Fiscal Rules and Countercyclical Policy
Frank Ramsey Meets Gramm-Rudman-Hollings1
Author: Mr. Evan C Tanner

Contributor Notes

Author’s E-Mail Address: etanner@imf.org
Publication Date:
01 Nov 2003
eISBN:
9781451875225
ISBN:
9781451875225
Language:
English
Keywords:
WP; government expenditure; accumulation average; end-period debt bj; debt ratio; primary surplus; endogenous expenditure regime; tax smoothing
Download PDF (238 KB)

Fiscal rules—legal restrictions on government borrowing, spending, or debt accumulation (like the Gramm-Rudman-Hollings Act in the United States)—have recently been adopted or considered in several countries, both industrial and developing. Previous literature stresses that such laws restrict countercyclical government borrowing, thus preventing intertemporal equalization of marginal deadweight losses of taxation—an idea associated with Frank Ramsey. However, such literature typically abstracts from persistent current deficits that are financed by future tax increases. Eliminating such deficits may substantially reduce tax rate variability—the very goal of countercyclical borrowing—even over a finite horizon. Thus, Gramm-Rudman-Hollings and Frank Ramsey are not necessarily enemies and they may even be good friends!

Abstract

Fiscal rules—legal restrictions on government borrowing, spending, or debt accumulation (like the Gramm-Rudman-Hollings Act in the United States)—have recently been adopted or considered in several countries, both industrial and developing. Previous literature stresses that such laws restrict countercyclical government borrowing, thus preventing intertemporal equalization of marginal deadweight losses of taxation—an idea associated with Frank Ramsey. However, such literature typically abstracts from persistent current deficits that are financed by future tax increases. Eliminating such deficits may substantially reduce tax rate variability—the very goal of countercyclical borrowing—even over a finite horizon. Thus, Gramm-Rudman-Hollings and Frank Ramsey are not necessarily enemies and they may even be good friends!

I. Introduction

Fiscal rules—legal restrictions on government borrowing, spending, or debt accumulation— have recently been adopted in several countries and are being discussed in several others (both industrialized and developing). While the details of fiscal rules may differ across countries, debates regarding their adoption involve similar issues.

Opponents of fiscal rules emphasize that they prevent the government from smoothing tax rates and expenditures over the business cycle, and may even prohibit discretionary countercyclical policy2. By contrast, their proponents argue that fiscal rules supplement weak institutions to promote fiscal responsibility and credibility.3 This issue may be especially important in those Latin American countries that have suffered from chronic fiscal indiscipline.4, 5

Economic theory should be able to help policy makers evaluate alternative fiscal policies, including legal restrictions on fiscal policy (balanced budget restrictions or debt ceilings). A sizeable literature already examines fiscal policy from a welfare theoretic perspective. Much of this work builds on Frank Ramsey’s (1927) insight that governments should equate the marginal deadweight losses from different tax sources. Robert Barro (1979) applied this insight to determine the optimal path of tax rates over time. He concluded that deadweight losses would be minimized under a policy of tax smoothing. Moreover, under such a policy, fiscal policy is optimally countercyclical in the sense that the government is permitted to borrow during economic recessions (but must save during upturns).6

In this vein, several recent papers have compared a balanced-budget fiscal rule like the U.S. Gramm-Rudman-Hollings (GRH) Act with the optimal (Ramsey) policy discussed above.7 Unsurprisingly, such research has generally confirmed that welfare under the Ramsey policy is higher than under the more restrictive fiscal rule.8

However, it may be relevant to compare such a balanced-budget restriction against a broader range of policies. As one alternative, a government might run primary deficits today but is nonetheless expected to finance its debt service with future surpluses. Since tax rates are not smoothed over time, such a policy is not optimal. Nonetheless, such a policy may more closely resemble those of actual governments than the optimal (Ramsey) policy. How would a country’s tax rates—level and variability—change if the government replaced such a policy with a GRH-like fiscal rule? Which regime would the country’s residents prefer?

This paper compares a restrictive fiscal law—one that prohibits government from borrowing more than the minimum required to keep debt-GDP constant—against several policy alternatives.9 Under a benchmark policy, tax rates are completely smooth. Also, a general fiscal reaction function that may resemble more closely policies of actual governments is considered. This policy links tax rates to debt (thus ensuring long-run solvency) but also allows for a constant (potentially deficit) component.

Uncertainty from two sources is assumed: cyclical output and access to credit. However permanent output is known and the ratio of government expenditures to permanent output is constant. Under these assumptions, countercyclical fiscal policy is synonymous with smooth tax rates. Welfare is assumed to fall when either the mean or the variance of tax rates rises.

Previous literature has stressed that the restrictive nature of a fiscal rule like Gramm-Rudman-Hollings hinders countercyclical public borrowing and thus also hinders tax smoothing. By contrast, this paper notes that the removal of persistent current primary deficits—through either a fiscal rule or a once-and-for-all fiscal reform—permits smoother tax rates than otherwise.10

Over an infinite horizon, the welfare gains implied by moving closer to a Ramsey regime—lower and less variable tax rates—should be immediately apparent. However, over shorter horizons, the issue is not as clear-cut. If policy makers choose to finance some level of government expenditures today by accumulating debt, taxes today may be lower, but they will also be more variable. Simulations presented in this paper suggest the magnitude of this tradeoff.11

Thus, under certain conditions, a fiscal rule may reduce tax rate variability—precisely the goal of countercyclical borrowing—even over a finite horizon. Put differently, Frank Ramsey and Gramm-Rudman-Hollings are not necessarily enemies. In fact, they may be good friends!12

The paper is organized as follows. Section II presents basic identities and discusses optimal fiscal policy over an infinite horizon under certainty. A general fiscal rule is introduced. Section III presents an alternative model in which the authority may favor current taxpayers at the expense of future ones. In Section IV, uncertainty in output and borrowing restrictions are introduced. Section V presents the central analysis of the paper: alternative fiscal regimes are defined and simulated. Section VI extends the framework to include variable government expenditures. Section VII presents some evidence regarding public sector size and expenditure variability. Section VIII summarizes and concludes.

