Effects of Fiscal Stimulus in Structural Models

Michael Kumhof; Dirk V Muir; Carlos de Resende; Jan in ‘t Veld; René Lalonde; Davide Furceri; Annabelle Mourougane; John Roberts; Stephen Snudden; Mathias Trabandt; Günter Coenen; Susanna Mursula; Christopher J. Erceg; Charles Freedman; Jesper Lindé; Werner Roeger; Douglas Laxton

doi:10.5089/9781451982169.001.A001

Author: Mr. Michael Kumhof, Mr. Dirk V Muir, Carlos de Resende¹, Jan in ‘t Veld², René Lalonde³, Davide Furceri⁴, Annabelle Mourougane⁵, John Roberts⁶, Stephen Snudden⁷, Mathias Trabandt⁸, Günter Coenen⁹, Susanna Mursula, Christopher J. Erceg¹⁰, Charles Freedman¹¹, Jesper Lindé¹², Werner Roeger¹³, and Mr. Douglas Laxton

¹ 0000000404811396https://isni.org/isni/0000000404811396International Monetary Fund
| ² 0000000404811396https://isni.org/isni/0000000404811396International Monetary Fund
| ³ 0000000404811396https://isni.org/isni/0000000404811396International Monetary Fund
| ⁴ 0000000404811396https://isni.org/isni/0000000404811396International Monetary Fund
| ⁵ 0000000404811396https://isni.org/isni/0000000404811396International Monetary Fund
| ⁶ 0000000404811396https://isni.org/isni/0000000404811396International Monetary Fund
| ⁷ 0000000404811396https://isni.org/isni/0000000404811396International Monetary Fund
| ⁸ 0000000404811396https://isni.org/isni/0000000404811396International Monetary Fund
| ⁹ 0000000404811396https://isni.org/isni/0000000404811396International Monetary Fund
| ¹⁰ 0000000404811396https://isni.org/isni/0000000404811396International Monetary Fund
| ¹¹ 0000000404811396https://isni.org/isni/0000000404811396International Monetary Fund
| ¹² 0000000404811396https://isni.org/isni/0000000404811396International Monetary Fund
| ¹³ 0000000404811396https://isni.org/isni/0000000404811396International Monetary Fund

Contributor Notes

Author’s E-Mail Address: mkumhof@imf.org, dlaxton@imf.org, dmuir@imf.org, smursula@imf.org

Publication Date:: 01 Mar 2010

eISBN:: 9781451982169

Language:: English

Keywords:: WP; consumption; inflation; GDP

Download PDF (683 KB)

The paper assesses, using seven structural models used heavily by policymaking institutions, the effectiveness of temporary fiscal stimulus. Models can, more easily than empirical studies, account for differences between fiscal instruments, for differences between structural characteristics of the economy, and for monetary-fiscal policy interactions. Findings are: (i) There is substantial agreement across models on the sizes of fiscal multipliers. (ii) The sizes of spending and targeted transfers multipliers are large. (iii) Fiscal policy is most effective if it has some persistence and if monetary policy accommodates it. (iv) The perception of permanent fiscal stimulus leads to significantly lower initial multipliers.

Abstract

I. Introduction

The global economy has over recent years suffered from a number of large negative demand shocks, which were initially driven by sharp declines in house and stock prices and a tightening of financial conditions (see Decressin and Laxton, 2009). The resulting collapse in output and the increase in unemployment also gave rise to a loss of confidence that intensified the downward pressures on the economy.

Governments and central banks responded by introducing measures to deal with liquidity and solvency problems in financial institutions. Central banks reduced interest rates to unprecedented levels to support aggregate demand in the face of an increase in private sector risk premia. They also used nonconventional measures in the form of quantitative easing and qualitative or credit easing to reduce risk premia and to provide liquidity. Despite all of this, credit remained tight and aggregate demand in many countries continued to weaken rapidly. There were negative spillovers from the weakening economies to those that had appeared to be more robust, and increased concern that the global economy might be moving into a period of deep and prolonged recession (IMF, 2009a).

With limited scope for further stimulus through monetary policy, attention turned to fiscal policy.¹ In this context, questions were raised as to how effective temporary government fiscal policy actions would be in lessening the depth and duration of the slowdown, and what the preferred mix of fiscal policy actions would be. Questions also started to be asked about the long-run sustainability of deteriorating fiscal positions, and about the potential long-run crowding-out effects of the debt accumulation resulting from the fiscal stimulus.

This paper addresses these questions, with the focus very much on the issue of short-run effectiveness, in other words on the size of fiscal multipliers, but with some discussion towards the end concerning long-run issues. The tools we use to address these questions are seven structural models of national economies and of the global economy that have been extensively used and tested over the years in a number of policymaking institutions, including the U.S. Federal Reserve, the European Central Bank, the International Monetary Fund, the European Commission, the OECD and the Bank of Canada.

Comparing the output of these diverse models provides a useful check on the robustness of the predictions produced by state-of-the-art macroeconomic models. We will demonstrate that there is indeed a considerable degree of agreement across theoretical models, particularly if one compares the outcomes with the wide variation in results in reduced-form empirical work. Furthermore, whenever there are substantial differences, the sources of these differences are fairly straightforward to identify. As we will discuss in Section II, this is quite different from the empirical literature, where there has been considerable disagreement over the effectiveness of different types of temporary fiscal measures in stimulating aggregate demand. We therefore feel that this work adds valuable information that can support the urgent decisions facing policymakers today.

There are four broad conclusions flowing from our analysis. First, there is no such thing as a simple fiscal multiplier. The response of the economy to temporary discretionary fiscal stimulus depends on a number of factors, including most importantly the type of fiscal instrument used and the extent of monetary accommodation of the higher inflation generated by the stimulus. Second, temporary expansionary fiscal actions can be highly effective, particularly when the fiscal instrument is spending or well-targeted transfers, and when in addition monetary policy is accommodative. Third, permanent stimulus, that is a permanent increase in deficits, is much more problematic than temporary stimulus. It leads to a long-run contraction in output, but in addition the perception that deficits will become permanent also substantially reduces short-run fiscal multipliers. Fourth, the G20 stimulus should have significant effects on global GDP in 2009 and 2010.

