for both. The lower weight for current domestic transfers and interest payments (0.68) reflects the savings leakage. Finally, the low weight (0.28) for capital transfers to the private sector and for net lending, defined to include equity participation in state enterprises, reflects the relatively high import propensity for investment goods and the apparently low sensitivity (0.56) 27 of private investment to fiscal instruments, irrespective of whether the latter are capital transfers, loans, or corporate taxes.

Tables 4 and 5 present for the Central Government and the public sector, respectively, the weighted and unweighted budget balances and their changes from one year to another for the years 1965–71; 1965 was chosen as the starting year for the analysis mainly because until that year the Italian authorities made no significant efforts to pursue an active policy of demand management through fiscal instruments.

Table 4.

Italy: First-Round Impact of the Central Government Budget on Gross National Product (GNP) and the Balance of Payments, at Current Prices, 1965–71

(In billions of lire)

article image
Sources: Ministero per il Bilancio e la Programmazione Economica e Ministero per il Tesoro, Relazione Generale Sulla Situazione Economica del Paese, 1965–71.

Includes net lending.

Deficits are given a positive sign, since they have a positive impact on GNP.

A positive value indicates an increase in the deficit.

TABLE 5.

Italy: First-Round Impact of the Public Sector Budget on Gross National Product (GNP) and the Balance of Payments, at Current Prices, 1965–71

(In billions of lire)

article image
Sources: Ministero per il Bilancio e la Programmazione Economica e Ministero per il Tesoro, Relazione Generale Sulla Situazione Economica del Paese, 1965–71.

Excludes net lending.

Deficits are given a positive sign, since they have a positive impact on GNP.

A positive value indicates an increase in the deficit.

Throughout the period, the WBBs for both levels of government show positive first-round contributions to GNP. However, if their year-to-year changes, rather than their absolute values, are taken as indicators of the character of fiscal policy, there appear to have been substantial swings in the direction of policy, especially at the central government level. In what follows, the budget will be considered to be expansionary in any given year if the change in the WBB is positive in that year, and contractionary if it is negative.

It is useful to discuss budgetary developments against the background of cyclical developments in the economy. At the beginning of 1965, the economy was at the trough of a major recession that had started almost two years earlier. The recovery got under way in 1965, gained momentum in 1966, and reached a peak in 1967. The growth rate declined in 1968, remained almost unchanged in 1969, and declined again sharply in 1970 and 1971. The budgets of both the Central Government and the public sector were strongly expansionary in 1965. In 1966 the budget became mildly contractionary for the public sector as a whole, while remaining expansionary for the Central Government. It exercised a clearly contractionary impact in 1967 when the economy was at the peak of the boom and turned expansionary again in 1968 in an attempt to counteract the incipient recession. The character of the policy differed markedly between the two levels of government in 1969 and 1970, with the budget of the Central Government showing a much larger swing than that of the public sector. While the latter was mildly expansionary in 1969 and almost neutral in 1970, the former was contractionary in 1969 and then again strongly expansionary in 1970. In 1971, with the gravity of the recession appearing in full evidence, the budget was clearly expansionary at both levels.

These differences in the effects of the budget at the two levels of government reflect different budgetary developments, on the one hand, and differences in the weighting schemes, on the other hand. In particular, since no information is available on net lending by the public sector as a whole, this item is not taken into account in computing the WBB for the public sector while it is included in the WBB for the Central Government, so that changes in net lending, which were substantial over the years 1969–71, explain, at least in part, the different behavior of the measure of fiscal impact at the two levels over those years. Another factor contributes to explain this difference: transfers from the Central Government to other levels of government are consolidated in the overall public sector accounts and are, therefore, replaced by other types of government expenditure that have different weights.