II. The Optimal Fiscal Rule: Certainty, Infinite Horizon

In any period, the government’s budget constraint is:

bt−1θ−pst=bt(1)

where b is the ratio of government debt to GDP, θ is the growth adjusted discount factor (1+r)/(1+ λ), r = real interest rate (constant), λ = permanent real GDP growth, λ < r, pst is the primary (non-interest) surplus (ratio to GDP). The intertemporal budget constraint is obtained by successive substitution of (1) over an infinite horizon:

b−1−∑t=0∞pst/θt−1=lim t→∞bt/θt−1(2)

The transversality (or “no-Ponzi game”) condition is:

lim t→∞bt/(1+r)t−1=0(3)

The primary surplus is the difference between tax ratio τ and noninterest expenditures γ. For convenience, we assume a constant and exogenous expenditure ratio.13

A general expression for fiscal regimes—including legally stipulated fiscal rules—is now introduced. Tax rates, primary expenditures and debt are linked according to:

τt=γ−к+βbt−1(4)

Fiscal policy is therefore summarized by the government’s choice of κ and β (for given values of initial b−1, and constants γ. λ and r). The term κ may be thought of as a “tax gap” (Blanchard et al. (1990)) reflecting either a decision to keep taxes low or inefficient tax collection.

In any period, losses are assumed to increase in both first and second derivatives: ϕ′>0, ϕ″>0.14 Thus, the government choose a path of tax rates over time τt that minimizes the discounted loss function subject to (3):15

∑t=0∞ϕ(τt)θ−t(5)

The intertemporal first order condition is: ϕ′ (τt) = ϕ′ (τt+1) for all t. Thus, a result similar to Barro (1979) obtains. For an optimum, κ=0 and β=(r−λ)/(1+λ). Doing so ensures both tax-smoothing (τt = τt+1, all t) and that the debt ratio remains constant at b−1.

Note that this result resembles the widely-used policy framework associated with Blanchard et. al. (1990) and Talvi and Végh (2000a).16 Both papers note that sustainable fiscal policy (satisfaction of (3) without inflation or default) requires a permanent primary surplus of [r−λ]/[1+λ]*bP (κ=0 and β = [r−λ]/[1+λ]). However, strict satisfaction of (3) does not require κ = 0 and β = [r−λ]/[1+λ]. Rather, sustainability only requires β > 0. But, the analysis shows conditions under which κ = 0 and β = [r−λ]/[1+λ] is an optimal policy.17

III. Favoring the Present Over The Future?

Much evidence suggests that governments do not pursue policies that resemble κ = 0 and β = [r−λ]/[1+λ]. For example, Table 1 presents selected recent fiscal variables—the primary surplus and debt (in percent of GDP) and real GDP growth—for several Latin American countries. In several key cases—Argentina, Brazil, Colombia, Peru, Uruguay—government debt grows substantially, while primary balances are persistently below values consistent with κ = 0 and β = [r−λ]/[1+λ], and in some years negative.

Table 1.

Primary Surplus, Debt and GDP Growth Selected Latin American Countries

article image
Sources: IMF Staff Reports and Recent Economic Developments (various); for Central American and Caribbean Countries, Offerdahl (2002).

Governments may instead favor present over future taxpayers.18 If they do so, their loss functions would instead contain two distinct components: ϕp(τ) for the present (P, t = 0 to J) and ϕF(τ) for the future (F, t = J+1 to ∞). As before, ϕi′>0, ϕi″>0, i = P, F. As above, governments choose a path of tax rates τt to minimize

∑t=0JϕP(τt)θ−t+∑t=J+1∞ϕF(τt)θ−t(5′)

subject to (3). And, as before, within periods P and F, the authority sets ϕi′ (τt) = ϕi′ (τt+1), i = P, F. Across periods P and F, the rate at which the authority prefers P over F is summarized by an interperiod marginal of substitution, P and F, µPF. This number is assumed to be constant. Thus, for an optimum, the authority must equate the ratio of average tax rates between periods P and F τP/τF to the marginal rate of substitution µPF.

To see that problem (5′) is compatible with a fiscal rule like (4), suppose that the government gives a tax cut κP to period P but requires period F to pay a surcharge- κF. Sustainability thus requires:

κP=−κF/{θJ+1−1}(6)

For simplicity, assume β is constant. If τP/τF = µPF, the optimal “tax break” for current taxpayers κP* is:

κP*={γ(1−μPF)+β(bP−bF)}/{1+μPF(θJ+1−1)}(7)

That is, κP* is a function of µPF, γ, r,λ, β, and debt in both periods, bP and bF respectively.

IV. Optimization Under Uncertainty: Output and Access to Credit

Uncertainty is an essential aspect of most policy environments. This paper assumes uncertainty in both output and access to credit markets. For simplicity, both forms of uncertainty are assumed to be exogenous. However, even if the uncertainty were instead to be endogenous, the qualitative implications of the analysis would be similar.

First, output Y is the sum of its (certain) permanent and (uncertain) temporary components:

Yt=YtP+υt(8)

where YtP = YtP (1+λ) is permanent (trend) output, and υt is mean-zero temporary income (deviation from trend) whose known variance is constant relative to YtP. Note that government spending as a fraction of permanent income γ remains constant.

Second, a random element to credit markets is introduced. The country is assumed to face a cutoff from access to borrowing with probability πc which is uniformly distributed between 0 and 1. If πc = 0, the country will have access to borrowing with certainty. If πc is 0.5, there is a 50 percent chance that the country will not be able to borrow.

If a government is denied access to credit in a period that it would have otherwise have borrowed, it must raise taxes in that period.19 At the same time, a cutoff from borrowing has an asymmetric effect: it limits a country’s deficit but not its surplus. Thus, suppose the country follows the tax smoothing policy discussed in the previous section, but sets an (ad hoc) target level of debt bP. With unfettered access to credit, taxes in each period are simply τ(U) = γ + (r−λ)/(1+λ)bP. By contrast, if borrowing is constrained, taxes are linked to the level of debt in the previous period according to τ(C)t = γ + (r−λt)/(1+λt)bt−1 where λt is prospective total growth in period t. Thus, taxes in any period are the minimum of τ(U) and τ(C)t.