The rest of the paper is organized as follows. Section II discusses the relative merits of using empirical evidence versus theoretical models to improve our understanding of the effects of fiscal policies, and it provides a brief survey of the two relevant literatures. Section III introduces the seven structural models, and the seven standardized specifications of temporary fiscal shocks. Section IV provides extensive multiplier estimates using simulations of the models. Section V discusses the effects of permanent changes of fiscal instruments and of government debt, using simulations of two of the models. Section VI uses simulations of three of the models to quantify the worldwide effects of the G20 stimulus packages that have been announced for 2009 and 2010. Section VII provides concluding comments.

II. Fiscal Multipliers: Empirical and Model-Based Evidence

Our knowledge of the effects of fiscal policy comes from two sources, reduced-form empirical exercises and structural models.

Reduced-form empirical work has produced empirical estimates of fiscal multipliers that are dispersed over a very broad range.² In studies that pay close attention to the identification of fiscal stimulus in the United States (Blanchard and Perotti 2002, and Romer and Romer 2008), a fiscal stimulus of one percent of GDP has been found to increase GDP by close to one percent on impact and by as much as 2 to 3 percent when the effect peaks a few years later. On the other hand, Perotti (2005) finds much smaller multipliers for European countries using the same identification strategy as in Blanchard and Perotti (2002). Cross-country studies often find small fiscal multipliers and in some cases multipliers with a negative sign (Christiansen 2008). The most notable studies with negative multipliers are found in the literature on expansionary fiscal contractions initiated by Giavazzi and Pagano (1990) and surveyed in Hemming, Kell and Mahfouz (2002). In their most recent work, Mountford and Uhlig (2009) again found substantial multipliers for the United States, with results that are comparable to Blanchard and Perotti (2002). On balance, the evidence provides some support for the view that, in the current environment where monetary policy remains accommodative, a well-executed global fiscal stimulus could provide an appreciable boost to aggregate demand in the world economy.

Empirical studies provide valuable information for policymakers, but they suffer from three major problems. First, the amount of available identifying information is often very small, making estimation results subject to considerable uncertainty. Second, there are many possibilities for omitted variable bias and reverse causation (most notably the two-way linkages between economic activity and fiscal balances), which reduce confidence in the results. Third, the amount of identifying information is far too small to allow us to say very much about issues like the interaction between monetary and fiscal policies, the distinction between different types of fiscal instruments, the distinction between automatic stabilizers and discretionary stimulus, leakages into imports, the effects of government financing constraints due to insufficient “fiscal space”, and other forces that cause variations in multipliers. Cross-country econometric studies of fiscal multipliers have not been able to adequately control for these possibilities, and this may be a major reason why they have tended to find lower multipliers than in the United States, and also lower multipliers than in the scenarios with accommodative monetary policy presented in this paper.

Given these difficulties with the empirical evidence, structural models could be a potentially valuable additional source of information. Structural models, particularly models that have been used heavily in policymaking institutions and that have therefore been applied to a variety of policy questions over the years, are identified using more than variation in fiscal policy. They can therefore bring a lot more evidence to bear in learning about the structure of the economy, and can then use that knowledge to deduce the likely effects of fiscal policy. This knowledge is reflected in the choice of the model structure itself, which would typically have been adapted to generate empirically valid correlations between key macroeconomic variables, and also in the calibration, which is always based on a great variety of sources of empirical evidence. Of course structural models also have weaknesses, most importantly the fact that there is only incomplete consensus on the most appropriate structural features and calibration, and these could have a material effect on the results. This however is precisely where this paper makes one of its most valuable contributions, by finding that there is substantial agreement across models on both the absolute and relative sizes of different types of fiscal multipliers. The other key contribution is that our analysis clarifies several key elements that should be important in enhancing the effectiveness of the stimulative fiscal actions.

Given the importance of this topic, several recent papers have used theoretical models to analyze the effects of fiscal stimulus. Cogan and others (2009) claim that the conclusions regarding fiscal stimulus of policy models currently used in practice, including FRB-US, are not robust, and that standard New Keynesian models such as Smets and Wouters (2007) produce much smaller multipliers. Our results do not support this view, as FRB-US is one of the seven models used in our comparison, and as its results are found to be broadly consistent with those of the other models. One explanation is that, unlike Cogan and others (2009), our comparison models feature hand-to-mouth agents, which are absent in Smets and Wouters (2007). But more importantly, Cogan and others (2009) focus mostly on the longer run multipliers of permanent fiscal shocks, while our focus is on the impact multipliers of temporary fiscal shocks, where the differences between models are much smaller. We will turn to an explicit comparison between the multipliers of temporary and permanent fiscal deficit shocks in Section V, where we find that the much larger negative wealth effects of permanent shocks substantially reduce their initial period multipliers. Corsetti and others (2009) discuss the possibility that the anticipation of post-stimulus spending reversals can help to crowd-in rather than crowd-out private consumption. Again, our design of fiscal stimulus in Section IV is as an explicitly temporary measure, and Section V contains the comparison with permanent increases in deficits that is consistent with the point made by Corsetti and others (2009). Christiano and others (2009) stress, as we do in this paper, that the government spending multiplier becomes very much larger when higher spending is accompanied by monetary accommodation. We note that all of the foregoing theoretical contributions focus almost exclusively on government spending as the single tool of fiscal policy, while this study allows for a number of other instruments.

III. Multipliers, Instruments, and Models

A. Definition of Fiscal Multipliers

The term fiscal multiplier has been used in a variety of ways in the literature. Broadly speaking, it describes the effects of changes in fiscal instruments on real GDP. Typically, it is defined as the ratio of the change in GDP to the change in the size of the fiscal instrument or the change in the fiscal balance. In this paper, we compare the effects on real GDP of different fiscal instruments. We therefore normalize the fiscal impulses in the experiments so that the size of the discretionary shock in each case represents an increase in expenditures or a decline in revenues equal to 1 percent of baseline, pre-stimulus GDP, for either one year or two years. Government deficits respond endogenously to the fiscal actions because of automatic stabilizers, so that the post-stimulus change in the deficits is less than the discretionary fiscal stimulus.³ We measure the multiplier as the percentage deviation of real GDP from baseline GDP as a result of the fiscal shock.

