## 1. Fiscal multipliers based on the direct effects of fiscal shocks .................................12

## Source details

**Canonical URL:** [1. Fiscal multipliers based on the direct effects of fiscal shocks .................................12](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2013/_wp13117.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2013/_wp13117.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2013/_wp13117.pdf.json)

---

### Major themes
- Title as provided: "Fiscal multipliers based on the direct effects of fiscal shocks .................................12"
- Supplied excerpt for this unit contains only the section heading and page marker: "12".
- No additional substantive text for this specific heading was provided in the supplied content unit.

### Context and purpose (from related supplied material)
- Objective: Quantitatively analyze effects of government revenue and expenditures on output dynamics in the Eastern Caribbean Economic and Monetary Union (ECCU).
- Contextual facts:
  - In 2008–11, total central government revenues averaged 26 percent of GDP while total expenditure was on average about 30 percent of GDP.
  - Quarterly series sample period: 1994:Q1 to 2009:Q4.
  - Number of quarterly observations available for each ECCU member is about 64, at most.

### Methodology overview (as supplied)
- Estimation framework:
  - Structural vector autoregression (SVAR) models using panel data for eight ECCU members.
  - Panel-VAR pooling justified by common currency and similar shock environment.
  - Fiscal shocks studied: tax revenue (tax), government consumption expenditure (con), government investment expenditure (inv).
- Data construction:
  - Fiscal variables transformed to natural logs, seasonally adjusted, and de-trended by subtracting a quadratic time trend; cyclical series named tax, con, inv, and gdp.
  - Quarterly GDP interpolated from annual GDP using the Chow and Lin (1971) method with quarterly proxies.
- Model specification:
  - Bivariate VARs (each fiscal variable with gdp) and a multivariate VAR (tax, inv, con, gdp).
  - Matrix A assumed lower triangular; ordering in multivariate VAR: inv → con → tax → gdp.
  - Control variables: current account-to-GDP ratio, monetary policy interest rate, public debt-to-GDP ratio, real exchange rate (log deviation), and U.S. GDP (log deviation).
  - Four lags used in estimations.
- Multiplier calculation details:
  - Impact multiplier (IM) at t=1: IM = (Δgdp1 / Δcon1) ÷ (CON/Y).
  - Cumulative multiplier (CM) computed over horizons; long-run multiplier operationally taken at 24 quarters.
  - Confidence intervals from 500 Monte Carlo simulations; reported multipliers are simulation means.

### Results — direct (bivariate) effects (exact figures preserved)
- Government consumption expenditure (bivariate):
  - Impact multiplier at t=1: 0.20.
  - 95 percent confidence interval at impact: −0.13 to 0.54.
  - One-year cumulative multiplier: about 0.42.
  - Statistical conclusion: not statistically different from zero over projection period.
- Government investment expenditure (bivariate):
  - Impact multiplier at t=1: 0.12.
  - Cumulative multiplier after four quarters: 0.37.
  - After four years multiplier about 0.44.
  - 95 percent confidence interval above zero for cumulative effects after a short horizon (statistically different from zero).
  - Interpretation: positive but < 1; substantial portion of impulse fades, likely due to high import content of investment.
- Tax revenue (bivariate):
  - Impact multiplier at t=1: about −0.5.
  - Subsequent periods: impulse-response turns positive but remains not statistically different from zero.
  - Interpretation: unexpected tax shock does not have a significant lasting effect on output in bivariate estimations.

### Results — multivariate (full interaction) effects (exact figures preserved)
- Multivariate VAR (24-quarter horizon):
  - Government investment expenditure multiplier:
    - Long-run value reaches 0.62 (statistically different from zero).
    - Interaction with other fiscal variables (notably induced increase in public consumption) reinforces investment shock effect.
  - Government consumption expenditure multiplier: not statistically different from zero.
  - Tax revenue multiplier: not statistically different from zero.
- Robustness / alternative orderings and monetary policy proxy:
  - Ordering tax before spending: investment long-run multiplier 0.63 (statistically different from zero); consumption and tax not different from zero.
  - Ordering consumption first: investment multiplier 0.59.
  - Replacing local policy rate with U.S. federal funds rate: investment long-run multiplier 0.54 (statistically different from zero); consumption and tax remain not different from zero.
- Summary statement in source conclusion: government investment multiplier reported as 0.60.