However, under such assumptions, κ=0 may no longer be optimal. Rather, if πc > 0, fiscal policy has a precautionary element: governments may want to self-insure against prospective borrowing cutoffs. As a simple two period example, suppose that a currently unconstrained government wishes to equate current (period 1) marginal deadweight loss with expected future (period 2) deadweight loss, by choosing the optimal surplus (r−λ)/(1+λ)b0−κ1* (κ1 < 0) in period 1. Therefore, κ1* will be chosen to satisfy:

ϕ′(τ(U)−κ1)=(1−πC) E{ϕ′(τ(U))}+πC E{ϕ′(τ(C)2)}(9)Current marginalFuture marginalFuture expecteddeadweight lossdeadweight loss,marginalunfettered accessdeadweight loss,to borrowingno access toborrowing

In (9), note that if πC =0, κ 1*=0. More generally, ∂ κ1*/∂ πC <0: as borrowing becomes more restricted (probalistically), the optimal primary surplus rises.20

To extend a framework like (9) to many periods also requires that an ad hoc debt target must be specified. Otherwise, the continually run surpluses and become a net creditor. Such a result is formally derived in Aiyagari, Marcet, Sargent and Seppälä (2002). Specifically, they show that under “natural” debt limits—“the maximum debt that could be repaid almost surely under an optimal tax policy”—the optimal tax rate will be zero, with expenditures financed entirely from interest receipts. By contrast, under an ad hoc debt limit, their model predicts that taxes will be positive and will follow a random walk, similar to Barro (1979)

V. A Comparison of Fiscal Regimes

As a practical matter, formulating a policy explicitly based on an optimizing framework may pose difficulties.21 For this reason, some have emphasized the importance of clear and simple policy rules—even if they are ad hoc. As discussed in further detail below, authors like Drazen (2000) have suggested that such simple rules—that is, balanced budget rules—may improve policy making since they provide a substitute for otherwise weak institutions.

In this section, three such rules are formulated and evaluated. First, under a benchmark policy regime (R0), tax rates are completely smooth (similar to optimizing under certainty) and the government is free to borrow over the business cycle while at the same time maintaining solvency. Second, under a restrictive fiscal law (R1), borrowing beyond the minimum required to keep debt-GDP constant is prohibited. Third, a general fiscal reaction function (R2) that may more closely resemble actual policies is considered. This policy links tax rates to debt (thus ensuring long-run solvency) but also allows for a constant (potentially deficit) component.

Successful implementation of (R0) may require strong institutions. However, if institutions are not sufficiently strong, and the authorities are not constrained by a fiscal law like (R1), they may instead choose a less restrictive fiscal regime like (R2).

As mentioned above, the loss function for taxation is assumed (in standard fashion) to increase in both first and second derivatives. Thus, losses increase with both the level and the variability of tax rates, and the authority will willingly trade off one against the other at some rate. In turn, tax rates and variability depend both on exogenous factors (that is, var(υt) and πc) and the choice of regime.

One way to compare regimes would be to assume a loss function with an explicit functional form. However, doing so may be unnecessarily restrictive. Instead, an alternative strategy is used: simulated values for the average and standard deviation of tax rates from different regimes are presented. Doing so thus yields the implicit marginal rate of substitution between tax rate levels and standard deviations at which the government would be indifferent between regimes.

A. The Regimes Defined

The fiscal regime that gives the government the most freedom to borrow over the business cycle while at the same time maintaining solvency is similar to the tax smoothing regime under certainty. The constant tax rate is:

τ(0)=γ+(r−λ)/(1+λ)bP(R0)

where bP is assumed to be equal to intial debt b−1. Access to borrowing is unfettered. To see that the debt remains in the long run close to bP and that solvency condition (3) is satisfied, note first that the primary surplus/GDP ratio in any period is:

ps(0)t=γ[1−wt]+(r−λ)/(1+λ)bP(10)

where wt = YtP/Yt is the ratio of permanent to total output in any period.22 Thus, the borrowing requirement beyond the minimum required to keep debt/GDP constant brt is:

br(0)t=γ[wt−1]+(r−λ)/(1+λ)[bt−1−bP](11)

Consider next a fiscal rule that explicitly ties today’s tax rates to the previous period’s debt. The tax rate is chosen when bt−1 is known but before Yt is known. According to (5) YtP is also the expected value of Yt. Thus, taxes are set according to:

τ(1)t=γ+(r−λt*)/(1+λt*)bt−1.(R1)

where λt* = [YtP/Yt−1 − 1] is expected output growth in any period. Thus, this rule aims ex-ante to maintain a constant debt/GDP ratio. Debt inherited from the previous period limits new borrowing more under (R1) than (R0). Ex post, governments may borrow or save.23 The primary surplus ratio in any period is:

ps(1)t=γ[1−wt]+[(1+r)/(1+λ*)]bt−1(12)

while the ex post borrowing requirement is:

br(1)t=γ[wt−1]+[(1+r)/(1+λ*)εt]bt−1(13)

where εt = (λt−λt*)/(1+λt) reflects the forecasting error in period t.

Regimes (R0) and (R1) appear similar: in both cases κ = 0 and β = r−λ/(1+λ). Therefore, average tax rates should be roughly equal across the two regimes. Tax rate variance is zero under (R0) but is positive under (R1). It is difficult to compare rule (R1) and (R0) without reference to institutional context.24 We therefore assume that rule (R1) is legally stipulated while (R0) is not. As Drazen (2000) argues, explicit rules can bolster otherwise weak credibility and institutions.

If institutions are weak and there is no fiscal rule, choice of bP may be problematic; there may be incentives to revise bP on a period-by-period basis. Without a rule like (R1), such revisions would likely be asymmetric: instead of adjusting, a government might simply raise its debt ceiling (bP).

However, (R0) may not be the relevant alternative to (R1). Instead, we may want to compare (R1) against a broader range of alternative fiscal regimes. Left without an explicit rule, a country’s institutional structure may not abide by optimality conditions κ = 0 and β = r−λ/(1+λ). Rather, the tax rule might take a more general form:

τ(2)t=γ−κ+βbt−1, κ>0, β>0.(R2)

While (R1) is a legal requirement, (R2) is not. As mentioned above κ > 0 reflects a tax gap. The primary surplus as a fraction of GDP is:

ps(2)t=−κ+βbt−1;(14)

Incremental new borrowing is

br(2)t=κ+[(r−λ)/(1+λ)−β]bt−1.(15)

B. Interaction Between Borrowing Constraints and Tax Regimes

As mentioned above, even under the constrained regime, countries may borrow in order to cover their forecasting error.25 Thus, under (R1), neither taxes nor borrowing will be affected. For regimes (R0) and (R2), if a borrowing constraint holds in a given period, borrowing is the maximum of what obtains under (R1) (namely, br(1)t = [wt −1] +[(1+r)/(1+λ*)εt]bt−1) and the amount that would have been borrowed otherwise; the tax rate is the minimum of τ(1)t = γ + (r−λ)/(1+λt *)bt−1 and the unconstrained rate.26 In this sense, governments benefit from past discipline: all else equal, lower debt levels imply smaller tax increases.

