## wpiea2020012-print-pdf

## Source details

**Canonical URL:** [wpiea2020012-print-pdf](https://www.imf.org/-/media/files/publications/wp/2020/english/wpiea2020012-print-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2020/english/wpiea2020012-print-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2020/english/wpiea2020012-print-pdf.pdf.json)

---

### Introduction and research questions
- Purpose: examine the output effects of fiscal spending shocks in low income countries (LICs) and explore determinants of fiscal multipliers in LICs.
- Key empirical questions:
  - Are the output effects of fiscal spending shocks in LICs different from those in AEs or EMs?
  - What factors affect the size of the output effects in LICs?

### Main findings (baseline)
- Output effects of government spending shocks in LICs are notably smaller than in AEs and marginally smaller than in EMs.
- Quantitative result: a positive government spending shock of one percentage point of GDP increases output by 0.1 percent in the same year; its impact disappears three years after the shock.
- On average, fiscal multipliers in LICs are less than half of those in AEs.
- Statistical significance:
  - Differences between AEs and LICs are statistically significant in the medium term (years two and three after the shock).
  - Differences between EMs and LICs are not statistically significant.

### Heterogeneity and determinants of multiplier size in LICs
- Multipliers are larger in LICs in the following instances:
  - during recessions;
  - under fixed exchange rate regimes;
  - with higher quality of institutions.
- Among these factors, the cyclical state of the economy is the most significant determinant.
- Under floating exchange rate regimes, positive output effects are not found even during recessions.

### Stylized facts across income groups
- Elements examined as relevant for multiplier size: (i) exchange rate regime; (ii) trade openness; (iii) public debt level; (iv) labor market rigidity; (v) size of automatic stabilizers; and (vi) institutional quality.
- Observed tendencies for LICs (relative to AEs and EMs):
  - slightly lower trade openness;
  - less flexible exchange rate regimes;
  - smaller automatic stabilizers;
  - lower quality of institutions;
  - median debt-to-GDP in LICs smaller than AEs but difference not large given variance;
  - no significant difference in labor market rigidity.

### Empirical methodology and data
- Identification of fiscal shocks: forecast errors of government spending using vintage data from the IMF’s World Economic Outlook (WEO) October forecasts; fiscal spending shocks constructed as actual government spending minus forecast (government spending measured as G/Y).
- Estimation method: local projection method (Jordà, 2005) following Auerbach and Gorodnichenko (2012, 2013).
- Baseline regression: estimated for horizons h = 0, ... , 3; relates future log real GDP changes to identified fiscal shock and controls with country and time fixed effects; impulse responses computed from estimated coefficients.
- Data:
  - Annual data for 1995–2017 from IMF WEO.
  - Sample focused on IMF-defined LICs, excluding resource rich countries.
  - Shocks from public investment and government consumption; shock variables trimmed at 1 percent; require observations to have both government investment and government consumption shock variables.
  - Resulting sample: 566 observations for 42 countries.
- Identification caveat: approach aims to reduce "fiscal foresight" problems and feedback from contemporaneous business cycle by using forecast errors; identification of fiscal shocks remains challenging.

### Output effects across income groups
- Temporal shape:
  - Output effects in EMs and LICs flatten a year after the shock, while in AEs the effects continue to rise over three years.
  - In AEs persistence likely reflects secondary (spillover) effects from fiscal spending to private sector activity; in less developed countries this spillover is weaker, making output effects in LICs short-lived.
- Statistical differences:
  - Differences between AEs and LICs are statistically significant at 5 percent level after t=2; differences for AEs and EMs are not statistically significant at any horizon.

### Role of the private sector
- Estimation of spending shocks on private consumption and investment shows:
  - Positive fiscal spending shocks foster private consumption and investment in AEs.
  - This fostering effect is not observed in EMs and LICs.
- Conclusion: different behavioral responses by the private sector likely explain smaller output effects in LICs.

