I. Introduction¹

There has been a renewed interest in the macroeconomic effects of fiscal policy since the 2007–08 food and fuel price hikes followed by the global financial crisis. While fiscal stimulus could smooth the negative effects of shocks, many economies were facing hard budget constraints. Several analysts, thus, feared that more debt accumulation would be counterproductive, as negative expectations about fiscal sustainability could raise market interest rates and end up having a contractionary effect on aggregate production. In this context, the adequacy of fiscal stimulus would depend on public debt levels and external financing needs—key determinants of the decision to consolidate fiscal accounts. Adding to the policy challenges, little information about the impact of fiscal stimulus on real activity (the so-called “fiscal multipliers”) has hampered the evaluation of trade-offs between stimulus and consolidation in several regions of the world. These issues have been debated publicly. For instance, recent IMF publications (IMF, 2012a and b)² have focused on how fiscal policy has influenced the economic outlook during the crisis. In particular, IMF (2012b) finds that fiscal multipliers appear to be higher than what is generally assumed by forecasters, pointing to still-pervasive uncertainty on how to calculate fiscal multipliers in a diverse country setting.

Against this backdrop, this paper develops a methodology to calculate fiscal multipliers using annual data and accounting for feedback effects from debt accumulation to growth. The policy discussion focuses on regions of the world with scant evidence on fiscal multipliers while addressing the following questions: Systematically accounting for the size of the public debt and its dynamic, how can growth-friendly fiscal consolidation be enacted in understudied regions of the world? Are improvements in tax collection less disruptive to growth than spending cuts? Is curbing current spending less disruptive to growth than curbing capital spending?

To answer these questions, the paper estimates short- and medium-term fiscal multipliers for Central American countries (for the first time in the literature), and a set of advanced, emerging market, oil exporter, low-income, heavily-indebted poor, and Sub-Saharan African countries. It builds on the structural vector autoregressive approach by Blanchard and Perotti (2002) to identify exogenous fiscal policy shocks using quarterly data and on the setup of Giavazzi (2007) to explicitly account for the effect of public debt feedback on fiscal multipliers. Because one important motivation for this paper is to provide estimates for Central American countries and other regions that do not have quarterly fiscal data, we depart from Blanchard and Perotti (2002) and use cointegrating techniques to identify key terms in the vector autoregression’s (VAR) variance-covariance matrix. The proposed modeling approach also assesses the impact of different modalities of fiscal consolidation—current and capital expenditure cuts or tax revenue increases—on economic growth for specific countries and various country grouping.³ To compare the estimated multipliers across countries, annual data are employed from 1972 to 2010 even for groupings of countries possessing quarterly fiscal data and previous estimates of fiscal multipliers.

The results show that fiscal multipliers are generally significantly different from zero (at least in the first three years following fiscal policy shocks). They imply a positive medium-term effect of fiscal consolidation on output in most Central American and low-income countries, more so if this consolidation is based on either increases in tax revenues or cutting government’s current spending (while, in general, cuts in capital spending are bad for short-and medium-term output). Advanced and emerging market economies are also shown to post better medium-term output effects when fiscal consolidation focuses on current spending cuts, although the medium-term effects of fiscal consolidation in these countries remain negative—results consistent with evidence in recent papers.

The contributions of the paper to the literature are twofold. First, the estimated fiscal multipliers cover a wide range of countries, from heavily indebted countries and low-income countries to emerging market and advanced countries. Given that many of these countries face the challenges of fiscal consolidation, the analysis takes debt feedback into account when estimating the multipliers. Second, given limited high frequency data across countries, the paper employs annual data and pays greater attention to the identification approach by using structural vector error-correction model, SVECM, techniques.

The rest of the paper is organized as follows. Section II briefly discusses past attempts to address identification issues when estimating fiscal multipliers. Section III presents an econometric model strategy for identifying fiscal shocks with annual data and debt feedback effects. Section IV provides a brief discussion of recent fiscal developments in Central America, Panama and the Dominican Republic (CAPDR), followed by a discussion of the estimated fiscal multipliers for countries in the region, as well as comparator country grouping. These groups include low-income countries (LIC), heavily indebted countries (HIPC), sub-Saharan Africa (SSA), emerging market countries (EME), oil producing countries, and advanced economics (AE). Section V concludes the paper.