C. Simulation Results

It should be immediately apparent—without simulations—that over an infinite horizon, the level of tax rates under (R0) and (R1) should be close to one another, but taxes are more variable under (R1).27 By contrast, under a regime like (R2), with values of κ and β that differ from their infinite horizon optima (0, (r−λ)/(1+λ), respectively) tax rates will be higher and more variable than under either (R0) or (R1).

However, such a distinction is not as clear-cut over shorter horizons (finite J). Policy makers choose (R2) if they want to provide a given level of government expenditures today but delay taxfinancing until some future date.28 Doing so may noticeably increase both tax rate variability and debt accumulation over this shorter horizon. That is, as κ and β move further away from their infinite horizon optima, tax rate variability rises, thus frustrating the very goal of countercyclical borrowing. Whether or not the costs of increased tax rate variability exceed the benefits of lower tax rate levels depends on the marginal rate of substitution implied by loss function ϕ(τ). Also, as κ and β move further away from their optimal values, more debt is accumulated, an important factor if credibility is imperfect.

Tables 2 through 4 present simulations of regimes (R0), (R1) and (R2). These simulations are intended to convey a flavor of how such regimes differ for shorter horizons–5, 10, and 20 years. The tables show the mean, variance, minima and maxima for three key variables: the tax rate (τt), the primary surplus (ratio to GDP) (pst), and the end of period debt (bJ, J = 5, 10, 20). In all cases, 500 random draws are taken.

Table 2.

Alternative Fiscal Regimes: Simulation Results

Probability of borrowing constraint πc=0.0

article image
Notes: Unconstrained regimes are: (R0) smoothing: τ(0)= γ +(r−λ)/(1+λ)bP; ps(0)t = γ[1–wt]+(r−λ)/(1+ λ)bP; (Rl) near-balanced budget:τ(1)t= γ +(r−λt*)/(1+ λt*)bt−1, ps(1)t = γ[1–wt]+[(1+r)/(1+ λ*)]bt−1;(R2) general reaction: τ(2)t= γ–κ+βbt−1, κ >0, β>0,ps(2)t = –κ + βbt−1;Constrained regimes are: (R0): τ(0C)t = max{τ(1)t, τ(0)t}, ps(0C)t = max{ps(1)t, ps(0)t}; (R2): τ(2C)t = max{τ(1)t, τ(2)t}, ps(2C)t = max{br(1)t, br(2)t};πC = probability of constraint in any period, τ = tax rate, γ = ratio of government expenditures to permanent output, r = interest rate, λ = growth of permanent output,λ*t = expected output growth, b=ratio of debt to output, wt = ratio, permanent/total output in period t;For all simulations, λ = 4%, r = 7%, b(initial)=bP=0.5, γ= 0.2, variance of temporary output = 0.5 * permanent output. Number of draws = 500.
Table 3.

Alternative Fiscal Regimes: Simulation Results

article image
Notes: Unconstrained regimes are: (R0) smoothing: τ(0)=γ+(r−λ)/(1+λ)bp, ps(0)t = γ[1–wt]+(r−λ)/(1+λ)bP; (R1) near-balanced budget:τ(1)t=γ+(r−λt*)/(1+λt*)bt−1, ps(1)t = γ[1–wt]+[(1+r)/(1+λ*)]bt−1. (R2) general reaction: τ(2)= γ−κ+βbt−1, κ >0, β>0, ps(2)t = −κ + βbt−1;Constrained regimes are: (R0): τ(0C)t = max{τ(1)t, τ(0)J, ps(0C)t = max{ps(1)t, ps(0)t}; (R2): τ(2C)t = max{τ(1)t, τ(2)t} ps(2C)t = max{br(1)t, br(2)t};πC = probability of constraint in any period, τ = tax rate, γ = ratio of government expenditures to permanent output, r = interest rate, λ = growth of permanent output, λ*t = expected output growth, b=ratio of debt to output, wt = ratio, permanent/total output in period t;For all simulations, λ = 4%, r = 7%, b(initial)=bP=0.5, γ= 0.2, variance of temporary output = 0.5* permanent output. Number of draws = 500.
Table 4.

Alternative Fiscal Regimes: Simulation Results

article image
Notes: Unconstrained regimes are: (R0) smoothing: τ(0)=γ+(r−λ)/(1+λ)bP., ps(0)t = γ[1–wt]+(r−λ)/(1+λ)bP; (R1) near-balanced budget:τ(1)t=γ+(r−λt*)/(1+λt*)bt−1 ps(1)t = γ [1–wt]+[(1+r)/(1+λ*)]bt−1;(R2) general reaction: τ(2)= γ – κ+βbt−1, κ >0, β>0,ps(2)t = –κ + βbt−1;Constrained regimes are: (R0): τ(0C)t = max{τ(1)t, τ(0)t}, ps(0C)t = max{ps(1)t, ps(0)t}; (R2): τ(2C)t = max{τ(1)t, τ(2)t}, ps(2C)t = max{br(1)t, br(2)t};πC = probability of constraint in any period, τ = tax rate, γ = ratio of government expenditures to permanent output, r = interest rate, λ = growth of permanent output, λ*t = expected output growth, b=ratio of debt to output, wt = ratio, permanent/total output in period t;For all simulations, λ = 4%, r = 7%, b(intial)=bP=0.5, γ= 0.2, variance of temporary output = 0.5* permanent output. Number of draws = 500.

As mentioned above, regimes may be compared by tradeoffs. Such tradeoffs are summarized in Table 5. Moving from (R2) to (R1), by how much will taxes rise (Δ Average)? By how much will the standard deviation fall (Δ Standard Deviation)? What is the tax increase required to ‘buy’ a one-percent reduction in the standard deviation? (Tradeoff Ratio = Δ Average/Δ Standard Deviation)?