A comparison of the behavior of the WBB with that of the simple deficit over the period 1965–71 suggests the following considerations:

(1) For the Central Government (Table 4), it appears that the weighted deficit was larger than the unweighted deficit over the years 1965–69 and smaller in 1970 and 1971. This was so because capital transfers and net lending, which in the present weighting scheme receive the smallest weight, increased much more rapidly than other categories of expenditure over the past two years. These items have the largest28 (—0.28) impact on the balance of payments, and therefore their rapid increase in 1970 and 1971 serves to explain the negative sign of the direct impact of the budget on the balance of payments in those years. In the other years, budgetary deficits exerted a favorable impact on the basic balance of payments. However, increases in the WBB were usually accompanied by a worsening of the balance of payments impact, and the reverse was true for decreases.

(2) For the public sector as a whole (Table 5), the WBB was always larger than its unweighted counterpart because the average weight for expenditure was larger, in absolute value, than that of revenue over the whole period. The two measures were essentially in agreement, however, as to direction and order of magnitude of changes of the budgetary impact from year to year. Basically, this reflected the fact that the composition of the public sector budget changed relatively little, especially on the revenue side, over the years 1965–71, as can be seen from Table 6.

Table 6.

Italy: Structure of the Public Sector Budgets, 1965–71

article image
Sources: Ministero per il Bilancio e la Programmazione Economica e Ministero per il Tesoro, Relazione Generale Sulla Situazione Economica del Paese, 1965–71.

Gross national product.

Sum of percentages does not add up to 100 because some minor revenue and expenditure items are omitted.

For wages and salaries, 1.0; for other purchases, 0.86.

Finland

In this case study, the public sector is defined to include the local authorities as well as the Central Government and the social security funds. A practical reason for limiting the analysis to the general government here is that up-to-date national accounting data are available only for the public sector as a whole. However, local authorities in Finland enjoy a considerable degree of fiscal autonomy stemming mainly from their ability to determine local tax rates. It is true that local tax revenues cover only about two thirds of local government expenditure, and the local authorities are thus dependent to a great extent on transfers from the Central Government and on borrowing. Nonetheless, because these transfers are generally used to finance current expenditure, the degree to which the Central Government can exert effective control is limited.29 In recent years, however, considerable effort has been made to improve coordination and planning for the public sector as a whole, as it is recognized that decisions about the central government budget must be taken within a framework of the prospective impact of the public sector as a whole.

The model (equations (1b) to (7b)) used in deriving the weights is an adaptation of the general model presented in Section II. For the purpose of this case study, some of the structural parameters of the model have been assumed, since a direct statistical estimation of the model was not possible owing to the lack of the necessary data; however, attempts have been made, using other econometric work, to check the reasonableness of the assumptions. The marginal consumption propensities are assumed to vary, depending on the different types of income. The marginal propensity to consume is assumed to be 1.0 for transfer receipts of households and was estimated to be 0.82 for all other household incomes by regressing household consumption expenditure (less transfer receipts of households) on household disposable income (less transfers) (equation (2b)).30 The marginal propensity to invest was estimated to be 0.87 by regressing private investment in manufacturing industries (Im) on the sum of corporate profits (net of corporate taxes) and net capital transfers to the private sector (equation (3b)).31 It would have been desirable to take account of net lending of the public sector as an important explanatory factor affecting private (residential) investment, but, unfortunately, figures for net lending are not available for a number of years. It was not possible to calculate statistical estimates for separate marginal import propensities. Instead, a marginal import propensity of 0.10 was assumed to be plausible for both private and public consumption (mi). Finland, with its relatively narrow capital goods sector, can be expected to have a rather high marginal propensity for investment expenditure, but the propensity is likely to be considerably lower for public investment expenditure (m3) than for private investment expenditure (m2), since a considerable part of the former consists of such projects as land reclamation and other infrastructural investments that have a relatively low import content. With these considerations in mind, it was assumed that m2 = 0.4 and m2 = 0.2 (equation (4b)). The data for 1964–72 indicate that the share (β) of corporate profits in net national income at factor cost less government income from property is about 0.07 (equation (6b)). The model32 that was used to derive the weights is therefore