### Fiscal policy instruments: government consumption vs public investment
- Decomposition findings:
  - Public investment shocks have a larger effect than government consumption shocks.
  - Output effects of public investment shocks are statistically significant over the horizon.
  - Government consumption shocks increase output only on impact; in LICs the output effect of public consumption shocks is relatively short-lived compared to AEs.
- Policy implication:
  - Cutting government consumption may adversely affect growth at the outset but would not necessarily drag down growth over the medium term.
  - Cutting capital spending produces adverse growth effects that build up over time.
  - Fiscal stimulus centered on government consumption yields only temporary output gains and can be wasteful if intended to stimulate growth.

### Determinants of the fiscal multiplier examined
- Determinants considered: business cycles, trade openness, exchange rate regimes, the level of public debt, labor market rigidities, the size of automatic stabilizers, the quality of institutions.
- Non-linear model specification:
  - Transition function G(z_it) = exp(−δ z_it) / (1 + exp(−δ z_it)), δ>0.
  - When z is the business cycle indicator, δ = 1.5; otherwise δ = 1.0.
  - Business cycle indicator: real GDP growth (normalized to zero mean and unit variance).
  - Exchange rate regimes: dummy for fixed vs floating using Ilzetzki et al. classification.

### Key empirical findings on determinants (LICs)
- Significant determinants: business cycle, exchange rate regimes, and quality of institutions.
- Business cycle state:
  - Output effects in LICs are highly state-dependent: larger during recessions than during booms.
  - During booms, positive spending shocks produce negative but generally not statistically significant effects.
  - During recessions, spending shocks produce positive, visible, and sizeable effects, statistically significant over the three-year estimation period.
- Government consumption vs public investment during cycles:
  - During booms: no statistically significant impact on growth for either spending type.
  - During recessions: both consumption and investment shocks have positive and statistically significant effects; public consumption effects are slightly lower and short-lived.
  - Policy implication: in recessions, cutting government consumption (rather than capital spending) would be less harmful for growth, particularly over the medium term.
- Exchange rate regimes:
  - Output effects tend to be larger under fixed exchange rate regimes; positive government spending shocks increase output under fixed regimes but do not significantly affect output under flexible regimes.
  - Even during recessions and for different spending types, fiscal spending shocks do not have a statistically significant output effect under floating exchange rate regimes.
- Quality of institutions:
  - Economies with higher institutional quality achieve a positive output impact of government spending shocks that remains positive until two years after the shock.
  - This impact is not observed in economies with lower institutional quality.
  - Robustness: similar results using the Public Investment Management Index.

### Factors not found significant in LICs
- Trade openness, public debt level, labor market rigidity, and the size of automatic stabilizers were not identified as significant determinants.
- Explanations and measurement caveats:
  - Trade openness: inconclusive effects; inefficient spending may not raise imports.
  - Public debt level: liquidity constraints can make fiscal stimulus raise consumption immediately, offsetting negative credibility effects.
  - Labor market rigidity: limited data quality and large informal sectors reduce measurement reliability.
  - Size of automatic stabilizers: measurement error using ratio of public spending to nominal GDP and potentially smaller stabilizers in developing countries.

### Dominance analysis: relative importance of determinants
- Method: dominance analysis (Budescu, 1993) applied to model including dummies for determinants above threshold (threshold = median for business cycle and institutional quality; exchange rate dummy as defined).
- Findings:
  - State of the business cycle has the largest impact on the size of fiscal multipliers in LICs.
  - On impact, the relative importance of the business cycle exceeds 80 percent; it remains dominant three years after the shock, though the relative importance of exchange rate regimes increases somewhat.
- Implication: output effects are notably lower during booms, under floating exchange rate regimes, and with weaker institutions; adverse output effects are larger during recessions, under fixed regimes, and with better institutions. The large role of the business cycle confirms the importance of countercyclical fiscal policy.

### Box: Initial capital stock and public investment shocks in LICs
- Question: Does initial capital stock affect output effect of public investment shocks?
- Measure: average of capital stock to GDP in t-1 to t-3 as initial capital stock; sample period 1995–2017.
- Result for LICs:
  - Output effects of public investment shocks are positive and significantly different from zero with higher initial capital stock levels.
  - Output effects are negative and insignificant with lower initial capital stock levels.
- Interpretation: when initial capital stock is too low, the private sector may not respond to fiscal policy shocks; heterogeneity exists and further investigation is needed.
- Data notes: capital stock comes from IMF–FAD (general government capital stock); robustness checks with alternative averaging and physical capital stock index deliver similar outcomes.