II. Past Attempts to Identify Discretionary Fiscal Policy Changes

A fiscal multiplier is the ratio between output change and an exogenous variation in the fiscal deficit with respect to their baseline values. Past work has shown that the size of the fiscal multiplier depends on the country, its business cycle stance, time period, special circumstances (including monetary and exchange rate regimes, its degree of financial integration, and the extent of openness, etc), and the methodology used to estimate them.⁴ In addition, instantaneous impacts and cumulative effects of fiscal shocks can differ. Multipliers can even be negative—a phenomenon named “contractionary fiscal expansions”—if, for instance, wider interest-rate spreads depress economic activity despite the initial positive impulse from larger net government demand.

Identifying exogenous fiscal policy shocks is an essential step for estimating the output multipliers. In essence, estimating fiscal multipliers requires isolating fiscal policy shocks from the initial influence of economic conditions, due to the bi-directional causality between government spending or tax revenue and growth. The most promising strand of research aiming to isolate discretionary fiscal policy shocks uses vector-autoregression methods (VARs). To date, this strand exhibits five main lines of inquiry for identification of policy shocks.

The first is the recursive approach originally introduced by Sims (1980) and applied by Fatás and Mihov (2001) in the study of fiscal policy, which assumes a causal ordering (from most exogenous to the most endogenous) of variables.

The second is the event-study approach (also called the fiscal-dummy approach) introduced by Ramey and Shapiro (1998). This strategy assumes that the U.S. government spending increases associated with the Korean War, the Vietnam War, and the Reagan military buildup were exogenous to the state of the economy. Thus, in those cases, there is no need to identify the structural form of the VAR model and the analysis can be based on a reduced-form VAR. This line of argumentation was also used more generally to study the effects of large unexpected increases in government defense spending by Edelberg et al. (1999), Eichenbaum and Fisher (2005), Perotti (2007), and Ramey (2007) among others. This approach is now generally called “the narrative” approach as recently developed by Romer and Romer (2010), and Favero and Giavazzi (2012). The framework typically, first attempts to determine the size and the timing of truly exogenous changes in fiscal outcomes (say, “legislated tax”) using information from policy documents. Then, the “narrative” shocks are shown (e.g. Favero and Giavazzi, 2011) to be orthogonal to the relevant information in the SVAR model, thus dispensing the inversion of the moving-average representation of a VAR for identification of fiscal shocks.

The third approach is related to the sign-restriction literature, which identify models by imposing restrictions on the acceptable impulse responses to shocks (Faust, 1998, and Canova and de Nicoló, 2002). The application to fiscal policy was introduced by Uhlig (2005) and Mountford and Uhlig (2005). The approach does not impose any zero restrictions on the impulse responses, instead restricting the sign of the impulse responses for a number of quarters after a shock occurred. Mountford and Uhlig (2005) address the problem of the positive correlation of output and revenue residuals by first identifying a business cycle shock and, then, requiring the government revenue shock to be orthogonal to the business cycle shock. An advantage of this approach is that unlike the recursive approach, it does not impose a zero restriction on the initial output response to a revenue shock. Also, unlike several structural VAR approaches, it does not require a two-step estimation procedure. However, a disadvantage of this approach is that by attributing a positive correlation between output and revenue residuals entirely to business cycle shocks, it rules out (by construction) non-Keynesian effects of fiscal policy (see, e.g., Giavazzi et al., 2000).

The fourth approach is the now popular normalization and restrictions on the contemporaneous relationships between variables in a Structural VAR (SVAR). While it has been widely applied to monetary policy (see Bagliano and Favero, 1998, for a review), it was first proposed by Blanchard and Perotti (2002) and extended by Perotti (2005). These authors use (i) institutional information about tax, transfers, and spending programs to construct parameters, and (ii) the reaction timing of policymakers to GDP shocks to identify parts of the variance-covariance matrix. They make the crucial assumption that policymakers and legislatures would take more than a quarter to learn about a GDP shock and decide which fiscal measures to take in response of a shock. This strategy cannot be used with annual data.