### Key quantitative settings and figures (exactly as supplied)
- Quarterly sample period: 1994:Q1 to 2009:Q4.
- Maximum quarterly observations per country: about 64.
- Estimations used four lags.
- Confidence intervals obtained from 500 Monte Carlo simulations.
- Long-run multipliers evaluated at 24 periods (quarters).
- Representative multiplier values reported:
  - Government consumption (impact, bivariate): 0.20; 95 percent CI: −0.13 to 0.54; one-year cumulative ~0.42.
  - Government investment (impact, bivariate): 0.12; cumulative after four quarters: 0.37; after four years: 0.44.
  - Tax revenue (impact, bivariate): about −0.5.
  - Government investment (multivariate long-run): 0.62.
  - Alternative multivariate estimates for investment long-run: 0.63, 0.59, 0.54 (depending on ordering and monetary policy proxy).

### Policy implications and conclusions (as supplied)
- Main conclusion:
  - Only the government investment expenditure multiplier is consistently statistically different from zero across estimations; its value is positive and less than one.
- Policy recommendation:
  - Countercyclical fiscal policies in the ECCU should focus on changes in public investment to stimulate economic activity, given limited monetary and exchange-rate policy autonomy.
- Caveats:
  - Investment multiplier < 1 implies much of the impulse expands demand for imports in small open ECCU economies.
  - Government consumption and tax revenue multipliers are consistently not statistically different from zero; hence they are less reliable policy levers for stimulating output in the ECCU.

*Source: _wp13117 - 1. Fiscal multipliers based on the direct effects of fiscal shocks .................................12*

### 1. Fiscal multipliers based on the direct effects of fiscal shocks .................................12

### 1. Fiscal multipliers based on the direct effects of fiscal shocks .................................12

### Major themes
- Title as provided: "Fiscal multipliers based on the direct effects of fiscal shocks .................................12"
- No additional substantive text from the source was provided for this content unit.

### What is present in the supplied content unit
- Section heading exactly as given.
- Page indicator exactly as given: "12".

### Notes
- The supplied content only includes the section heading and page marker. No findings, numerical estimates, projections, empirical results, methodology details, or policy recommendations were included in the supplied excerpt.

*Source: _wp13117 - 1. Fiscal multipliers based on the direct effects of fiscal shocks .................................12*

### 2. Fiscal multipliers taking into account interaction of variables ................................14

### Major themes
- Title as provided: "Fiscal multipliers taking into account interaction of variables ................................14"
- No additional substantive text from the source was provided for this content unit.

### What is present in the supplied content unit
- Section heading exactly as given.
- Page indicator exactly as given: "14".

### Notes
- The supplied content only includes the section heading and page marker. No findings, numerical estimates, projections, empirical results, methodology details, or policy recommendations were included in the supplied excerpt.

*Source: _wp13117 - 2. Fiscal multipliers taking into account interaction of variables ................................14*

### References .............................................................................................................

### _wp13117 - References

### I. Introduction
- Purpose: Quantitatively analyze effects of government revenue and expenditures on output dynamics in the Eastern Caribbean Economic and Monetary Union (ECCU).
- Contextual facts preserved from source:
  - In 2008–11, total central government revenues averaged 26 percent of GDP while total expenditure was on average about 30 percent of GDP.
  - Data sample period referenced for quarterly series: 1994:Q1 to 2009:Q4.
  - Number of quarterly observations available for each ECCU member is about 64, at most.

### II. The role of fiscal multipliers
- Definitions and interpretation:
  - Public spending multiplier > 1: final increase in output larger than initial increase in public spending.
  - Multiplier < 1: initial increase in aggregate demand eroded by crowding out, higher imports, or other counteracting effects.
  - Tax multiplier expected negative; tax multiplier < −1 implies collecting one unit of taxes causes a decrease in economic activity larger than one unit.
  - Tax multiplier between 0 and −1 indicates partial recovery of economic activity after tax extraction.
- Policy relevance:
  - Accurate estimates are essential to design fiscally sustainable countercyclical policies; overestimating multipliers can compromise public-finance sustainability, and underestimating tax multipliers can lead to unexpected erosion of the tax base.
- Literature note:
  - Estimation of fiscal multipliers is uncertain; see Blanchard and Leigh 2013 and referenced crisis-era intervention literature.