Throughout the simulations, it is assumed that there is a coordinated global monetary policy response. Monetary policy in each country is determined by an interest rate rule, where interest rates are allowed to either adjust freely in line with the central bank’s reaction function or are held fixed for one or two years.

On those occasions when the economy is operating at or near full capacity and fiscal stimulus is inappropriate, one would expect monetary policy to set interest rates on the basis of the reaction function in order to prevent the fiscal stimulus from leading to an overheated economy and upward pressure on inflation. That is, central banks will raise interest rates to offset the expansionary and inflationary implications of the fiscal expansion in such circumstances. In contrast, when the economy is in a serious recession, with interest rates at their floor of zero or a small positive value and with a risk of deflation, monetary policy can be used to hold interest rates constant during the period of fiscal expansion so that the two macroeconomic policies act in a complementary way.

B. The Seven Fiscal Instruments

The simulations of the structural models examine changes in seven fiscal instruments. These are

an increase in government consumption expenditures
an increase in government investment expenditures
an increase in general lump sum transfers
an increase in lump sum transfers targeted to hand-to-mouth households⁴
a decrease in labor tax rates
a decrease in consumption tax rates
a decrease in corporate income tax rates

C. The Seven Structural Models

Six institutions participated in this project using seven structural models—the European Commission (QUEST), the International Monetary Fund (GIMF), the Board of Governors of the Federal Reserve System (with two models, FRB-US and SIGMA), the Bank of Canada (BoC-GEM), the European Central Bank (NAWM), and the OECD (OECD Fiscal). Of the seven models, four are global (BoC-GEM, GIMF, QUEST and SIGMA), NAWM is a two-region model (United States and the euro area), FRB-US is a U.S. only model, and OECD Fiscal is a euro-area-only model. Six of the models are dynamic stochastic general equilibrium (DSGE) models, while FRB-US is based on the polynomial adjustment cost (PAC) framework. This explains that model’s fairly smooth consumption response to stimulus shocks despite a high share of hand-to-mouth households. Two of the models, QUEST and GIMF, use their annual versions for the simulations presented here, while the remainder uses quarterly versions.

Table 1 summarizes the key model features of the seven models, including the number of regions covered, the proportion of hand-to-mouth households, the type of monetary policy rule, certain special features, and references to papers that more thoroughly outline the models and their properties.

Table 1:

Key Model Features

^{^*}Households: HTM = hand to mouth; CC = credit constrained; FIN = finite lives; INF = infinitely lived.

^{^**}BoC-GEM has a non-Ricardian link between net foreign assets and government debt.

^{^***}IFB = inflation-forecast-based rule.

Table 1:

Key Model Features

	# of regions	Hand-to- Mouth %	Monetary Policy	Special Features ^*	References
QUEST	3	20% HTM 20% CC	Taylor Rule	Households: HTM, CC, INF; productive gov. investment; housing.	Ratto, Roeger, and in’t Veld (2009)Roeger and in’t Veld (2009)
GIMF	5	25%	IFB ^***	Households: HTM, FIN; productive gov. investment; financial accelerator.	Kumhof and Laxton (2007)Kumhof and others (2010)Freedman and others (2010)
FRB-US	1	40%	Taylor Rule	Households: HTM, INF; very detailed breakdown; substantial work on empirics.	Brayton and Tinsley (1996)
SIGMA	2	50%	Taylor Rule	Households: HTM, INF.	Erceg, Guerrieri and Gust (2005)
BoC-GEM	6	15%	IFB	Households: HTM, INF ^**; productive gov. investment; separate sectors for oil and non-oil commodities.	Lalonde and Muir (2007)Lalonde, de Resende, and Snudden (2010)
NAWM	2	25%	Taylor Rule	Households: HTM, INF; HTM can smooth consumption through cash.	Coenen, McAdam and Straub (2008)
OECD Fiscal	1	25%	Taylor Rule	Households: HTM, INF; endogenous government debt risk premium.	Furceri and Mourougane (2010)

^{^*}Households: HTM = hand to mouth; CC = credit constrained; FIN = finite lives; INF = infinitely lived.

^{^**}BoC-GEM has a non-Ricardian link between net foreign assets and government debt.

^{^***}IFB = inflation-forecast-based rule.

Table 1:

Key Model Features

	# of regions	Hand-to- Mouth %	Monetary Policy	Special Features ^*	References
QUEST	3	20% HTM 20% CC	Taylor Rule	Households: HTM, CC, INF; productive gov. investment; housing.	Ratto, Roeger, and in’t Veld (2009)Roeger and in’t Veld (2009)
GIMF	5	25%	IFB ^***	Households: HTM, FIN; productive gov. investment; financial accelerator.	Kumhof and Laxton (2007)Kumhof and others (2010)Freedman and others (2010)
FRB-US	1	40%	Taylor Rule	Households: HTM, INF; very detailed breakdown; substantial work on empirics.	Brayton and Tinsley (1996)
SIGMA	2	50%	Taylor Rule	Households: HTM, INF.	Erceg, Guerrieri and Gust (2005)
BoC-GEM	6	15%	IFB	Households: HTM, INF ^**; productive gov. investment; separate sectors for oil and non-oil commodities.	Lalonde and Muir (2007)Lalonde, de Resende, and Snudden (2010)
NAWM	2	25%	Taylor Rule	Households: HTM, INF; HTM can smooth consumption through cash.	Coenen, McAdam and Straub (2008)
OECD Fiscal	1	25%	Taylor Rule	Households: HTM, INF; endogenous government debt risk premium.	Furceri and Mourougane (2010)

^{^*}Households: HTM = hand to mouth; CC = credit constrained; FIN = finite lives; INF = infinitely lived.

^{^**}BoC-GEM has a non-Ricardian link between net foreign assets and government debt.

^{^***}IFB = inflation-forecast-based rule.