Y=C+I+Gw+Gi+XM(1b)
CSh=6.5(4.5)+0.82(101.7)(W+Pu+Q+InoTdwTdpTrh)(2b)R2¯=0.999DW=2.53
Im=36.9(0.12)+0.87(4.51)(Pc+ShTc)(3b)R2¯=0.68DW=1.0
M=m0+0.10(C+Gc)+0.40I+0.20Gi(4b)
W+Pu+Q=0.93(YDEPTi+SfGIP)(5b)
Pc=0.07(YDEPTi+SfGIP)(6b)
BP=(XM)+TfGwf+(TrfSw)+(TrkfSkf)(7b)+(BfNLf)+SP+K

The solution of this model in terms of the various budget items provides estimates of weights indicating the first-round impact of these items on GNP and on the balance of payments. These weights are presented in Table 7. Among revenue items, the weight for personal direct taxes (and current transfers from the private sector) ranks highest (–0.74), followed closely by the weight for indirect taxes. The weight for corporate taxes is materially smaller (–0.52), since this tax is assumed to affect only private investment.

Table 7.

Finland: Weights for Budget Items

article image

Includes social security contributions.

Includes net contributions of public enterprises and rent, interest, and dividends from the private sector.

Capital transfers to and from the private sector are assumed to have a one-to-one impact on private investment.

The weights for expenditure are, as expected, much larger in absolute value than those for revenue items. The ranking of expenditure weights is, in general, the same as in the case study for Italy. There are, however, some exceptions. The weight for gross capital formation is lower than that for current purchases because of the assumed higher marginal propensity to import investment goods. Current transfers to households receive a weight (0.90) that is higher than that of household income from interest on the public debt (0.74) because the receivers of transfers are assumed to have a higher marginal propensity to consume.

A comparision of the unweighted and weighted budget balances 33 and their respective changes are presented in Table 8 for the years 1965–72. This comparison shows that, whereas the unweighted budget was in surplus for every year except 1965, the weighted budget was in deficit throughout the period.

Table 8.

Finland: First-Round Impact of the Public Sector Budget on Gross National Product (GNP) and the Balance of Payments, at Current Prices, 1965–72

(In millions of markkaa)

article image
Source: Central Statistical Office, National Accounting for Finland.

Deficits are given a positive sign, since they have a positive impact on GNP.

A positive value indicates an increase in deficit (a decrease in surplus).

On balance, and particularly in the later years, the effects of fiscal policies, judged on the basis of the changes in WBB, seem to have been appropriately countercyclical. In 1965, however, when activity was at a relatively high level, and capacity was highly utilized, the change in the WBB indicated an expansionary fiscal stance. Again in the following year, when the growth of output was below trend, the fiscal impact was mildly contractionary. Fiscal impact was only slightly expansionary in 1967, although the growth of output had continued to be sluggish. The changes in the WBB for subsequent years indicate that fiscal policy supported the revival in 1968 and then turned appropriately contractionary in the following two boom years. In 1971 the boom slackened, and fiscal policy changed again in an expansionary direction. In 1972 the change in the WBB indicated that fiscal policy was much more expansionary, helping to ensure that the trough of recession was very shallow. Particularly in 1967, and again in 1971, changes in the unweighted budget surplus would have given an incorrect impression of the impact of fiscal policies because current expenditure on goods and services, which have the highest positive impact on GNP, increased their share in total public sector expenditure (see Table 9).

Table 9.

Finland: Structure of the Public Sector Budgets, 1965–72

article image
Sources: Central Statistical Office, National Accounting for Finland; Table 7.

Gross national product.

Sum of percentages does not add up to 100 because some minor revenue and expenditure items are omitted.

The direct impact of fiscal policy on the balance of payments has been generally favorable, except in 1965, but it was more favorable in the years in which the budget exerted a contractionary influence on GNP. The indirect repercussions on the balance of payments of the budget impact on GNP in 1967, 1968, 1971, and 1972 may have reduced, if not outweighed, the direct impact, while in the other years they are likely to have reinforced it.

Austria

For this case study, data are available to permit analyzing fiscal developments at both the federal and total public sector levels. As for Finland, the scope for coordinated fiscal policy between the two levels is quite limited in Austria. The state and local governments enjoy a considerable degree of fiscal independence, and the arrangements for coordinating their fiscal policies with that of the Federal Government generally have been informal and voluntary. However, the influence of the Federal Government has been increasing, and officials of the local governments now hold discussions on fiscal policy decisions with the Federal Ministry of Finance on a regular basis.