Table 5.

Summary, Simulation Results, Tradeoffs

article image
Note: Δ Average = Difference in average tax rate (τ), (R2) minus (Rl); Δ Standard Deviation = Difference in standard deviation of tax rate (τ), (R2) minus (Rl); Tradeoff ratio = Δ Average/Δ Standard Deviation. Negative number indicates tax hike required to obtain a one-percent decrease in standard deviation.

In all cases, the initial debt ratio is assumed to be 50% (bP = .5); permanent growth λ is assumed to be 4 percent; the variance of temporary income is assumed to be 5% of permanent income; the constant interest rate is 7% (r=.07), and the permanent spending ratio is γ = 20 percent. Assumption that borrowing restrictions will be imposed with probabilities πc = 0, 0.3, and 0.5 are presented in Tables 2, 3, and 4, respectively. For regime (R2), all tables present alternative values for κ and β, namely κ = .03, β=.8 and κ = .05, β=.8.

Consider first the case of no borrowing constraints πc=0 in Table 2. Note that for (R0) the (benchmark) constant tax rate is τ = .2144 (21.44%) over all horizons. Moving to the near-balanced budget regime (R1) reduces tax rates (but only slightly) while variability becomes positive.29 For example, during the first five years (J = 5) tax rate variance rises from zero under (R0) to 0.05 percent of GDP under (R1).

Thus, in this period, tax rates range from a minimum of 18.9 percent to a maximum of about 23.5 percent under (R1). For longer horizons (J = 10, 20) the variance of τ rises, as does the gap between the between minima and maxima. Note also that, under both regimes, the primary surpluses range from 1.1 to 1.3 percent of GDP. And, end-period debt bJ remains on average close to its initial value of 0.5 under these regimes.

Unsurprisingly, for horizons presented here, taxes are lower under regime (R2)—in both cases—than either (R0) or (R1); tax rate variability is about the equal under (R1) and (R2) for J = 5; However, as J rises to 10 and 20, tax rate variability does as well. For example, for κ = 0.03—a regime under which primary deficits average about 2 percent of GDP—tax rates in the first five years (J = 5) range from just under 16% to just over 20%. However, over a ten-year horizon (J = 10), tax rates range from about 14% to 22%, and for a twenty-year horizon (T = 20) tax rates range from about 12% to about 25%. In all cases, tax rates become more variable κ is increased to 0.05 (primary deficits of 3 to 4 percent of GDP).

Unsurprisingly, debt accumulation is also substantially greater under (R2), and debt builds up ever more as the period grows. For example, in the case of J = 5, debt accumulation averages about 60% of GDP for κ = 0.03 and 65% for κ = 0.05. For J = 10, the end-period debt ratio bJ rises to about 72% and 85%; for J = 20, the debt ratio rises to just under 100% and 127%, respectively.

Thus, as Table 5 shows, for J=10 the tax rate increase required to obtain a one percent decrease in the standard deviation appears to be prohibitive: 35% for κ = 0.03 and 10% for κ = 0.05. However, as the horizon grows to J = 20, this tradeoff drops to dramatically: 3.1% for κ = 0.03 and 2.5% for κ = 0.05.

As Tables 3 and 4 show, if uncertain borrowing constraints are assumed (πc=0.3 and πc=0.5), regime (R0) becomes slightly less attractive relative to (R1): both the level and variability of tax rates under (R0) rise for the cases shown.30 Unsurprisingly, (R0) debt accumulation is greater under πC = 0 than πC > 0 (since borrowing is not always available).Tables 3 and 4 also show, that the (R2) regimes become less attractive relative to both (R0) and (R1). As πc rises from 0 to 0.3 and 0.5, so do both the level and variability of tax rates under (R2).

Furthermore tax rates are always more variable under (R2) than (R1) if πc >0. For example, for κ = 0.03—a regime under which primary deficits average between 0.5 and 1 percent of GDP—tax rates in the first five years (J = 5) range from about 16.5% to about 21%. However, over a ten-year horizon (J = 10), tax rates range from about 15% to just under 24%, and over a twenty-year horizon (J = 20) tax rates range from about 13% to about 27%. As before, tax rates become more variable when κ is increased to 0.05 (primary deficits of 1.6 to 2 percent of GDP). Note also that debt buildup under (R2) falls when πc rises, but is nonetheless substantially higher than under (R0) or (R1).

Table 5 reveals that, for πc = 0.3 and 0.5, the tax rate increases required to obtain a one percent decrease in the standard deviation are substantially lower than for the case of πc = 0. For example, in the case of πc=0.5, J = 10, these tradeoffs are: 2.4% for κ = 0.03 and 1.6% for κ = 0.05. For a longer horizon of J = 20, these tradeoffs are 3.4% and 1.1%, respectively.

VI. Extension: Variable Government Expenditures

In the preceding discussion, expenditures have been assumed to be constant (exogenously set) fraction of permanent GDP. While such an assumption is standard in the literature, it is made primarily for convenience rather than realism.31 Typically, expenditures also suffer cuts during adverse periods. Thus, consider a more general framework. Expenditures and taxes, without borrowing constraints, are determined by:32

τt=γP−κ+ωβbt−1(16a)
γt=τt+κ−(1−ω)βbt−1(16b)

where γP and γt are permanent and total government expenditures respectively, and 0 ≤ ω ≤ 1. In expressions (16a) and (16b) the exogenous (long-run), fiscal adjustment is distributed between taxes and expenditures according to ω: if ω = 1, the entirety of the adjustment falls on taxes (equivalent to the previous section’s model). By contrast, if ω = 0 all adjustment falls on expenditures. In this case, τ is constant and γP= τ + κ.33

For πc > 0, the corresponding expressions are:

τtC=max[γP−κ+ωβbt−1,γP−ω(r−λt*)/(1+λt*)bt−1](17b)
γtC=min [τt+κ−(1−ω)βbt−1,τt−(1−ω)(r−λt*)/(1+λt*)bt−1](17b)

In (17b), if ω = 0, if a borrowing government is denied access to credit, it cuts expenditures in that period.