IV. Fiscal Multipliers for Temporary Stimulus

A. Introduction to Simulation Results

Figures 1 to 42 show the effects on U.S. real GDP, inflation, real interest rates, consumption and investment of the various instruments of U.S. fiscal stimulus. The simulations examine both one year of fiscal stimulus (figures 1 to 21) and two years of fiscal stimulus (figures 22 to 42). For each duration there are three charts for each of the seven fiscal instruments. Each chart compares the cases of no monetary accommodation (top panel), one year of monetary accommodation (middle panel) and two years of monetary accommodation (bottom panel). Not all models are used in all of these experiments. For example, the OECD Fiscal model is not used in any of these shocks since it covers only the euro area, the experiments with consumption tax cuts can be done only in QUEST and GIMF, while those with corporate income tax cuts can be done with those two models as well as SIGMA and BoC-GEM.

Figures 43 to 84 show the same results for the euro area or the EU, depending on the model.⁵ Most of these simulations were done using four models—QUEST, GIMF, NAWM and OECD Fiscal. The experiments with corporate income tax cuts in Europe were done only with QUEST and GIMF.

Figures 85 through 87 illustrate the results of two years of fiscal stimulus in the form of government investment expenditures in Europe and the United States on the GDP, inflation, real interest rates, consumption and investment of the two economic areas, assuming two years of monetary accommodation. These results are for fiscal stimulus in each area but not in both areas at the same time. This allows us to more easily compare fiscal multipliers for domestic stimulus in these two economic areas.

The magnitude of the fiscal multipliers in these simulations is highly dependent on both the structural features and details of the calibration of the underlying models. The impossibility of accounting for such details in empirical work may be an important reason why reduced-form empirical estimates are dispersed over such a wide range. The factors that are most evident in the simulations and that will be examined in some detail later in this section are the following: (i) the extent of monetary accommodation; (ii) the persistence of the fiscal stimulus; (iii) the type of fiscal instrument used and the channels through which the instrument operates; (iv) the population share of hand-to-mouth households; (v) the effect of targeting transfers to hand-to-mouth households; (vi) the role of economic openness; (vii) the degree of nominal rigidity of prices and wages; and (viii) the size of automatic stabilizers.

B. Government Investment - Detailed Discussion

Before examining each of these factors, we will examine in some detail two sets of simulation results, those pertaining to government investment expenditures in the United States and Europe, in order to illustrate the conclusions that can be drawn from this analysis.

Figure 1 shows the effects on U.S. real GDP of one year of fiscal stimulus in the form of government investment expenditures, figure 2 shows the effect of the same action on U.S. inflation and real interest rates, and figure 3 sets out its effects on U.S. consumption and investment. Figures 22 to 24 do the same for a program of two years of government investment expenditures.

As noted, the top panel in each figure shows the results of temporary fiscal stimulative actions under no monetary accommodation, the middle panel under one year of monetary accommodation and the bottom panel under two years of monetary accommodation. If the fiscal stimulus were undertaken in circumstances in which output was at or very near to capacity, one would expect the central bank to respond to the stimulus in line with the typical Taylor-type reaction function and to raise its policy interest rate in a way that would lessen the impact on output and inflation of the inappropriately-timed fiscal stimulus. In contrast, in circumstances in which aggregate demand is very weak, the economy is well below capacity output and is expected to remain there for some time, and in which the policy interest rate is at or near the zero lower bound, a policy of monetary accommodation for one or preferably two years would be appropriate to the circumstances. By not acting to offset the stimulative effects of fiscal policy, the central bank allows the temporary fiscal actions to have a considerably larger effect on the economy than if it acted in line with its normal reaction function.

That said, the central bank must be sensitive to the risk that its inaction could result in inflation expectations ratcheting upward to a level above the long-term inflation objective and remaining there. It would therefore be very helpful in such circumstances for the central bank to have policy credibility so that inflation expectations remain anchored at a low level, for the monetary accommodation to be viewed as temporary, and for the stimulative fiscal actions also to be viewed as temporary and as appropriate in the context of a very weak economy.⁶ In the absence of such a favorable environment, monetary accommodation could have adverse effects over time and result in inflation pressures that lead to the inflation rate overshooting its target and eventually requiring appreciable monetary tightening. The clearly positive benefits of monetary accommodation throughout this paper must be understood in the context of a favorable monetary policy environment, and the results would have to be qualified in the absence of such an environment.

We begin by examining the case of one year of fiscal stimulus in figure 1. For the case of no monetary accommodation, the multipliers for the first year are similar in all six models, ranging between 0.9 and 1.2. As we move from no monetary accommodation to one year of monetary accommodation to two years of monetary accommodation, the multipliers become larger and the differences between the results of the various models become more noticeable. In the case of two years of fiscal stimulus (figure 22), the differences across the models are even more noticeable, particularly in the case of two years of monetary accommodation, where the spread is much wider between the largest multiplier (about 2.2 in GIMF) and the smallest multiplier (about 1.1 in FRB-US).

The reasons for these disparities relate in large part to differences in the effects of the fiscal stimulus on inflation and the real interest rate. Thus, returning to the case of one-year fiscal stimulus in figure 2, we see that the effect on inflation differs considerably across the models, with the largest effect coming from GIMF, especially with monetary accommodation. This is mostly due to the fact that GIMF has somewhat smaller nominal rigidities than the other models. As a result, the effect on the real interest rate with monetary accommodation is largest in GIMF, with a decline of about 0.5 percent in the case of two years of monetary accommodation, followed by BoC-GEM at 0.4 percent. Qualitatively similar, but quantitatively even more pronounced, differences in the effects on inflation and real interest rates can be found for the simulations of two years of fiscal stimulus in figure 23. Interestingly, the variation of real interest rate movements also tends to be significant across models in the case of no monetary accommodation, mainly reflecting differences in the reaction functions embedded in the various models.

Significant differences in the models can also be seen in the behavior of consumption and investment. Figures 3 and 24 show that in most cases GIMF shows the largest effects on consumption and FRB-US shows the largest effects on investment. The consumption effects in models with a significant share of hand-to-mouth households (or of credit-constrained and finitely-lived households) are larger. Hand-to-mouth agents in NAWM can smooth consumption much more effectively than in other models because they can use cash for that purpose.