The fifth approach takes into consideration the longer-run properties of fiscal variables, economic activity, and other endogenous variables that enter the VAR system, generally in the form of a vector error-correction model (VECM). The VECM captures these long-run properties of the VAR system variables through their cointegrating relationships. Hence, as shown in Pagan and Pesaran (2007), the VECM provides a further identification tool. This approach can be seen as an extension of the Blanchard and Quah (1989) methodology as there is a clear correspondence between SVARs and VECMs.^5,6

III. Identifiying Fiscal Policy Shocks with Debt Feedback and Annual Data

As discussed in the previous section, there are several ways to undertake identification of fiscal policy shocks in SVARs. However, because many countries do not have high frequency information as a group, this paper develops a way to identify exogenous fiscal policy shocks using annual data.⁷ In particular, we nest the traditional short-run restrictions and take into account the long-run properties of the model through the use of cointegrating relationships (Pagan and Pesaran, 2008; and Dungrey and Fry, 2009) to identify fiscal shocks. Further, we follow Favero and Giavazzi (2007) which extends the Blanchard and Perotti (2002) and Perotti (2005) SVAR to account for a government’s budget constraint. We consider the following SVAR model

The evolution of the debt ratio depends on the government budget constraint

Where the vector of endogenous variables, Y_t = [g_t, t_t, y_t, reer_t, i_t]′, includes government spending (g_t) defined as the sum of government consumption and investment (excluding interest payment), net revenue excluding interest receipt on government debt (t_t), real output (y_t), real effective exchange rate (reer_t), and the yield on government securities (reer_t). ${\epsilon }_t={\left[{\epsilon }_t^g,{\epsilon }_t^t,{\epsilon }_t^y,{\epsilon }_t^{\mathit{er}},{\epsilon }_t^i\right]}^{\prime }$ is the vector of structural shocks to the endogenous variables and $e_t={\left[e_t^g,e_t^t,e_t^y,e_t^{\mathit{er}},e_t^i\right]}^{\prime }$ is the corresponding innovation. A₀ is the matrix of contemporaneous parameters, L is the lag operator, A(L) is the matrix of VAR parameters, and B is the structural matrix associated with innovations.

Identification is achieved with assumptions about policy decision lags and estimated elasticities through cointegrating properties. As defined in Banchard and Perotti (2002) and in Perotti (2005), observed fiscal policy reactions (expenditure and tax):

Or

are functions of (i) automatic responses of spending/ tax to output, exchange rate, and financial shocks; (ii) systematic discretionary responses of fiscal policy to macroeconomic system; and (iii) random discretionary fiscal policy shocks. The relation between the structural shocks and the innovation is thus given by:

Or in an equation format: A₀e_t = Bε_t.

A just-identification of [5] would require $\frac{5\left(5+1\right)}{2}=15$ restrictions on the A₀ matrix (left-hand-side of [5]), implying that 5 additional restrictions are needed.

1. $\begin{array}{cc}{\alpha }_{g,i}={\alpha }_{t,i}=0& \left[6\right]\end{array}$

   on the grounds that interest payments on government debt are excluded from the definition of expenditures and taxes used in the model. Furthermore, the contemporaneous impact of interest rates on taxes is likely to be small or close to zero in practice.⁸

2. The interest rate on government debt depends on the fiscal stance and the exchange rate, but (contemporaneously) less on output. Thus:

   $\begin{array}{cc}{\gamma }_{i,y}=0& \left[7\right]\end{array}$

3. We now need (at least) 2 restrictions. Most studies using high frequency data have assumed that either expenditures or taxes do not respond to economic activity within a quarter (α_g,y = 0 and α_t,y = 0). Such assumption may not hold for annual data and that is why we use cointegration analysis to impose the missing identification assumptions. Suppose that there is at least one cointegrating relationship (which is likely to be the case, given that, by construction, all system variables enter in level and generally follow I(1) processes), then estimates of the automatic response of taxes to changes in the economic environment and the exchange rate or the automatic response of government spending to economic or exchange rate shocks could be obtained.