### III. Methodology and data
- Overall approach:
  - Estimation based on structural vector autoregression (SVAR) models using panel data for the eight ECCU members.
  - Panel-VAR chosen because ECCU members share a single currency, high government share in the economy, high trade openness, high public indebtedness, and similar exogenous shocks, allowing pooling.
  - Fiscal multipliers derived from impulse response functions; statistical significance assessed via Monte Carlo confidence intervals.
  - Three fiscal multipliers studied: shocks to tax revenue (tax), government consumption expenditure (con), and government investment expenditure (inv).
- Data construction and transformations:
  - Fiscal variables: tax revenues (total tax revenue net of taxes on international trade, and transfers and direct subsidies to households), government consumption expenditure (net of investment, interest payments, and pension contributions), and government investment expenditure (as reported).
  - Variables transformed to natural logs, seasonally adjusted, and de-trended by subtracting a quadratic time trend. Resulting cyclical series named tax, con, inv, and gdp.
  - Real GDP compiled annually; quarterly GDP interpolated using Chow and Lin (1971) method with quarterly indicator series.
  - Quarterly proxies used for interpolation: tourism stayovers by ECCU member, real compensation to employees in the United States, real U.S. GDP, U.S. index of industrial production, and banks’ claims on the private sector in each ECCU economy.
- Interpolation (Chow and Lin) essentials:
  - High-frequency proxies X used to distribute low-frequency y into high-frequency series z while enforcing consistency with annual aggregates via a distribution matrix D.
  - Practical steps: regress low-frequency y on aggregated X to estimate β (equations (9), (10), (11) in source), obtain residuals, and distribute residuals according to distribution matrix.

### IV. Methodology for estimation of fiscal multipliers
- SVAR model setup:
  - Initial bivariate VARs include two endogenous variables: (con or inv or tax) and gdp. Matrix A assumed lower triangular (recursive structure per Blanchard and Perotti 2002).
  - Control variables (Z) included: current account-to-GDP ratio, monetary policy interest rate, public debt-to-GDP ratio, real exchange rate (log deviation), and U.S. GDP (log deviation).
  - When estimating a particular multiplier, the other two fiscal variables are included as control variables.
  - All estimations included four lags; SIC suggested five lags but four lags used for comparability and because results do not differ substantially.
- Multivariate VARs:
  - Endogenous variables: tax, inv, con, and gdp; control variables as above.
  - Ordering per Blanchard and Perotti: inv → con → tax → gdp (investment contemporaneously affects consumption, taxes, and GDP; consumption affects tax and GDP contemporaneously; taxes affect GDP contemporaneously).
  - Robustness checks: alternative orderings and replacing local policy rate with U.S. federal funds rate.
- Multiplier calculations:
  - Impact multiplier (IM) at t=1: IM = (Δgdp1 / Δcon1) ÷ (CON/Y) where Δgdp1 and Δcon1 are impulse-response contemporaneous changes, and CON/Y is average ratio of government consumption to GDP in the sample (formula (13) in source).
  - Cumulative multiplier (CM): ratio of cumulative effect on GDP to cumulative change in fiscal variable, corrected by average fiscal-to-GDP ratio (formula (14) in source). Long-run multiplier operationally taken at 24 quarters because multipliers stabilize by then.
  - Confidence intervals obtained from 500 Monte Carlo simulations; reported multipliers are the simulation means.

### V. Results — direct (bivariate) effects
- Government consumption expenditure (direct/bivariate):
  - Impact multiplier at t=1: 0.20.
  - 95 percent confidence interval at impact: −0.13 to 0.54.
  - One-year cumulative multiplier: about 0.42.
  - Statistical conclusion: not statistically different from zero over projection period.
- Government investment expenditure (direct/bivariate):
  - Impact multiplier at t=1: 0.12.
  - Cumulative multiplier after four quarters: 0.37.
  - 95 percent confidence interval above zero for cumulative effects after short horizon (statistically different from zero).
  - After eight quarters multiplier stabilizes; after four years multiplier about 0.44.
  - Interpretation: positive but < 1; substantial portion of impulse fades, likely due to high import content of investment.
- Tax revenue (direct/bivariate):
  - Impact multiplier at t=1: about −0.5.
  - Subsequent periods: multiplier turns positive in impulse-response path but remains not statistically different from zero according to its confidence interval.
  - Interpretation: an unexpected shock in taxes does not have a significant lasting effect on output in bivariate estimations.