Important recent papers such as Cogan and others (2009) have expressed skepticism about fiscal multipliers, based on the fact that for permanent government spending shocks households experience large and immediate negative wealth effects due to their anticipation of a much higher tax burden in the future. In that case multipliers are well below one and start to decline immediately. This contrasts sharply with our results because stimulus only lasts for one or two years, so that the wealth effects are minimal while the direct stimulus to aggregate demand is large. We will return to this point in Section V.

Again because of the temporary nature of the stimulus, the persistence of consumption (and investment) beyond the stimulus period is small. Given the small wealth effects, only hand-to-mouth (and to some extent finitely-lived) households respond significantly to the income generating effects of the stimulus, and stop doing so when it expires. We do observe some persistence in BoC-GEM, GIMF and QUEST, but this is driven by the supply side, as these three models feature productive government investment that has highly persistent output effects. These are, however, not very large given the short duration of the stimulus period.

We can also examine the output effects of an equal sized fiscal stimulus in the four models that have a separate region for Europe (figures 43 and 64). For the case of no monetary accommodation, the multipliers for the first year are, again, very similar across models, around 1, and in a slightly narrower range than in the case of the United States. Once again, more monetary accommodation leads to larger GDP responses, while the variation across models is less than in the case of the United States. Consider the case of two years of fiscal stimulus, with 2 years of monetary accommodation, where the largest multiplier is from OECD Fiscal (about 1.6), followed by QUEST (about 1.5), and the smallest is from NAWM (about 1.1). The multipliers in these cases are lower than their U.S. counterparts because the real interest rate exhibits much smaller movements as a result of smaller inflation responses (see figure 23 versus figure 65 for the case of two years of fiscal stimulus). The reason is an assumption of stronger nominal rigidities in the price and wage processes for Europe compared to the United States, resulting in greater inflation inertia. This is of course based on observed differences in inflation inertia in the data. The three models that produce simulations for both the United States and Europe (QUEST, NAWM and GIMF) do assume significant differences in nominal rigidities between these two regions.

European consumption and investment behavior can be seen in figures 45 and 66. Consumption is usually more important, with the exception of OECD Fiscal, where investment expands more strongly, because it features a stronger link between the real interest rate and the level of investment than in the other models.

C. The Role of Monetary Accommodation

For each figure showing the effect of fiscal stimulus on real GDP, moving from the top panel (no monetary accommodation) to the middle panel (one year of monetary accommodation) to the bottom panel (two years of monetary accommodation), there is a tendency for multipliers to increase. The reason is that, in addition to their direct effects on aggregate demand, fiscal stimulus measures also have indirect effects on aggregate demand through their impact on real interest rates. The fiscal action leads to an increase in inflationary pressure as aggregate demand increases, which in turn leads to a movement in real interest rates. With no monetary accommodation, the inflation pressures lead to an upward movement in real interest rates and thereby offset, in part, the effects of the fiscal stimulus on GDP. In contrast, with monetary accommodation and interest rates held constant, the increases in inflation give rise to decreases in real interest rates. As a result, accommodative monetary policy complements the fiscal policy stimulus and intensifies its effects on real GDP. The indirect effects differ more across models than the direct effects, because of the differences in the size of the linkage between aggregate demand and inflation. In the case of the United States, inflationary pressures and, consequently, real interest rate movements, are largest in GIMF and BoC-GEM, and smallest in FRB-US and NAWM. In the case of Europe, they are largest in GIMF (as with the United States), and then, generally, followed by QUEST, OECD Fiscal and NAWM. Overall, monetary accommodation is less effective in Europe, mostly because of stronger nominal rigidities than in the United States.

D. The Persistence of Fiscal Stimulus

The size of the multiplier will depend on the expected persistence of the fiscal stimulus measure. Our focus for most of the paper is on fiscal expansions that are, and are perceived to be, temporary, and that therefore do not result in long-run crowding out of private expenditures.⁷ In such cases, a two-year expansion will have significantly larger multiplier effects than a one-year expansion in the first year, but only under monetary accommodation. The reason is that a more persistent boost to demand creates higher inflation over a longer period, thereby causing a more powerful reduction of real interest rates.

Compare, for example, figures 22 and 23 with figures 1 and 2 for the United States and figures 64 and 65 with figures 43 and 44 for Europe. In each case, the expectation of two years of fiscal stimulus results in a large increase in the size of the multiplier even in the first year of the stimulus. But in the case of Europe the differences are smaller than in the United States, as the stronger nominal rigidities in Europe limit the effect of higher demand on inflation and real interest rates.

As we will see later, fiscal expansions that are expected to last indefinitely typically have much smaller multipliers. This is because they result in a much larger increase in the present discounted value of taxes, and therefore a much larger negative wealth effect that crowds out private spending from the outset. Stimulus is therefore at its most effective over an intermediate horizon and in combination with monetary policy. In that case its direct demand and income generating effects, and its interactions with monetary accommodation, are more powerful than its wealth effects.

E. Government Spending versus Taxes and Transfers

Different types of fiscal measures operate on aggregate demand through different channels. Thus, government investment expenditures and government consumption expenditures impact directly on aggregate demand while increases in transfers and reductions in taxes operate mainly through their effects on personal disposable incomes, as well as through their effects on incentives in the case of changes in distortionary taxes. It is widely accepted in the literature that fiscal measures that have direct effects on aggregate demand have larger multipliers than those whose initial impact operates through their effects on private-sector spending behavior. This is confirmed by our simulations. For the United States, compare figures 22 and 25, on the one hand, with figures 28, 31, 34, 37 and 40 on the other. For Europe, compare figures 64 and 67, on the one hand, with figures 70, 73, 76, 79 and 82 on the other. A number of results are consistent across all models.

First, the multipliers from government investment and consumption expenditures, which are roughly similar in size, are clearly larger than the multipliers from transfers, labor income taxes, consumption taxes and corporate taxes.