More precisely, with one cointegrating vector,⁹ the structural VECM counterpart of the baseline model [1] is:

where a = A₀α and α is the loading parameter.

With such cointegrating relationship between government spending, output growth, and exchange rate, the remaining two coefficients can be obtained by:¹⁰

Parameters of [3’] can be proxied by estimates from [9]. With this strategy, all fiscal shocks are identified and the matrix A₀ can be fully estimated.

We now extend the previous model to six endogenous variables in which we split total expenditure (excluding interest payment on debt) into capital expenditure (labeled by “ca”) and current spending (labeled by “cu”). The expanded equation [5] with 6 endogenous variables is:

Hence, this will require $\frac{6\left(6+1\right)}{2}=21$ restrictions in which 15 are already obtained.

Following [6] and [7], we have α_ca,i = α_cu,i = α_t,i = 0 and γ_i,y = 0 respectively. Finally, using co-integration properties, the two remaining coefficient are determined.

IV. Estimation Results¹¹

A. Fiscal Policy in Central America as a case study

Fiscal policy was, in general, procyclical in Central America (including Panama and the Dominican Republic or CAPDR) through the late 1990s. As argued by Talvi and Végh (2000), this stylized fact seems to be valid for most of Latin America and the Caribbean. The authors show that procyclical fiscal policy contributed to deepen business cycles in the region, as governments generally relax their policies during booms and restrict them during busts due to weak institutions, unfavorable political economy, and volatile access to international capital markets.

However, following a decade of adjustment, CAPDR used counter-cyclical fiscal policy to withstand the late 2000s crises. The food and fuel price hikes (2007–08) and the recent global financial crisis hit CAPDR severely. Thanks to strong policy buffers built in the 2000s, appropriate counter-cyclical fiscal policy responses allowed CAPDR withstand the adverse effects of the crises. Lopez-Mejia (2012) argues that most countries accommodated lower import-related tax collections (import duties and indirect taxes levied on imports) and lower domestic-tax receipts during the 2008–09 downturn; and that at the same time, government expenditures increased (except for the Dominican Republic) to support demand.¹²

Despite the general countercyclical response in the region, there are clusters within CAPDR, with additionally important cross-country differences. Indeed, Morra (2011) shows that the size of the fiscal stimulus varied across countries. For instance, in Nicaragua the fiscal stimulus was close to zero and in Panama it was quite large. Moreover, among the countries in the region that have implemented some type of fiscal stimulus during the global financial crisis, the measures did not seem to produce similar outcomes, probably because of differences in the degree of policy credibility, financial integration, and dollarization, among other factors.

• Download Figure
• Download figure as PowerPoint slide

View Full Size

• Download Figure
• Download figure as PowerPoint slide

• Download Figure
• Download figure as PowerPoint slide

Following a generally countercyclical fiscal policy in CAPDR, public debt could be reduced in some countries, given their high vulnerability to weather and external shocks, which would require policy buffers.¹³ (Figure 1) In addition, while fiscal policy can boost growth through its multiplier, high public debt could dampen its effectiveness. Finally, reducing debt would help to improve the countries’ sovereign credit risk rating.¹⁴ While consolidation is a major issue in some of these countries, others may need to raise taxes from very low levels to improve the provision of public services. In this context, knowing how fiscal policy changes would affect economic performance in the region is an important priority.

Central America and Regional Peers: Selected Economic Indicators

Citation: IMF Working Papers 2013, 136; 10.5089/9781484304488.001.A001

Sources: IMF World Economic Outlook; and authors’ calculations.

• Download Figure
• Download figure as PowerPoint slide

View Full Size

Central America and Regional Peers: Selected Economic Indicators

Citation: IMF Working Papers 2013, 136; 10.5089/9781484304488.001.A001

Sources: IMF World Economic Outlook; and authors’ calculations.