VII. Extension: Public Sector Size and Volatility

Fiscal reforms generally envisage permanent cuts of less productive expenditures.34 Doing so helps transfer resources to either higher priority public expenditures, the private sector (through tax cuts), or both. This is perhaps the most widely recognized benefit of such an adjustment. However, doing so may also permit essential public goods and services to be provided more smoothly—with fewer cuts or interruptions.

Moreover, the previous discussion suggests that level of permanent government expenditures γP and their volatility should be related. For example, under the endogenous expenditure regime (ω=0) the average level of government expenditures (γp= τ+κ) and the variability of expenditures var(γ) should be positively related.35 For more general cases, (0 ≤ ω ≤ 1) higher γP should raise the variability of both expenditures and revenues.

To investigate this issue in Latin America, Figure 1 presents a plot of the level of real consumption expenditures (relative to GDP) against its coefficient of variation (variance/mean). (The average ratio of government consumption/GDP thus proxies for the permanent expenditure ratio γ). According to this chart, casual observation may favor such a positive relationship among Latin American countries.

Figure 1.
Figure 1.

Latin America: Public Consumption/GDP Level and Coefficient of Variation

Citation: IMF Working Papers 2003, 220; 10.5089/9781451875225.001.A001

Source: IMF - WEOAverage Level, 1980-2001 (in percent)

What is the relationship between the level of government expenditures and real GDP volatility?36 In the traditional public finance literature, stabilization was a key role of the public sector. Moreover, according to recent evidence presented by Fatas and Mihov (2001), amongst industrialized countries, a larger public sector is associated with lower output variability. However, such a relationship is not evident for Latin America. Figure 2 presents a plot of the level of real consumption expenditures (relative to GDP) against the variance of real GDP. According to this chart, casual observation may also favor such a positive relationship among Latin American countries between these two variables.

Figure 2.
Figure 2.

Latin America: Public Consumption/GDP and Output Volatility

Citation: IMF Working Papers 2003, 220; 10.5089/9781451875225.001.A001

Source: IMF - WEOAverage Level, 1980-2001 (in percent)

VIII. Summary and Conclusions

This paper attempted to clarify several widely-held but informal notions regarding restrictive fiscal rules and the conduct of fiscal policy over the business cycle. Fiscal rules (like Gramm-Rudman-Hollings) are often cast as an “enemy” of the first-best (Ramsey) optimum of tax smoothing. Of course, in any welfare comparison, it is essential to be clear about exactly what are the alternatives under consideration. Fiscal policy in many emerging markets—and particularly in Latin America—is plagued by budgetary rigidities, weak tax administration, and volatile tax rates, expenditures, and debt/GDP ratios.

As a theoretical construct, the benefits of a Ramsey-style tax smoothing regime are clear: over an infinite horizon, consumers benefit from lower and less variable tax rates. A more difficult question involves shorter horizons. If a persistent tax gap is eliminated (through once-and-for-all measure tax and expenditure measures) will tax rates or expenditures become appreciably smoother? Simulations presented in this paper suggested that the answer to this question is “yes.” Moreover, while such once-and-for-all measures may be distinct from a balanced-budget law or other fiscal restriction, the two may nonetheless complement one another. In these ways, Gramm-Rudman-Hollings and Frank Ramsey may be “friends” rather than “enemies.”

While the assumptions in this paper were simple, more realistic ones might be used in future work. For example, both expenditures and taxes might share some of the burden in further simulations. Also, future work might specify consumer preferences and the production technology more fully.

There were also some key topics that, while omitted, would be fruitful extensions in future work. For example, an extension of this work might include a motivation for default and endogenous borrowing constraints, as discussed in previous sections. Also, the model might be extended to include changes in the price level, interest rate changes, or both (according to, for example, the recently developed “fiscal theory of the price level”). Ultimately, economic theory ought to be able to compare a GRH-like rule against policies that countries currently pursue. The agenda for future research on this topic thus remains sizeable.

References

  • Aiyagari, S.Rao, Albert Marcet, Thomas J. Sargent, and Juha Seppälä, 2002, “Optimal Taxation without State-Contingent Debt,” Journal of Political Economy, Vol. 110, No. 6 (December), pp. 1220–54.

    • Search Google Scholar
    • Export Citation

  • Aizenman, Joshua, Michael Gavin, and Ricardo Hausmann, 2000, “Optimal Tax And Debt Policy with Endogenously Imperfect Creditworthiness,” Journal of International Trade and Economic Development, Vol. 9, No. 4 (December), pp. 367–95.

    • Search Google Scholar
    • Export Citation

  • Alesina, Alberto, and Silvia Ardagna, 1998, “Fiscal Adjustments: Why They Can Be Expansionary,” Economic Policy: A European Forum, Vol. 0, No. 27 (October), pp: 487–517.

    • Search Google Scholar
    • Export Citation

  • Barro, Robert J., 1979, “On the Determination of the Public Debt,” Journal of Political Economy, Vol. 87, No. 5 (Part 1, October 1979), pp. 940–71.

    • Search Google Scholar
    • Export Citation

  • Bayoumi, Tamim, and Barry Eichengreen, 1995, “Restraining Yourself: The Implications of Fiscal Rules for Economic Stabilization,” International Monetary Fund Staff Papers, Vol. 42, No. 1 (March), pp. 32–48.

    • Search Google Scholar
    • Export Citation

  • Blanchard, O. J., et al., 1990, “The Sustainability of Fiscal Policy: New Answers to an Old Question,” OECD Economic Studies, 15 (Autumn), pp. 7–34.

    • Search Google Scholar
    • Export Citation

  • Chalk, Nigel, and Richard Hemming, 2000, “Assessing Fiscal Sustainability in Theory and Practice,” International Monetary Fund Working Paper No. 00/81.

    • Search Google Scholar
    • Export Citation

  • Chalk, Nigel A., 2002, “Structural Balances and All That: Which Indicators to Use in Assessing Fiscal Policy,” International Monetary Fund Working Paper No. 02/101.

    • Search Google Scholar
    • Export Citation

  • Chari, V.V., and Patrick J. Kehoe, 1999, “Optimal Fiscal and Monetary Policy,” NBER Working Paper No. W6891 (January).

  • Croce, Enzo, and V. Hugo Juan-Ramón, 2003, “Assessing Fiscal Sustainability: A Cross-Country Analysis,” International Monetary Fund Working Paper 03/145 (June).