Second, multipliers are small for general transfers, labor income taxes and corporate taxes, and somewhat larger (but still small relative to government expenditures) for consumption taxes.

Third, only targeted transfers (figures 31 and 73), which we will discuss again shortly, come close to having multipliers similar to those of government expenditures.

F. Different Taxes and Transfers

The responses to tax and transfer-based stimulus measures depend critically on two factors, the behavior of hand-to-mouth households and the relative distortions caused by different fiscal instruments.

We start our discussion of the importance of hand-to-mouth households under the assumption that stimulus is based on general transfers, which cause no distortions. Expenditures of hand-to-mouth households respond strongly to transfer changes in all models, while other households respond to the temporary nature of the transfer change largely by adjusting their saving behavior. In all but two of the models these other households are infinitely-lived, and therefore see future transfer cuts as exactly offsetting the current transfer increase, with their consumption not affected at all. In GIMF a temporary increase in transfers has some effects because households with finite lives interpret part of it as an increase in lifetime wealth. BoC-GEM generates a similar non-Ricardian feature by positing a link between net foreign assets and government debt. But for realistic planning horizons of finitely-lived agents this effect, while significant, is much smaller than the effect on hand-to-mouth households. The share of the latter in the population, and the precise nature in which they are constrained, are therefore critical determinants of the response of the economy to transfer shocks, and also to tax shocks.

Turning to distortionary tax cuts as the instrument of stimulus, these have additional effects on supply and therefore on inflation and real interest rates, but these effects differ significantly across instruments. For example, the output effects of temporary cuts in labour income tax rates are not very large, and in fact are smaller than for general transfers under monetary accommodation. This is because they increase potential output, which reduces the inflationary effects of the stimulus and therefore the effectiveness of monetary accommodation. Similar comments apply to cuts in corporate income taxes. In contrast, the simulations suggest that the incentives for increased household expenditures from temporary cuts in consumption taxes are significantly larger.

G. General Transfers versus Targeted Transfers

Hand-to-mouth households have a much higher marginal propensity to consume out of current income than other households. This has two implications. First, models that have a high share of hand-to-mouth households have a higher multiplier for transfers (and also for taxes). Second, transfers that can be targeted to hand-to-mouth households provide a much more powerful stimulus than general transfers.

Temporary increases in general transfers are presented in figures 7 and 28 for the United States and figures 49 and 70 for Europe. The multipliers for this measure are small in absolute terms across all models, but are larger in those models that have a higher share of hand-to-mouth households. Thus, FRB-US (40 percent for US), SIGMA (50 percent for US) and QUEST (40 percent for US and EU – this includes both hand-to-mouth and credit-constrained households) tend to have larger multipliers than the other three models – BoC-GEM (15 percent for US, 25 percent for EU), GIMF (25 percent for US and EU), NAWM (25 percent for US and EU) and OECD Fiscal (25 percent for EU).

Temporary increases in targeted transfers are presented in figures 10 and 31 for the United States and figures 52 and 73 for Europe. Multipliers increase sharply relative to general transfers, and in fact targeted transfers multipliers are close to (but somewhat lower than) those of government expenditures. The increase in multipliers from targeting transfers tends to be greatest in those models that have the lowest percentage of hand-to-mouth households, especially BoC-GEM and GIMF. The reason is that shifting the total value of the increase in transfers from the general public to the targeted groups leads to a larger increase in the disposable incomes of the targeted groups when they are a smaller proportion of the population.

H. United States versus Europe

Temporary fiscal stimulative actions have greater effects on output, inflation, and real interest rates in the United States than in Europe. Figures 85, 86 and 87 show the effects of two years of stimulative government investment expenditures in Europe (top panel – four models) and the United States (bottom panel – six models) along with two years of monetary accommodation. The larger effects in the United States could be due to a number of factors, and this section will explore which of them is most important. First, Europe is more open than the United States, and therefore the leakage to imports is larger. Second, automatic stabilizers play a larger role in Europe than in the United States, and therefore the leakage from the discretionary fiscal stimulus into higher taxes and lower transfers is greater in Europe. Third, the degree of nominal rigidities is larger in Europe than in the United States, and therefore the effect of expansionary fiscal actions on the rate of inflation is lower in Europe than in the United States, leading to smaller downward movements in the real interest rate in Europe than in the United States under monetary accommodation.

We analyze the importance of these three factors by examining the effects of a two-year, one percent of baseline GDP, increase in government consumption expenditures with different calibrations of the IMF’s model, GIMF. For openness we examine this shock relative to a baseline for the euro area, while for the remaining two factors we examine it relative to a baseline for the United States.

Consider first, the degree of openness of the economy. The top panel of figure 88 shows the results for the original calibration of the euro area in GIMF, with exports from and imports into the euro area of about 17.4 and 17.5 percent of GDP. The bottom panel of figure 88 presents the results of the fiscal shock for a recalibrated level of exports and imports as a share of GDP of one half of the original calibration (and much smaller than the calibrated share for the U.S. economy). With the more open calibration of the euro area, real GDP is 1.1 and 1.0 percent higher than baseline in years 1 and 2, respectively, in the case of two years of monetary accommodation. In the case of the more closed economy, real GDP is somewhat larger, at 1.2 and 1.1 percent higher than baseline. In the less open economy, the increase in demand falls more heavily on the domestic sector and also leads to a larger inflation response and therefore a larger decline in real interest rates in the case of monetary accommodation. The conclusion that the more closed economy has a somewhat higher fiscal multiplier holds for all the temporary fiscal stimulus measures considered in this paper, and for the same type of experiment on the U.S. economy.

Consider next the levels of nominal rigidities and of automatic stabilizers. Figure 89 examines three U.S. calibrations: the original calibration (top panel); an alternative where U.S. nominal rigidities have the same parameter values as in the euro area, that is they are 50 percent higher than in the original calibration (middle panel); and a third calibration in which the weight on the output gap in the fiscal policy rule is the same as in the euro area, that is it equals 0.49 instead of 0.34, the original U.S. calibration (bottom panel). Comparing the results of the original calibration with that for higher nominal rigidities, we see that inflation responds less to aggregate demand, and the real interest rate therefore also moves less under monetary accommodation. Thus, the multiplier is significantly lower with higher nominal rigidities, especially with two years of monetary accommodation. The U.S. results with higher nominal rigidities are closely in line with those seen in the GIMF simulations for Europe.