• Download Figure
• Download figure as PowerPoint slide

Central America and Regional Peers: Selected Economic Indicators

Citation: IMF Working Papers 2013, 136; 10.5089/9781484304488.001.A001

Sources: IMF World Economic Outlook; and authors’ calculations.

• Download Figure
• Download figure as PowerPoint slide

C. Overall Results

The results suggest that, in general, fiscal consolidation in less developed countries hurts output only in the short-term. (Figures 2 to 4) The negative short-term effect is largest for advanced economies (AEs), significant for emerging markets (EMEs), and small for less developed economies—a result consistent with the evidence presented in Ilzetzki, Mendoza, and Végh (2010). Fiscal consolidation based on expenditure cuts tends to produce the best medium-term output effects in advanced and emerging economies, but not necessarily in less developed countries. Results for advanced economies (AEs) and emerging markets (EMs) broadly match the evidence in other papers (Appendix, Table 6), which validates the novel method we propose.¹⁹ The cumulative multipliers—the ratios of cumulative increase in the net present value of GDP and the cumulative increase in the net present value of the fiscal variable—are computed from the impulse responses (Figures 5–11). Dashed lines show the boundaries of the 90 percent confidence interval levels (estimated by Monte Carlo simulations with bootstrapped standard errors and 500 repetitions). In general, the impact and cumulative multipliers are statistically significant from zero at the 90 percent confidence level through each horizon—but often not at the 95 percent confidence interval—with wider error bands at longer horizons.

Real GDP Growth Effects of a One-Percentage-Point Increase in Expenditure or Cut in Tax Revenue–A SVECM Approach with Debt Feedback ^{1/, 2/, 3/} (Impact)

Citation: IMF Working Papers 2013, 136; 10.5089/9781484304488.001.A001

Source: Authors’ estimates.^1/ Response of output growth to a one-standard deviation of expenditure and tax shocks, rescaled to output growth response to a one-percentage-point increase in expenditure and in taxes. The cumulative multipliers are obtained as ratios of net present values as defined in Section IV.B^2/ The identification scheme utilizes the institutional assumption (Blanchard and Perroti, 2002) and the Long-run employs coefficient from the system long-run cointegration estimates.^3/ “Tax revenue” referrers to total tax collection and tax base (perhaps at given tax rates).

• Download Figure
• Download figure as PowerPoint slide

View Full Size

Real GDP Growth Effects of a One-Percentage-Point Increase in Expenditure or Cut in Tax Revenue–A SVECM Approach with Debt Feedback ^{1/, 2/, 3/} (Impact)

Citation: IMF Working Papers 2013, 136; 10.5089/9781484304488.001.A001

Source: Authors’ estimates.^1/ Response of output growth to a one-standard deviation of expenditure and tax shocks, rescaled to output growth response to a one-percentage-point increase in expenditure and in taxes. The cumulative multipliers are obtained as ratios of net present values as defined in Section IV.B^2/ The identification scheme utilizes the institutional assumption (Blanchard and Perroti, 2002) and the Long-run employs coefficient from the system long-run cointegration estimates.^3/ “Tax revenue” referrers to total tax collection and tax base (perhaps at given tax rates).

• Download Figure
• Download figure as PowerPoint slide

Real GDP Growth Effects of a One-Percentage-Point Increase in Expenditure or Cut in Tax Revenue–A SVECM Approach with Debt Feedback ^{1/, 2/, 3/} (Impact)

Citation: IMF Working Papers 2013, 136; 10.5089/9781484304488.001.A001

Source: Authors’ estimates.^1/ Response of output growth to a one-standard deviation of expenditure and tax shocks, rescaled to output growth response to a one-percentage-point increase in expenditure and in taxes. The cumulative multipliers are obtained as ratios of net present values as defined in Section IV.B^2/ The identification scheme utilizes the institutional assumption (Blanchard and Perroti, 2002) and the Long-run employs coefficient from the system long-run cointegration estimates.^3/ “Tax revenue” referrers to total tax collection and tax base (perhaps at given tax rates).