In this case study no attempt has been made to estimate any of the parameters econometrically, but, again, attempts have been made to check the assumptions used. The form of the model (equations (lc) to (7c)) used in deriving the weights for Austria is basically the same as that used in the case study for Finland (equations (lb) to (7b)) with the exception of equations (2c) to (4c). In the consumption function (equation (2c)), recipients of government transfers are assumed to have a marginal propensity to consume of less than 1. In the investment function (equation (3c)), a one-to-one correspondence is assumed between private investment and net capital transfers to the private sector. The import function (equation (4c)) takes into account the probability that the marginal propensity to import is relatively small for public consumption expenditure. For investment it has been suggested that the marginal propensity of the private sector to import is about double that of the public sector.

The model used for Austria is as follows:

Y=C+I+Gc+Gw+Gi+XM(1c)
C=c0+0.75(W+Pu+Q+IngTdwTdpTrh)+0.90Sh(2c)
I=i0+0.70(PcTc)+Sk(3c)
M=m0+0.30C+0.40I+0.10Gc+0.20Gi(4c)
W+Pu+Q=0.90(YDEPTi+SfGIP)(5c)
Pc=0.10(YDEPTi+SfGIP)(6c)
BP=(XM)+TfGwf+(TrfSw)+(TrkfSkf+(BfNLf)+SP+K(7c)

The solution of this model provides the weights that are presented in Table 10. The ranking of the weights is similar to that in the case study for Finland, although their absolute values are notably different.

Table 10.

Austria: Weights for Budget Items

article image

Includes employees’ social security contributions.

Government wage payments are lumped together with other current expenditure on goods and services because separate information on them is not available in published sources.

Including other nonprofit institutions.

Zero for the public sector as a whole.

For both the Federal Government and the total public sector, the unweighted budget balance showed a surplus in every year from 1966 to 1972 except in 1967 and 1968 (Tables 11 and 12). The WBB, on the other hand, showed a deficit in every year. The impact of fiscal policy, as indicated by the changes in the WBB, remained expansionary throughout the period but varied considerably in size. In 1967, when the economy was still in a recessionary stage, the policy was strongly expansionary. Subsequently the Austrian economy enjoyed the longest period of sustained growth since World War II and the expansionary impact was smaller, whether measured at the federal or the total public sector level.

Table 11.

Austria: First-Round Impact of the Federal Budget on Gross National Product (GNP) and the Balance of Payments, at Current Prices, 1966–72

(In billions of schillings)

article image
Sources: Amtsbehelf zum Bundesfinanzgesetz, 1972, Vol. 1, pp. 307–31; Table 10.

Estimates.

Deficits are given positive signs, since they have positive impact on GNP.

A positive value indicates an increase in deficit (a decrease in surplus).

Table 12.

Austria: First-Round Impact of the Public Sector Budget on Gross National Product (GNP) and the Balance of Payments, at Current Prices, 1966–72

(In billions of schillings)

article image
Sources: Amtsbehelf zum Bundesfinanzgesetz, 1972, Vol. 1, pp. 307–31; Table 10.

Estimates.

Deficits are given positive signs, since they have a positive impact.

A positive value indicates an increase in deficit (a decrease in surplus).

The direct impact of fiscal policy on the balance of payments was favorable for every year except 1967, despite the fact that the first-round effects on GNP were expansionary. The indirect repercussions on the balance of payments of the expansionary effect of the budget on GNP are likely to have substantially reduced, if not outweighed, this direct impact.

IV. An Evaluation of the Approach

The WBB approach presented here offers a compromise between two requirements that are often in conflict with each other. On the one hand, the tool of fiscal analysis that is used should be conceptually simple and capable of being estimated easily from generally available empirical information. On the other hand, it should provide an accurate measure of the impact of the budget on other major economic aggregates.