Higher automatic stabilizers reduce the effect of the government discretionary fiscal action through a decrease in general transfers to households as GDP increases. That is, the overall increase in the fiscal deficit is smaller when automatic stabilizers are larger. However, while there is some decline in the fiscal multiplier with larger automatic stabilizers, the magnitude is not very large.

Overall, it appears that the smaller fiscal multipliers in Europe relative to the United States in GIMF are mostly a result of the higher nominal rigidities in Europe, with the relative openness of European economies and the larger automatic stabilizers playing smaller roles.

V. Permanently Higher Deficits and Debt

In this section we demonstrate that, while the case for temporary fiscal stimulus, as outlined in the previous section, is strong, the case for a permanent increase in fiscal deficits is much weaker in terms of its short-run effectiveness, and its long-run consequences could be negative. For illustrative purposes, we focus on the United States, although the results hold qualitatively for each of the different regions in the models discussed below.

A. Permanently Higher Deficits and Short-Run Multipliers

We start by using the GIMF and QUEST models to compare the short-run effects of temporary and permanent increases in deficits and debt, using the United States as our example. GIMF features finitely-lived households while QUEST features infinitely-lived households.⁸ Figure 90 illustrates the differences in multipliers between a one-year fiscal stimulus using government consumption expenditures, and a permanent change in government consumption expenditures of the same size, one percent of baseline GDP. We assume that the permanent increase in government consumption is accompanied by a permanent increase in the government’s interest-inclusive deficit to GDP ratio equal to 1 percent of GDP. Given our assumptions about nominal growth rates, the latter leads in the long run to a 20 percent increase in the debt to GDP ratio. Higher long-run debt implies that additional interest charges will eventually exceed the 1 percent increase in the deficit. We assume that labor income taxes are adjusted to service these interest charges as well as the increase in government spending in the long run. Figure 91 illustrates the model’s dynamic transition between the short run and the long run when there is a permanent increase in the debt to GDP ratio. Given that this is a long-run scenario, we assume for all experiments in this section that there is no monetary accommodation.

The temporary fiscal stimulus has a one-year multiplier of about 1.0 in GIMF and 0.8 in QUEST (top panel of figure 90). Following the withdrawal of the fiscal stimulus, GDP remains slightly below its baseline value for some period of time in order to put downward pressure on inflation and bring it back to baseline from the levels reached during the period of fiscal stimulus.

The effect on GDP of a permanent change in the fiscal instrument is shown in the bottom panel of figure 90. The fiscal multiplier for the first year is on the order of 0.7 in GIMF and 0.3 in QUEST, considerably less than the multiplier from the temporary fiscal stimulus. The numbers fall to 0.5 and less than 0.1, respectively, in the second year. In QUEST they return to zero soon after that, remaining there subsequently, while in GIMF they turn negative in the long run due to crowding-out. We note that the multipliers reported by Cogan and others (2009) are very similar to those in the bottom panel of figure 90. Part of their skepticism about the effectiveness of fiscal stimulus can therefore be attributed to the fact that they assume permanent rather than temporary increases in spending.

To understand the differences between short-run stimulus and permanent deficits, we return to the fact that our permanent deficits experiment involves much higher labor income taxes in the long run. As shown in Figure 91, this has three effects.

First, the large increase in the present discounted value of taxes leads to a negative wealth effect that immediately starts to crowd out private demand. This is the main reason behind the much smaller first-year multiplier for the permanent measure, and for the fact that the multiplier thereafter starts to fall back towards zero.

Second, if taxes are distortionary, this exacerbates the crowding-out effects. The more distortionary is the tax, the greater will be the effect on potential GDP. Thus, the use of capital taxes would have a larger effect than labor taxes because of the higher long-run supply elasticity of capital. This logic also carries over to government investment spending. If this was cut in the long run to stabilize debt, there would be significant negative consequences for the long-run level of potential output because of the contribution of government infrastructure investment to private sector productivity.⁹

Third, in GIMF, due to finitely-lived households, part of the increase in government debt is perceived as net worth, and therefore crowds out alternative investments, specifically physical capital and (net) foreign assets, as well as resulting in a permanent increase in the world real interest rate. This makes long-run multipliers negative, as well as further reducing short-run multipliers of the permanent fiscal measure.¹⁰

B. Permanently Higher Deficits and Long-Run Crowding-Out

In this subsection we use GIMF to simulate the long-run effects of a permanent 0.5 percent increase in interest-inclusive deficits that increases long-run government debt by 10 percent of GDP. GIMF is calibrated so that a one percentage point increase in the U.S. government-debt-to-GDP ratio leads to an approximately one basis point increase in the U.S. and world real interest rate. This is at the lower end of the range of estimates (1-6 basis points) reported by Laubach (2003), Engen and Hubbard (2004) and Gale and Orszag (2004). The remaining models in this paper, except for OECD Fiscal, do not feature a link between debt and real interest rates. These models still generate some long-run crowding-out effects, but only due to the effects of higher distortionary taxes in the long run.¹¹

The simulations assume that deficits are initially increased through a reduction in taxes or an increase in transfers, unlike the simulations in the previous subsection. As debt and interest charges rise, the primary deficit has to fall to keep the overall deficit increase at 0.5 percent of GDP, and this is assumed to be implemented through an offsetting increase in taxes or reduction in transfers.