• Download Figure
• Download figure as PowerPoint slide

Real GDP Growth Effects of a One-Percentage-Point Increase in Expenditure or Cut in Tax Revenue–A Structural VAR Approach with Debt Feedback ^{1/, 2/, 3/,} (Cumulative)

Citation: IMF Working Papers 2013, 136; 10.5089/9781484304488.001.A001

Source: Authors’ estimates.^1/Response of output growth to a one-standard deviation of expenditure and tax shocks, rescaled to output growth response to a one-percentage-point increase in expenditure and in taxes. The cumulative multipliers are obtained as ratios of net present values as defined in Section IV.B^2/The identification scheme utilizes the institutional assumption (Blanchard and Perroti, 2002) and the Long-run employs coefficient from the system long-run cointegration estimates.^3/“Tax revenue” referrers to total tax collection and tax base (perhaps at given tax rates).

• Download Figure
• Download figure as PowerPoint slide

View Full Size

Real GDP Growth Effects of a One-Percentage-Point Increase in Expenditure or Cut in Tax Revenue–A Structural VAR Approach with Debt Feedback ^{1/, 2/, 3/,} (Cumulative)

Citation: IMF Working Papers 2013, 136; 10.5089/9781484304488.001.A001

Source: Authors’ estimates.^1/Response of output growth to a one-standard deviation of expenditure and tax shocks, rescaled to output growth response to a one-percentage-point increase in expenditure and in taxes. The cumulative multipliers are obtained as ratios of net present values as defined in Section IV.B^2/The identification scheme utilizes the institutional assumption (Blanchard and Perroti, 2002) and the Long-run employs coefficient from the system long-run cointegration estimates.^3/“Tax revenue” referrers to total tax collection and tax base (perhaps at given tax rates).

• Download Figure
• Download figure as PowerPoint slide

Real GDP Growth Effects of a One-Percentage-Point Increase in Expenditure or Cut in Tax Revenue–A Structural VAR Approach with Debt Feedback ^{1/, 2/, 3/,} (Cumulative)

Citation: IMF Working Papers 2013, 136; 10.5089/9781484304488.001.A001

Source: Authors’ estimates.^1/Response of output growth to a one-standard deviation of expenditure and tax shocks, rescaled to output growth response to a one-percentage-point increase in expenditure and in taxes. The cumulative multipliers are obtained as ratios of net present values as defined in Section IV.B^2/The identification scheme utilizes the institutional assumption (Blanchard and Perroti, 2002) and the Long-run employs coefficient from the system long-run cointegration estimates.^3/“Tax revenue” referrers to total tax collection and tax base (perhaps at given tax rates).

• Download Figure
• Download figure as PowerPoint slide

Impact and Cumulative Output Multipliers of Capital and Current Expenditures – A Structural VAR Approach with Debt Feedback

Citation: IMF Working Papers 2013, 136; 10.5089/9781484304488.001.A001

Source: Authors’ estimates.Note: Reponse of output growth to a one-standard deviation of expenditure and tax shocks, rescaled to output growth response to a one-percentage-point increase in expenditure and in taxes. The identification scheme utilizes the institutional assumption (Blanchard and Perroti, 2002) and the Long-run employs coefficient from the system long-run cointegration estimates. “Tax revenue” referrers to total tax collection and tax base (perhaps at given tax rates). The cumulative multipliers are obtained as ratios of net present values as defined in Section IV.B

• Download Figure
• Download figure as PowerPoint slide

View Full Size

Impact and Cumulative Output Multipliers of Capital and Current Expenditures – A Structural VAR Approach with Debt Feedback

Citation: IMF Working Papers 2013, 136; 10.5089/9781484304488.001.A001

Source: Authors’ estimates.Note: Reponse of output growth to a one-standard deviation of expenditure and tax shocks, rescaled to output growth response to a one-percentage-point increase in expenditure and in taxes. The identification scheme utilizes the institutional assumption (Blanchard and Perroti, 2002) and the Long-run employs coefficient from the system long-run cointegration estimates. “Tax revenue” referrers to total tax collection and tax base (perhaps at given tax rates). The cumulative multipliers are obtained as ratios of net present values as defined in Section IV.B