### Robustness checks
- Local projection method noted as more robust to misspecification.
- Alternative shock measures: forecasts from October of the previous year and April of the same year tested; results "remain broadly unchanged with the alternative shock series."
- Additional controls: current and lagged output growth innovations added; main results remain robust.
- Sign of shocks: introducing dummy for positive shocks shows "no statistically significant differences between positive and negative shocks."
- Controls and samples: different control combinations, lag lengths, and re-estimating using longest available sample period per country yield results "broadly the same."
- Business cycle measure: using output gap instead of growth rate leaves main results "unchanged."

### Main empirical conclusions
- Cross-group comparison:
  - "The output effects of government spending shocks in LICs are notably smaller than those in AEs and slightly smaller than those in EMs."
  - "On average, the fiscal multipliers in LICs are less than half of those in AEs."
- Fiscal instrument heterogeneity:
  - Government consumption shocks: effects are temporary.
  - Public investment shocks: "have a larger and long-lasting effect."
- State dependence and structural factors:
  - Larger output effects found (i) during recessions; (ii) under fixed exchange rate regimes; (iii) with higher institution quality.
  - Under floating exchange rate regimes, "positive output effects are not found even during recessions."
  - The state of the economy is most important in determining the size of the output effects.

### Policy implications and recommendations
- When estimating output effects in LICs, "the state of the economy and the country’s structural characteristics, such as exchange rate regimes and quality of institutions, should be considered, preferably with different weights."
- Because output effects of fiscal spending are low in LICs, "the output costs of fiscal adjustment appear not very high."
  - Output costs of fiscal adjustment in LICs "may not be as large as previously thought," especially if adjustments are:
    - adopted outside of a recession;
    - under a floating exchange rate regime;
    - based on cutting public consumption;
    - and/or accompanied by reform to enhance institutions.
- Countercyclical policy: "State dependent output effects" imply that undertaking countercyclical fiscal policy is essential.
- Exchange rate regime: under floating exchange rate regimes, "the output effects of fiscal spending appear non-existent."
- Institutional quality: LICs "should enhance institutional quality to promote growth stimulus of fiscal spending."

*Source: wpiea2020012-print-pdf*

### REFERENCES ___________________________________________________________________________________ 22

### wpiea2020012-print-pdf - REFERENCES ___________________________________________________________________________________ 22

### Introduction and research questions
- Purpose: examine the output effects of fiscal spending shocks in low income countries (LICs) and explore determinants of fiscal multipliers in LICs.
- Key empirical questions:
  - Are the output effects of fiscal spending shocks in LICs different from those in AEs or EMs?
  - What factors affect the size of the output effects in LICs?

### Main findings (baseline)
- Output effects of government spending shocks in LICs are notably smaller than in AEs and marginally smaller than in EMs.
- Quantitative result: a positive government spending shock of one percentage point of GDP increases output by 0.1 percent in the same year; its impact disappears three years after the shock.
- On average, fiscal multipliers in LICs are less than half of those in AEs.
- Statistical significance:
  - Differences between AEs and LICs are statistically significant in the medium term (years two and three after the shock).
  - Differences between EMs and LICs are not statistically significant.

### Heterogeneity and determinants of multiplier size in LICs
- Multipliers are larger in LICs in the following instances:
  - during recessions;
  - under fixed exchange rate regimes;
  - with higher quality of institutions.
- Among these factors, the cyclical state of the economy is the most significant determinant.
- Under floating exchange rate regimes, positive output effects are not found even during recessions.

### Stylized facts across income groups
- Six elements examined as relevant for multiplier size: (i) exchange rate regime; (ii) trade openness; (iii) public debt level; (iv) labor market rigidity; (v) size of automatic stabilizers; and (vi) institutional quality.
- Observed tendencies for LICs (relative to AEs and EMs):
  - slightly lower trade openness;
  - less flexible exchange rate regimes;
  - smaller automatic stabilizers;
  - lower quality of institutions;
  - median debt-to-GDP in LICs smaller than AEs but difference not large given variance;
  - no significant difference in labor market rigidity.