    • Search Google Scholar
    • Export Citation

  • Drazen, Allan, 2000, Political Economy in Macroeconomics, Princeton, NY: Princeton University Press.

  • Drazen, Allan, 2002, “Fiscal Rules From a Political Economy Perspective,” presented at Conference on Rules-Based Fiscal Policy in Emerging Market Economies, Oaxaca, Mexico, February 14–16, 2002.

    • Search Google Scholar
    • Export Citation

  • Fatas, Antonio, and Ilian Mihov, 2001, “Government Size and Automatic Stabilizers: International and Intranational Evidence,” Journal of International Economics, Vol. 55, No. 1 (October), pp. 3–28.

    • Search Google Scholar
    • Export Citation

  • Gavin, Michael, and Roberto Perotti, 1997, “Fiscal Policy in Latin America,” in Bernanke, Ben S. and Rotemberg (July), NBER Macroeconomics Annual 199, pp. 11–61.

    • Search Google Scholar
    • Export Citation

  • Guardia, Eduardo, R. and Roberto P. Messenberg, 2002, “Fiscal Rules and Fiscal Adjustment in Brazil,” presented at Conference on Rules-Based Fiscal Policy in Emerging Market Economies, Oaxaca, Mexico, February 14–16, 2002.

    • Search Google Scholar
    • Export Citation

  • Kopits, G., and S. Symansky, 1998, Fiscal Policy Rules, IMF Occasional Paper 162 (Washington, D.C.: International Monetary Fund).

  • Leeper, Eric M., 1991, Equilibria under “Active” and “Passive” Monetary and Fiscal Policies,” Journal of Monetary Economics, Vol. 27, No. 1 (February), pp. 129–47.

    • Search Google Scholar
    • Export Citation

  • Lucas, Robert E. Jr., and Nancy Stokey, 1983, “Optimal Fiscal and Monetary Policy in an Economy Without Capital,” Journal of Monetary Economics, Vol. 12 (July), pp. 55–93.

    • Search Google Scholar
    • Export Citation

  • McCallum, Bennet T., 1984, “Are Bond-Financed Deficits Inflationary? A Ricardian Analysis,” Journal of Political Economy, Vol. 92 (February), pp. 123–35.

    • Search Google Scholar
    • Export Citation

  • Milesi-Ferretti, Gian Maria, 2000, “Good, Bad, or Ugly? On the Effects of Fiscal Rules with Creative Accounting,” IMF Working Paper 00/172 (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation

  • Ramsey, Frank, 1927, “A Contribution to the Theory of Taxation,” Economic Journal, Vol. 37, pp 47–61.

  • Schmitt-Grohe, Stephanie, and Martin Uribe, 1997, “Balanced-Budget Rules, Distortionary Taxes, and Aggregate Instability,” Journal of Political Economy, Vol. 105, No. 5 (October), pp. 976–1000.

    • Search Google Scholar
    • Export Citation

  • Stockman, David R., 2001a, “Balanced-Budget Rules: Welfare Loss and Optimal Policies,” Review of Economic Dynamics, Vol. 4, No. 2, pp. 438–59.

    • Search Google Scholar
    • Export Citation

  • Stockman, David R., 2001b, “Default, Reputation, and Balanced-Budget Rules,” unpublished manuscript, University of Delaware, Department of Economics, http://www.buec.udel.edu/stockmad/.

    • Search Google Scholar
    • Export Citation

  • Talvi, Ernesto, and Carlos Vegh (eds.), 2000a, “Cómo Armar el Rompecabezas Fiscal: Nuevos Indicadores de Sostenibilidad,” (Washington: Inter-American Development Bank).

    • Search Google Scholar
    • Export Citation

  • Talvi, Ernesto, and Carlos A. Végh, 2000b, “Tax Base Variability and Procyclical Fiscal Policy,” NBER Working Paper No. 7499, (Cambridge, Massashusetss: National Bureau of Economic Research).

    • Search Google Scholar
    • Export Citation
1

This paper benefited from insightful comments by Adolfo Barajas, Kevin Carey, Enzo Croce, Mercedes Da Costa, Andrew Feltenstein, Hugo Juan-Ramón, Michael Kumhof, Gene Leon, Eduardo Ley, Michel Robe, and André Santos. All errors are my own responsibility.

2

Fiscal rules may also encourage governments to use questionable accounting procedures (Milesi-Ferretti (2000)).

3

Drazen (2000, 2002) argues that fiscal rules compensate for inherent pro-deficit biases. Aizenman, Gavin, and Hausmann (2000) and Stockman (2001a,b) formally model the credibility gains from fiscal rules.

4

For example, among Latin American countries, fiscal rules have been recently enacted in Argentina, Brazil, Colombia, Peru and Chile, and they have been proposed in elsewhere.

5

Empirical evidence on fiscal rules (largely for industrialized countries) supports both claims. For example, Bayoumi and Eichengreen (1995) suggest that jurisdictions (U.S. states) that have more restrictive fiscal rules do also run smaller deficits. At the same time, fiscal policy in such jurisdictions tends to be more procyclical in nature than jurisdictions without fiscal rules.

6

In work after Barro (1979)—for example Lucas and Stokey (1983), Chari and Kehoe (1999), and Aiyagari, Marcet, Sargent and Seppälä (2002)—a “Ramsey approach” came to be understood as one in which the authority directly maximizes a the utility of a representative consumer over an infinite horizon. However, the position taken in this paper is that a “Ramsey approach” may also be one where the authority indirectly maximizes utlity, including by minimizing a deadweight loss function (as did Barro (1979)). Note also that the “countercyclical policy” in this paper refers specifically to the smoothing of taxes or expenditures. Issues of endogenous countercyclical spending (automatic stabilizers) or discretionary policy are left for another paper.

7

Reference to Gramm-Rudman-Hollings (GRH), a legislative act that has been modified and weakened, is for rhetorical purposes only. Fiscal rules outside the United States may differ considerably either from GRH or the fiscal rule proposed in this paper.

9

Ex-ante, the rule permits just enough borrowing to keep the debt ratio constant. However, since tax rates are set before output is known, ex post borrowing also reflects a forecasting error. Such net borrowing might be thought of as variations in public sector bank deposits.