• Download Figure
• Download figure as PowerPoint slide

Impact and Cumulative Output Multipliers of Capital and Current Expenditures – A Structural VAR Approach with Debt Feedback

Citation: IMF Working Papers 2013, 136; 10.5089/9781484304488.001.A001

Source: Authors’ estimates.Note: Reponse of output growth to a one-standard deviation of expenditure and tax shocks, rescaled to output growth response to a one-percentage-point increase in expenditure and in taxes. The identification scheme utilizes the institutional assumption (Blanchard and Perroti, 2002) and the Long-run employs coefficient from the system long-run cointegration estimates. “Tax revenue” referrers to total tax collection and tax base (perhaps at given tax rates). The cumulative multipliers are obtained as ratios of net present values as defined in Section IV.B

• Download Figure
• Download figure as PowerPoint slide

Surprisingly, the positive long-run effect of tax increases on output tend to be higher than the positive effect of cutting spending in heavily indebted poor countries (HIPCs), low-income countries (LICs), and sub-Saharan Africa (SSA). This result suggests that the stylized fact that fiscal stabilization focused on spending cuts have better growth outcomes may not apply to poorer economies. Such an outcome may be caused by inefficient tax administration in those countries and observed increases in tax revenues may be the result of improved efficiency with little distortive impacts. Most Central American countries appear to have larger tax and spending multipliers in the medium term than other countries in a similar stage of development. Specific estimates for these countries are shown in Table 1 and Figures 5-11 in the appendix. The following sections go over these results in more detail.

Table 1.

Cumulative Output Multipliers with Debt Feedback and Financial Constraints: Current and Capital Expenditure

Note: Response of output growth to capital and current expenditure (net of interest) cut (shocks). Upper-band and lower-band are 90th percentile and 10th percentile, respectively. Breakdown of total expenditure between current expenditure (net of interest) and capital expenditure were not available for all country groups from 1970 to 2011.

Table 1.

Cumulative Output Multipliers with Debt Feedback and Financial Constraints: Current and Capital Expenditure

			Current expenditure		Capital expenditure
			Impact	Cumulative	Impact	Cumulative
Central America-Dominacan Republic and Panama
	Costa Rica		-0.0436	0.7926	-0.3835	-0.7264
		Lower-band	-0.035	0.951	-0.326	-0.530
		Upper-band	-0.0523	0.6341	-0.4410	-0.9225
	Dominican Republic		-0.0392	0.1555	-0.2161	-0.4236
		Lower-band	-0.031	0.187	-0.184	-0.309
		Upper-band	-0.0471	0.1244	-0.2485	-0.5380
	El Savador		-0.0453	0.7257	-0.3604	-0.7064
		Lower-band	-0.039	0.871	-0.306	-0.516
		Upper-band	-0.0521	0.5805	-0.4145	-0.8972
	Guatemala		-0.1705	0.3384	-0.3990	-0.5267
		Lower-band	-0.145	0.406	-0.339	-0.385
		Upper-band	-0.1961	0.2707	-0.4589	-0.6689
	Honduras		-0.0845	0.7310	-0.3783	-0.5925
		Lower-band	-0.072	0.877	-0.322	-0.433
		Upper-band	-0.0971	0.5848	-0.4350	-0.7524
	Nicaragua		-0.0486	0.7374	-0.2117	-0.5836
		Lower-band	-0.041	0.885	-0.180	-0.426
		Upper-band	-0.0559	0.5899	-0.2435	-0.7411
	Panama		-0.1867	-0.3951	-0.7006	-1.0033
		Lower-band	-0.159	-0.288	-0.595	-0.732
		Upper-band	-0.2147	-0.5018	-0.8056	-1.2742

Note: Response of output growth to capital and current expenditure (net of interest) cut (shocks). Upper-band and lower-band are 90th percentile and 10th percentile, respectively. Breakdown of total expenditure between current expenditure (net of interest) and capital expenditure were not available for all country groups from 1970 to 2011.

View Table

Table 1.