While the WBB approach requires considerably more information and analysis than that based on the simple budget balance, it seems more satisfactory than the latter because it allows for the different impact of the various budget items and, therefore, for the effects of changes in the composition of the budget over time. If the weights are significantly different from one another and substantial shifts occur in the composition of the budget from one period to the next, changes in the WBB are a more reliable indicator of the character of fiscal developments than are changes in the simple budget balance. This is borne out clearly by the case studies, which show that the changes in the two balances can be different not only in size but also in sign.34 Another advantage of the WBB approach is that it permits a separation of the budget impact on GNP from that on the balance of payments. In this paper, only the direct impact of the budget on the basic balance of payments has been taken into account; however, the approach can easily be extended to include the repercussion on the balance of payments of the first-round impact of the budget on GNP.

Whether or not the WBB approach should be preferred in any given case to a full-multiplier analysis, based on a complex econometric model, depends mainly on two factors: (1) data or time constraints, which often prevent the construction and simulation of more realistic macroeconometric models, and (2) the purpose of the analysis. Being limited to first-round effects, the WBB is not suitable for an analysis aimed at quantifying the effects of the budget over time. For the latter purpose, there seems to be no satisfactory substitute for a full-multiplier analysis using a model into which appropriate lag structures have been introduced. The WBB approach appears most appropriate when the aim of the analysis is to evaluate, in broad terms, the expansionary or contractionary character of a given budget or of a set of fiscal policy measures.

The accuracy of the WBB as a measure of the first-round impact of the budget obviously depends on the merits of the model from which the weights are derived. The models presented in this paper undoubtedly have several limitations, some of which make them quite unsatisfactory for analyzing the effects of fiscal policies in certain economic conditions. A major limitation is the fact that prices do not appear explicitly in any of them. Therefore, the models are more suitable for analyzing fiscal policy in recessionary than in inflationary conditions. When fiscal policy is used primarily as an anti-inflationary weapon, an attempt should be made to specify the main channels through which fiscal changes are likely to affect prices. This is likely to increase the complexity of the exercise and to require computer simulation of the model.

A further shortcoming of the models is the fact that they do not take into account the monetary repercussions of fiscal changes. Because of the existence of the budget constraint,35 changes in the budget balance must result in borrowing from either the private sector or the central bank. In borrowing from the private sector, there are likely to be repercussions on interest rates (and, therefore, possibly on investment) and on the stock of private wealth (and, therefore, possibly on aggregate consumption). In borrowing from the central bank, the monetary base changes and there are likely to be further effects on prices, interest rates, and wealth. If these interrelationships between fiscal, monetary, and other macroeconomic aggregates were incorporated explicitly into the models, the weights derived from their reduced forms would, of course, be different from the ones presented here and would provide better estimates of the first-round impact of the various budget items. This appears to be one of the more promising areas of research for improving the WBB approach.

*

Mr. Borpujari, economist in the Western Division of the Middle Eastern Department, is a graduate of the Universities of Madras and Delhi and received his doctorate in economics from Cambridge University. Formerly he taught at Ramjas College of the University of Delhi.

Mrs. Teresa Ter-Minassian, a graduate of the University of Rome and of Harvard University, is an economist in the Fiscal Analysis Division of the Fiscal Affairs Department. She is presently on leave from the central bank of Italy.

In addition to colleagues in the Fund, the authors are indebted to Michael V. Posner for helpful comments.

1

For a survey of techniques of fiscal analysis, see Joergen Lotz, “Techniques of Measuring the Effects of Fiscal Policy,” OECD Economic Outlook: Occasional Studies (Paris, July 1971).

2

For a discussion of different concepts of budget balance used in the literature, see Raja J. Chelliah, “Significance of Alternative Concepts of Budget Deficit,” Staff Papers, Vol.XX (1973), pp. 741–84.

3

For a survey of fiscal policy models, see Alan Peacock and G. K. Shaw, The Economic Theory of Fiscal Policy (London, 1971).

4

In symbols:

WBB=Σj=1mWiEiΣi=1nWiRi
andΔWBB=Σj1mWiΔEiΣi=1nWiΔRi

where

Wj is the jth expenditure weight

Wi is the ith revenue weight

Ej is the jth budgetary expenditure item

Ri is the ith budgetary revenue item

j = 1, 2, …, m

and i = 1, 2, …, n

where m and n, respectively, stand for the total number of expenditure and revenue items in the given budget.