### VI. Results — multivariate (full interaction) effects
- Multivariate VAR results (24-quarter horizon, interactions included):
  - Government investment expenditure multiplier:
    - Long-run value reaches 0.62 (statistically different from zero).
    - Interpretation: interaction with other fiscal variables (notably induced increase in public consumption) reinforces investment shock effect.
  - Government consumption expenditure multiplier: not statistically different from zero.
  - Tax revenue multiplier: not statistically different from zero.
- Robustness and alternative orderings:
  - Ordering tax before spending: public investment long-run multiplier 0.63 (statistically different from zero); consumption and tax not different from zero.
  - Ordering consumption first: investment multiplier 0.59.
  - Replacing local monetary policy rate with U.S. federal funds rate: investment multiplier long-run 0.54 (statistically different from zero); consumption and tax remain not different from zero.

### VII. Key quantitative findings and technical settings (exact figures preserved)
- Sample and estimation parameters:
  - Quarterly sample period: 1994:Q1 to 2009:Q4.
  - Maximum quarterly observations per country: about 64.
  - Estimations used four lags.
  - Confidence intervals obtained from 500 Monte Carlo simulations.
  - Long-run multipliers evaluated at 24 periods (quarters).
- Representative multiplier values reported:
  - Government consumption (impact, bivariate): 0.20; 95 percent CI: −0.13 to 0.54; one-year cumulative ~0.42 (not statistically different from zero).
  - Government investment (impact, bivariate): 0.12; cumulative after four quarters: 0.37; after four years: 0.44.
  - Tax revenue (impact, bivariate): about −0.5 (statistically different from zero at impact); subsequent path not statistically different from zero.
  - Government investment (multivariate long-run): 0.62 (statistically different from zero).
  - Alternative multivariate estimates: investment long-run 0.63 (tax before spending), 0.59 (consumption first), 0.54 (using U.S. federal funds rate in place of local policy rate).
  - Summary statement in conclusion: government investment multiplier reported as 0.60 in the paper’s concluding summary.

### VIII. Policy implications and conclusions
- Main conclusion:
  - Only the government investment expenditure multiplier is consistently statistically different from zero across estimations; its value is positive and less than one.
- Policy recommendation:
  - Countercyclical fiscal policies in the ECCU should focus on changes in public investment to stimulate economic activity, given limited monetary and exchange-rate policy autonomy.
- Caveats emphasized by the source:
  - Investment multiplier < 1 implies much of the impulse expands demand for imports in small open ECCU economies.
  - Government consumption and tax revenue multipliers are consistently not statistically different from zero; hence they are less reliable policy levers for stimulating output in the ECCU.

*Source: Authors’ calculations and analysis as presented in the IMF working paper content unit.*

### References

### References

### Listed publications
- Blanchard, Olivier, and Roberto Perotti, 2002, “An Empirical Characterization of the Dynamic Effects of Changes in Government Spending,” Quarterly Journal of Economics, Vol. 17, No. 4, pp. 1329–68.
- –––––, and Daniel Leigh, 2013, “Growth Forecasts and Fiscal Multipliers,” IMF Working Paper WP/13/1 (Washington: International Monetary Fund).
- Chow, Gregory, and An-Loh Lin, 1971, “Best Linear Unbiased Interpolation, Distribution, and Extrapolation of Time Series by Related Series,” Review of Economics and Statistics, Vol. 53, pp. 372–75.
- Freedman, Charles, Michael Kumhof, Douglas Laxton, and Jaewoo Lee, 2009, “The Case for Global Fiscal Stimulus,” IMF Staff Position Note 09/03 (Washington: International Monetary Fund).
- Hall, Robert, 2009, “By How Much Does GDP Rise If the Government Buys More Output?” NBER Working Paper 15496 (Cambridge: National Bureau of Economic Research).
- Ilzetzki, Ethan, Enrique Mendoza, and Carlos Vegh, 2011, “How Big (Small?) Are Fiscal Multipliers?” IMF Working Paper 11/52 (Washington: International Monetary Fund).
- Perotti, Roberto, 2004, “Public Investment: Another (Different) Look,” unpublished, Bocconi University.
- Spilimbergo, Antonio, Steve Symansky, Olivier Blanchard, and Carlo Cotarelli, 2008, “Fiscal Policy for the Crisis,” IMF Staff Position Note 08/01 (Washington: International Monetary Fund).
- –––––, and Martin Schindler, 2009, “Fiscal Multipliers,” IMF Staff Position Note 09/11 (Washington: International Monetary Fund).

*Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2013/_wp13117.pdf*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2013/_wp13117.pdf_