Table 2 presents the effects on real GDP in the United States, the rest of the world, and globally. As investment in the additional government debt crowds out saving in physical capital and (net) foreign assets, the world real interest rate rises by 9 to 11 basis points, and this results in a decline in global long-run real GDP of between 0.3 percent and 0.7 percent. The effects on real GDP in the United States are significantly larger than in the rest of the world. In the case of lower long-run transfers (-0.30 in the United States and -0.17 in the rest of the world), this is due to a negative wealth effect on U.S. consumption demand, from two sources. First, the United States pays higher interest charges on a pre-existing stock of foreign liabilities. And second, due to crowding-out it adds further to that stock of liabilities. In the case of taxes, the differences between the United States and the rest of the world are larger because tax distortions increase in the United States but not elsewhere. The largest distortions arise in the case of corporate income taxes, due to their effect on capital accumulation, with a long-run GDP effect of -0.71 (-0.23 in the rest of the world). The GDP effect of higher labor income taxes is -0.48 (-0.21 in the rest of the world), and that of consumption taxes is -0.38 (-0.18 in the rest of the world). This corresponds closely to the rankings of alternative taxes by their distortionary effects in the public finance literature.

Table 2:

Effects of a Permanent 10 Percentage Point Increase in the U.S. Government Debt to GDP Ratio

Table 2:

Effects of a Permanent 10 Percentage Point Increase in the U.S. Government Debt to GDP Ratio

	US	RoW	Global
Financed by a Cut in General Transfers
Real GDP	-0.30	-0.17	-0.20
Real Exchange Rate	0.88	-0.88	…
Real Interest Rate	0.10	0.10	0.10
Current Account to GDP	-0.32	0.10	…
Investment	-0.56	-0.42	-0.45
Goverment Deficit to GDP	0.48	-0.00	0.11
Private Saving to GDP	0.12	0.04	0.06
General Transfers to GDP	-0.14	-0.06	-0.08
Financed by an Increase in Consumption Taxes
Real GDP	-0.38	-0.18	-0.23
Real Exchange Rate	0.79	-0.79	…
Real Interest Rate	0.10	0.10	0.10
Current Account to GDP	-0.33	0.10	…
Investment	-0.64	-0.45	-0.49
Goverment Deficit to GDP	0.48	-0.00	0.11
Private Saving to GDP	0.11	0.05	0.06
Consumption Tax Rate	0.35	0.00	0.08
Financed by an Increase in Corporate Income Taxes
Real GDP	-0.71	-0.23	-0.33
Real Exchange Rate	-0.01	0.01	…
Real Interest Rate	0.09	0.09	0.09
Current Account to GDP	-0.30	0.09	…
Investment	-1.92	-0.47	-0.79
Goverment Deficit to GDP	0.48	-0.00	0.11
Private Saving to GDP	-0.03	0.04	0.03
Corporate Income Tax Rate	1.35	0.00	0.30
Financed by an Increase in Labor Income Taxes
Real GDP	-0.48	-0.21	-0.27
Real Exchange Rate	0.74	-0.75	…
Real Interest Rate	0.11	0.11	0.11
Current Account to GDP	-0.36	0.11	…
Investment	-0.77	-0.50	-0.56
Goverment Deficit to GDP	0.48	0.00	0.11
Private Saving to GDP	0.08	0.05	0.05
Labor Income Tax Rate	0.44	0.00	0.10

Table 2:

Effects of a Permanent 10 Percentage Point Increase in the U.S. Government Debt to GDP Ratio

	US	RoW	Global
Financed by a Cut in General Transfers
Real GDP	-0.30	-0.17	-0.20
Real Exchange Rate	0.88	-0.88	…
Real Interest Rate	0.10	0.10	0.10
Current Account to GDP	-0.32	0.10	…
Investment	-0.56	-0.42	-0.45
Goverment Deficit to GDP	0.48	-0.00	0.11
Private Saving to GDP	0.12	0.04	0.06
General Transfers to GDP	-0.14	-0.06	-0.08
Financed by an Increase in Consumption Taxes
Real GDP	-0.38	-0.18	-0.23
Real Exchange Rate	0.79	-0.79	…
Real Interest Rate	0.10	0.10	0.10
Current Account to GDP	-0.33	0.10	…
Investment	-0.64	-0.45	-0.49
Goverment Deficit to GDP	0.48	-0.00	0.11
Private Saving to GDP	0.11	0.05	0.06
Consumption Tax Rate	0.35	0.00	0.08
Financed by an Increase in Corporate Income Taxes
Real GDP	-0.71	-0.23	-0.33
Real Exchange Rate	-0.01	0.01	…
Real Interest Rate	0.09	0.09	0.09
Current Account to GDP	-0.30	0.09	…
Investment	-1.92	-0.47	-0.79
Goverment Deficit to GDP	0.48	-0.00	0.11
Private Saving to GDP	-0.03	0.04	0.03
Corporate Income Tax Rate	1.35	0.00	0.30
Financed by an Increase in Labor Income Taxes
Real GDP	-0.48	-0.21	-0.27
Real Exchange Rate	0.74	-0.75	…
Real Interest Rate	0.11	0.11	0.11
Current Account to GDP	-0.36	0.11	…
Investment	-0.77	-0.50	-0.56
Goverment Deficit to GDP	0.48	0.00	0.11
Private Saving to GDP	0.08	0.05	0.05
Labor Income Tax Rate	0.44	0.00	0.10

VI. Effects of G-20 Fiscal Packages

Table 3 sets out the components of the fiscal stimulus packages that are being implemented over 2009 and 2010 by the G20 countries.¹² Japan, emerging Asia and the United States have announced the largest packages, while the G20 countries in the euro area, Africa and Latin America have the smallest packages.

Topics

Countries

Series

About

Resources

Contributor Notes

Abstract

I. Introduction

II. Fiscal Multipliers: Empirical and Model-Based Evidence

III. Multipliers, Instruments, and Models

A. Definition of Fiscal Multipliers

B. The Seven Fiscal Instruments

C. The Seven Structural Models

IV. Fiscal Multipliers for Temporary Stimulus

A. Introduction to Simulation Results

B. Government Investment - Detailed Discussion

C. The Role of Monetary Accommodation

D. The Persistence of Fiscal Stimulus

E. Government Spending versus Taxes and Transfers

F. Different Taxes and Transfers

G. General Transfers versus Targeted Transfers

H. United States versus Europe

V. Permanently Higher Deficits and Debt

A. Permanently Higher Deficits and Short-Run Multipliers

B. Permanently Higher Deficits and Long-Run Crowding-Out

VI. Effects of G-20 Fiscal Packages