5

The average expenditure (revenue) weight is derived by multiplying the weights of the various expenditure (revenue) items by their respective shares in total expenditure (revenue) and totaling the products. In symbols:

AEW=ΣjWiEi/E

where AEW is the average expenditure weight and E is total expenditure.

6

If the import is effected through a domestic firm, there is some positive effect on domestic output and employment; however, this effect in any case will be a relatively small part of the total impact of government expenditure abroad.

7

For an analysis of various types of lag in fiscal policy, see Albert Ando, E. Carey Brown, Robert M. Solow, and John Kareken, “Lags in Fiscal and Monetary Policy” (Part Π: “Lags in Fiscal Policy,” by Albert Ando and E. Carey Brown), in Stabilization Policies, Commission on Money and Credit (Englewood Cliffs, New Jersey, 1963), pp. 97–163.

8

For another example of the use of a static macro-model in the assessment of the first-round impact of the budget on GNP and the balance of payments, see Lotz, op. cit., pp. 25–27.

9

This criticism applies to Bent Hansen’s multiplier analysis in Fiscal Policy in Seven Countries, 1955–1965, Organization for Economic Cooperation and Development (Paris, 1969). An analysis that utilizes the Hansen model but is limited to first-round effects can be found in Assar Lindbeck, “Fiscal Policy as a Tool of Economic Stabilisation—Comments to an OECD Report,” Kyklos, Vol. XXIII (1970), pp. 7–32.

10

See, for example, R. A. Musgrave, “On Measuring Fiscal Performance,” The Review of Economics and Statistics, Vol. XLVI (1964), pp. 213–20; W.A.B. Hopkin and W. A. H. Godley, “An Analysis of Tax Changes,” National Institute Economic Review, National Institute of Economic and Social Research (May 1965), pp. 33–42; D. A. L. Auld, “A Measure of Australian Fiscal Policy Performance, 1948–49 to 1963–64,” The Economic Record, Vol. 43 (1967), pp. 333–53; Edward M. Gramlich, “Measures of the Aggregate Demand Impact of the Federal Budget,” in Budget Concepts for Economic Analysis, ed. by Wilfred Lewis, Jr. (The Brookings Institution, Washington, 1968), pp. 110–27; Elliott R. Morss and Alan T. Peacock, “The Measurement of Fiscal Performance in Developing Countries,” in Quantitative Analysis in Public Finance, ed. by Alan T. Peacock (New York, 1969), pp. 171–97; Michael Artis, “Fiscal Policy for Stabilization,” Chapter 8 in The Labour Government’s Economic Record: 1964–1970, ed. by Wilfrid Beckerman (London, 1972), pp. 262–99; and House of Commons, Seventh Report from the Expenditure Committee: Public Expenditure and Economic Management (London, February 15, 1972), p. 18. The last two references deal with the first-round effects, while all the others use a full-multiplier analysis.

11

Nicholas Kaldor, Essays on Value and Distribution (London, 1960), pp. 227–36; and Luigi L. Pasinetti, “Rate of Profit and Income Distribution in Relation to the Rate of Economic Growth,” The Review of Economic Studies, Vol. XXIX (October 1962), pp. 267–79.

12

Although there is evidence that changes in the distribution of income by types affect aggregate consumption, sometimes it may not be possible to get econometric estimates of the various marginal propensities to consume because of the lack of data on the distribution of personal income taxes by types of income.

13

When information is available on the direct import content of government expenditure, it may be preferable to subtract it from total imports and to introduce it separately into the balance of payments equation. The coefficients of the government expenditure items will then reflect only their indirect import content.

14

If the assumption of constancy appears to be inconsistent with empirical evidence in any particular case study, α, β, and γ can be taken to be (exogenous) variables. This, of course, implies that the weights may vary from one year to another.