Cumulative Output Multipliers with Debt Feedback and Financial Constraints: Current and Capital Expenditure

			Current expenditure		Capital expenditure
			Impact	Cumulative	Impact	Cumulative
Central America-Dominacan Republic and Panama
	Costa Rica		-0.0436	0.7926	-0.3835	-0.7264
		Lower-band	-0.035	0.951	-0.326	-0.530
		Upper-band	-0.0523	0.6341	-0.4410	-0.9225
	Dominican Republic		-0.0392	0.1555	-0.2161	-0.4236
		Lower-band	-0.031	0.187	-0.184	-0.309
		Upper-band	-0.0471	0.1244	-0.2485	-0.5380
	El Savador		-0.0453	0.7257	-0.3604	-0.7064
		Lower-band	-0.039	0.871	-0.306	-0.516
		Upper-band	-0.0521	0.5805	-0.4145	-0.8972
	Guatemala		-0.1705	0.3384	-0.3990	-0.5267
		Lower-band	-0.145	0.406	-0.339	-0.385
		Upper-band	-0.1961	0.2707	-0.4589	-0.6689
	Honduras		-0.0845	0.7310	-0.3783	-0.5925
		Lower-band	-0.072	0.877	-0.322	-0.433
		Upper-band	-0.0971	0.5848	-0.4350	-0.7524
	Nicaragua		-0.0486	0.7374	-0.2117	-0.5836
		Lower-band	-0.041	0.885	-0.180	-0.426
		Upper-band	-0.0559	0.5899	-0.2435	-0.7411
	Panama		-0.1867	-0.3951	-0.7006	-1.0033
		Lower-band	-0.159	-0.288	-0.595	-0.732
		Upper-band	-0.2147	-0.5018	-0.8056	-1.2742

Note: Response of output growth to capital and current expenditure (net of interest) cut (shocks). Upper-band and lower-band are 90th percentile and 10th percentile, respectively. Breakdown of total expenditure between current expenditure (net of interest) and capital expenditure were not available for all country groups from 1970 to 2011.

V. Concluding Remarks

Given the pressing need for fiscal consolidation in many countries in the current global economic environment, this paper estimates the real effect of fiscal consolidation in various country groupings, while offering some more detailed discussion of Central American countries—a region with scant evidence on fiscal multipliers. The estimation procedure modified traditional structural VAR modeling to account for debt feedback effects and to enable the use of annual data. In particular, given that the use of annual data precludes using identification assumptions based on the quarterly timing of fiscal policy reactions to output changes, we use cointegration techniques and error-correction modeling to identify exogenous fiscal shocks and estimate fiscal multipliers. The inclusion of debt as a ratio to GDP in the estimation (as suggested by Favero and Giavazzi, 2007) allows for attenuating effects from changes in the fiscal deficit. For instance, as public debt declines as a result of a smaller fiscal deficit, interest rates would also decline, potentially undoing part of the initial negative impulse. The paper also compares the results from applying the proposed procedure and data frequency to advanced and emerging market economies with estimates from other recent papers, which validates the methodology proposed here.

Our results are comparable to the literature and suggest that fiscal consolidation provides some medium-term growth benefit in many countries and regions. The positive effects of fiscal consolidation tend to be inversely related to a country’s stage of development. Fiscal consolidation tends to have a strong positive effect in low income countries (especially HIPCs), although that is also the case in some mid-income countries, as Costa Rica (or Panama and the Dominican Republic if fiscal consolidation is based on improved tax system efficiency). Estimates for advanced and emerging market economies confirm that fiscal consolidation tend to hurt short- and medium-term output.

The composition of the fiscal consolidation effort also matters. For instance, tax revenue increases appear to yield greater medium-term output growth than current spending control in several Central American countries and LICs, including HIPCs. However, the public investment multiplier is found to be larger than tax multipliers. Estimates for CAPDR show that cutting capital spending would hurt short- and medium-term growth significantly, while curbing current spending would raise output in Nicaragua, Honduras, El Salvador, and (to a lesser extent and only in the medium term) Guatemala. Nicaragua and Honduras show greater benefits from cutting current spending than raising tax revenues.