### Empirical methodology and data
- Identification of fiscal shocks: forecast errors of government spending using vintage data from the IMF’s World Economic Outlook (WEO) October forecasts; fiscal spending shocks constructed as actual government spending minus forecast (government spending measured as G/Y).
- Estimation method: local projection method (Jordà, 2005) following Auerbach and Gorodnichenko (2012, 2013).
- Baseline regression (estimated for horizons h = 0, ... , 3) relates future log real GDP changes to identified fiscal shock and controls with country and time fixed effects; impulse responses computed from estimated coefficients.
- Data:
  - Annual data for 1995–2017 from IMF WEO.
  - Sample focused on IMF-defined LICs, excluding resource rich countries.
  - Construction details: shocks from public investment and government consumption; shock variables trimmed at 1 percent; require observations to have both government investment and government consumption shock variables.
  - Resulting sample: 566 observations for 42 countries.
- Note on identification: approach aims to reduce "fiscal foresight" problems and feedback from contemporaneous business cycle by using forecast errors; caveat remains that identification of fiscal shocks is challenging.

### Supporting empirical context and comparisons
- Literature context:
  - Previous evidence generally finds larger multipliers in downturns and identifies determinants such as exchange rate regime, trade openness, public debt, labor market rigidity, size of automatic stabilizers, and institutional quality.
  - Prior findings for developing countries: short-term fiscal multiplier of 0.2 for developing countries in some studies; estimates for sub-Saharan Africa tend to be smaller than EMs and AEs.
- Sample sizes reported for comparative figures:
  - AEs: sample size 507 (17 countries).
  - EMs: sample size 742 (55 countries).
  - RR countries excluded; AEs data from OECD Economic Outlook.

*Source: wpiea2020012-print-pdf*

### 20. Notably, the shape of the output effects differs significantly among the income

### wpiea2020012-print-pdf - 20. Notably, the shape of the output effects differs significantly among the income

### Output effects across income groups
- Output effects in EMs and LICs flatten a year after the shock, while in AEs the effects continue to rise over three years.
- In AEs the persistence likely reflects secondary (spillover) effects from fiscal spending to private sector activity; in less developed countries this spillover is weaker, making output effects in LICs short-lived.
- Differences between AEs and LICs: the differences of output effects between AEs and LICs are statistically significant at 5 percent level after t=2; differences for AEs and EMs are not statistically significant at any horizon.

### Role of the private sector
- Estimation of spending shocks on private consumption and investment (Annex I) shows:
  - Positive fiscal spending shocks foster private consumption and investment in AEs.
  - This fostering effect is not observed in EMs and LICs.
- Conclusion: different behavioral responses by the private sector likely explain smaller output effects in LICs.

### Fiscal policy instruments: government consumption vs public investment
- Government spending shocks decomposed into consumption and investment shocks:
  - Public investment shocks have a larger effect than government consumption shocks (consistent with Ilzetzki et al., 2013).
  - Output effects of public investment shocks are statistically significant over the horizon.
  - Government consumption shocks increase output only on impact; in LICs the output effect of public consumption shocks is relatively short-lived compared to AEs.
- Note: responses are to an unanticipated shock to government spending of 1 percentage point of GDP; confidence bands are 90 percent.

### Policy implications for fiscal consolidation and stimulus
- Cutting government consumption may adversely affect growth at the outset but would not necessarily drag down growth over the medium term.
- Cutting capital spending would produce adverse growth effects that build up over time, reflecting a less productive capacity.
- Fiscal stimulus centered on government consumption yields only temporary output gains and can be wasteful if intended to stimulate growth.

### Determinants of the fiscal multiplier examined
- Determinants considered (as discussed in IMF (2014)):
  - Business cycles,
  - Trade openness,
  - Exchange rate regimes,
  - The level of public debt,
  - Labor market rigidities,
  - The size of automatic stabilizers,
  - The quality of institutions.
- Non-linear model specification:
  - Transition function G(z_it) = exp(−δ z_it) / (1 + exp(−δ z_it)), δ>0.
  - When z is the business cycle indicator, δ = 1.5; otherwise δ = 1.0.
  - Business cycle indicator: real GDP growth (normalized to zero mean and unit variance).
  - Exchange rate regimes: dummy variable for fixed vs floating using Ilzetzki et al. classification.