10

In the model, the deficit is eliminated by raising taxes. Potential examples of such a policy might include a one-time tax rate increase or improvement in tax collection. More broadly, one-time, permanent expenditure reductions may also help reduce the deficit. In this sense, there may be a distinction between a fiscal rule and a fiscal reform. A fiscal rule, according to Kopits and Symansky (1998), is a permanent restriction on fiscal policy, while a reform occurs at one point in time. Of course, the two measures may complement one another. And, as a legal matter, the two may be combined. For example, Brazil’s Fiscal Responsibility Law (FRL) not only limits borrowing but also mandated a one-time, permanent reduction in public sector employment. For further details, see Guardia and Messenberg (2002).

11

Whether or not the costs of increased tax rate variability exceeds the benefits of lower tax rate levels depends on the precise form of the loss function (the degree of risk aversion).

12

Note that, in Latin America fiscal policy in the region has generally been procyclical even without fiscal rules (see, for example, Gavin and Perotti (1997), Talvi and Végh (2000b)).

13

Other authors have used such an expression; see, for example, Leeper (1991). In this section, since growth is constant, actual and permanent output are identical. Note also that the assumption of constant γ is made for simplicity. An extension to the case of variable expenditure ratios is presented later in the paper.

14

Previous literature in this area typically specifies household preferences. For a closed economy doing so is important, since intertemporal fluctuations in the tax rate may affect intertemporal allocations of consumption and leisure. By contrast, for an open economy such a specification would be inessential, since the government smooths its expenditure stream over time through international capital markets (the current account). In this context, the loss function should be interpreted as one of collection costs on a commodity tax – for example tax evasion. While consumers might wish to smooth such costs over time, incomplete markets prohibit them from doing so.

15

Default, through inflation or otherwise, is not considered in this paper.

16

This idea is widely used in policy literature regarding fiscal sustainability. See, for example, Chalk and Hemming (2000), Chalk (2002) and Croce and Juan-Ramón (2002).

17

However, as discussed above, a sustainable policy κ > 0 and β < [r−λ]/[1+λ] has an important drawback: debt and tax rates increase over time. In some cases, the discounted debt may also grow, but only initially. Note also that formally debt per se is not “bad” if the borrower sticks to its chosen values of κ and β and lenders continue to lend—no matter what. Such a case is formally equivalent to one discussed by McCallum (1984).

18

One kind of default is unanticipated inflation. For simplicity, we do not consider such issues in this paper.

19

Introducing a random exclusion from borrowing also indirectly brings in the issue of default. If borrowers know with certainty that they will be cut off from all borrowing in the present and future, and there is no other default penalty, the borrower will default. By contrast, if the government knows with certainty that they will never be cut off from credit markets, default is less likely than otherwise. In the scheme presented here, governments find themselves somewhere between these two extremes. More realistically, the probability of a borrowing cut-off should be modeled as a function of debt itself. Such a task is left for another paper.

20

In expression (9), the current period marginal loss from taxation is ϕ′ (τ(U)1–κ1) = ϕ′[γ+(r−λ)/(1+λ)b0–κ1], the future marginal loss, unconstrained case is ϕ′(τ(U))2 = (1–πC)ϕ′[γ+(r−λ)/(1+λ)b0], and the future marginal loss, constrained case is ϕ′(τ(C))2 = ϕ′[γ+(r−λ2)/(1+λ2)(b1+κ1)], where b0=bP.

21

The optimal zero-tax policy derived by Aiyagari, Marcet, Sargent and Seppälä (2002), discussed above, is such an example.

22
As a probability limit, E(1–wt) converges to zero. Thus, solvency is ensured, since discounted debt converges over an infinite horizon to:
lim t→∞bt/θt=γ∑t=0∞[wt−1]/θt
23

We might assume that such ex post variations in the debt ratio are reflected in changes in government bank deposits or other liquid assets.

24

Schmitt-Grohe and Uribe (1997) make a similar comparison in the context of a growth model.

25

As mentioned above, such borrowing might reflect reductions in government bank deposits.

26

More precisely, borrowing and taxes under borrowing constraints are: br(0C)t = max{br(1)t, br(0)t}, τ(0C)t = min{τ(1)t, τ(0) t}, br(2C)t = max{br(1)t, br(2)t}, and τ(2C)t = min{τ (1)t, τ (2)t}.

27

Taxes under (R0) can vary but only if there are borrowing constraints.

28

This policy may be thought of as a naïve application of problem (5′) that ignores tax rate variability.

29

Note that results for (R1) are invariant to πc.

30

Note that, at some point, an increase in πc should decrease both tax levels and variability under (R0) and (R2), since πc =1 is the same as an (R1) regime.

31

It is easier to make welfare statements about tax rates than about expenditures. For example, government expenditures in the form of lump-sum transfers have no welfare implications.

32

Note that, under (R0), both taxes and expenditures are exogenous.

33

Presumably, political considerations would determine the value of ω. However, this topic is left for another paper.

34

That is, alignment between taxes and primary expenditures—removal of the “tax gap” κ — may be achieved by a once-and-for-all reduction in γ. Moreover, as Alesina and Ardanga (1998) suggest, fiscal adjustments that emphasize expenditure reduction rather than tax increases are both more durable and more likely to increase economic growth.

35

For discussions of related issues in Latin America see Gavin and Perotti (1996) and Talvi and Végh (2000b).

36

This idea is not undisputed. For example, in the traditional public finance literature stabilization was one of the public sector’s key roles. More recently, Fatas and Mihov (2001) provide evidence—for industrialized countries—that a larger public sector is associated with lower output variability.

Same Series

Other IMF Content

Other Publishers

Cover IMF Working Papers
Fiscal Rules and Countercyclical Policy: Frank Ramsey Meets Gramm-Rudman-Hollings
Author: Mr. Evan C Tanner
Volume/Issue:
Volume 2003: Issue 220
Publisher:
International Monetary Fund
ISBN:
9781451875225
ISSN:
1018-5941
Pages:
29
DOI:
https://doi.org/10.5089/9781451875225.001
  • View in gallery

    Latin America: Public Consumption/GDP Level and Coefficient of Variation

  • View in gallery

    Latin America: Public Consumption/GDP and Output Volatility

Latin America: Public Consumption/GDP Level and Coefficient of Variation

Latin America: Public Consumption/GDP and Output Volatility