15

Other incomes are, of course, determined residually from equations (6) to (9):

Q=(1αβγ)(YDEPTi+SfGIP)
16

See, for instance, Roger E. Bolton, “Predictive Models for State and Local Government Purchases,” in The Brookings Model: Some Further Results, ed. by James S. Duesenberry and others (Chicago, 1969), pp. 221–67, and John F. Helliwell and others, Government Sector Equations for Macroeconomic Models, Bank of Canada, Staff Research Study, No. 4 (Ottawa, 1969).

17

Variables are listed in order of first appearance, and all values are expressed in current prices.

18

Tdw, Tdp, Tc and Ti, are endogenous in Model II.

19

The multiplier, derived from Model I, is 1/[1(1m1){c1α+c2[1αγβ(1δ)]}(1m2)i1β].

20

The same assumption is made for subsidies, which, therefore, receive a weight equal in absolute value (and opposite sign) to that of indirect taxes.

21

An attempt was made to take into account the effect of indirect taxes (net of subsidies) on prices by assuming that they are entirely shifted onto consumers and by introducing into the model a price equation reflecting this assumption, such as

Pc=P+TiSfC

where Pc = consumer price index

P = general price index net of indirect taxes

C = aggregate private consumption

This formulation, however, had the disadvantage of introducing a nonlinearity that made its algebraic solution quite complex, and it was therefore eventually abandoned in the model discussed in the text.

22

Ideally, indirect budgetary effects on exports and on private capital flows should also be taken into account in specifying the model. However, these effects are ignored in the present exercise because of the difficulty of estimating them empirically.

23

The multiplier, derived from Model II, is1/[1(1m1)](1ti){c1(α+γ)(1tw)}+c2(1tp)[1αβγ+δβ(1tc)]}(1m1)i1(1tc)(1ti)].

24

Model II is not used in the case studies because reliable information on the revenue impact of discretionary changes was not available.

25

The various functions show high R2s and t-ratios that are well above the 1 per cent significance level. There is little evidence of serial correlation.

26
YD=YDEPTi+SfGIPTcScTdTrf+Sh+Ing
27

The low coefficient is probably related to two factors: first, only recently the authorities have pursued an active policy of stimulating investment through fiscal instruments, especially in the form of increased equity participation in public enterprises; second, the corporation tax is levied at rather low legal rates and the effective rate is even lower because of widespread tax evasion.

28

In absolute value and excluding the direct government transactions abroad, which constitute a very small proportion of the budget.

29

For more details on the Finnish public sector, see Eino H. Laurila, “On the Development of the Public Sector Since the War,” Unitas (Helsinki), Vol. 41 (1969), pp. 71–80; and Organization for Economic Cooperation and Development, “Annex I: The Use of Fiscal Policy in Demand Management,” Economic Surveys—Finland (Paris, May 1970), pp. 37–42.

30

Equation (2b) refers to the annual data for 1958–72.

31

Equation (3b) refers to the annual data for 1961–72 excluding 1964 and 1965; these years were excluded because their capital transfer data were exceptionally large. The D-W statistic is low, but, considering the small size of the sample, it does not necessarily indicate autocorrelation.

32

The notation is the same as in Section II and in the case study for Italy.

33

The exclusion of net lending to the private sector gives a downward bias to these figures. Net lending could be included in the analysis by assigning to it the same weight as capital transfers, but this is not essential for the present purpose of illustrating the WBB approach to fiscal analysis.

34

See the data covering 1969, 1970, and 1972 for Austria and 1967 and 1971 for Finland.

35

The first systematic treatment of the implications of the existence of a government budget constraint can be found in Carl F. Christ, “A Simple Macro-economic Model with a Government Budget Restraint,” The Journal of Political Economy, Vol. 76 (1968), pp. 53–67. A more recent contribution is by Bent Hensen, “On the Effects of Fiscal and Monetary Policy: A Taxonomic Discussion,” The American Economic Review, Vol. LXIII (September 1973), pp. 546–71, where he extends his previous OECD model to incorporate a simple monetary sector and a budget constraint, in order to analyze, along with pure fiscal effects, the monetary repercussions of budgetary changes.

  • Collapse
  • Expand