### Key empirical findings on determinants (LICs)
- Significant determinants: business cycle, exchange rate regimes, and quality of institutions.
- Business cycle:
  - Output effects in LICs are highly state-dependent: larger during recessions than during booms.
  - During booms, positive spending shocks produce negative but generally not statistically significant effects.
  - During recessions, spending shocks produce positive, visible, and sizeable effects, statistically significant over the three-year estimation period.
  - Low effects in booms likely reflect crowding out of private investment.
- Government consumption and capital spending shocks:
  - During booms: no statistically significant impact on growth.
  - During recessions: both consumption and investment shocks have positive and statistically significant effects, but public consumption effects are slightly lower and short-lived.
  - Policy implication: if fiscal adjustment is needed during recessions, cutting government consumption (rather than capital spending) would be less harmful for growth, particularly over the medium term.
- Exchange rate regimes:
  - Output effects tend to be larger under fixed exchange rate regimes; positive government spending shocks increase output under fixed regimes but do not significantly affect output under flexible regimes.
  - Even during recessions and for different spending types (public investment and government consumption), fiscal spending shocks do not have a statistically significant output effect under floating exchange rate regimes (Annex III).
- Quality of institutions:
  - Economies with higher institutional quality achieve a positive output impact of government spending shocks that remains positive until two years after the shock.
  - Such an impact is not observed in economies with lower institutional quality.
  - Robustness: similar results using the Public Investment Management Index (Dabla-Norris et al., 2012).

### Factors not found significant in LICs
- Trade openness, public debt level, labor market rigidity, and the size of automatic stabilizers were not identified as significant determinants (Annex IV).
- Explanations and measurement caveats offered:
  - Trade openness: inconclusive effects; inefficient spending may not raise imports.
  - Public debt level: liquidity constraints can make fiscal stimulus raise consumption immediately, offsetting negative credibility effects.
  - Labor market rigidity: limited data quality and large informal sectors reduce measurement reliability.
  - Size of automatic stabilizers: measurement error using ratio of public spending to nominal GDP and potentially smaller stabilizers in developing countries.

### Dominance analysis: relative importance of determinants
- Method: dominance analysis (Budescu, 1993) applied to model including dummies for determinants above threshold (threshold = median for business cycle and institutional quality; exchange rate dummy as defined).
- Findings:
  - State of the business cycle has the largest impact on the size of fiscal multipliers in LICs.
  - On impact, the relative importance of the business cycle exceeds 80 percent; it remains dominant three years after the shock, though the relative importance of exchange rate regimes increases somewhat.
- Implication: output effects are notably lower during booms, under floating exchange rate regimes, and with weaker institutions; adverse output effects are larger during recessions, under fixed regimes, and with better institutions. The large role of the business cycle confirms the importance of countercyclical fiscal policy.

### Box: Initial capital stock and public investment shocks in LICs
- Question: Does initial capital stock affect output effect of public investment shocks?
- Measure: average of capital stock to GDP in t-1 to t-3 as initial capital stock; sample period 1995–2017.
- Result for LICs (contrary to Izquierdo et al. findings for AEs/EMs):
  - Output effects of public investment shocks are positive and significantly different from zero with higher initial capital stock levels.
  - Output effects are negative and insignificant with lower initial capital stock levels.
- Interpretation: in LICs, when initial capital stock is too low, the private sector may not respond to fiscal policy shocks; heterogeneity exists and further investigation is needed.
- Data notes: capital stock comes from IMF–FAD (general government capital stock); robustness checks with alternative averaging and physical capital stock index deliver similar outcomes.

*Source: Authors’ calculations (from wpiea2020012-print-pdf).*

### 37. These results are robust in several directions. Although the local projection method

### wpiea2020012-print-pdf - 37. These results are robust in several directions. Although the local projection method

### Robustness checks
- Methodological note: The local projection method is noted as more robust to misspecification (Jordà, 2005).
- Government spending shocks:
  - Alternative shock measures tested: forecasts from October of the previous year and April of the same year used to compute forecast errors instead of forecasts made in October of the same year.
  - Results: "Our results remain broadly unchanged with the alternative shock series."
  - Additional controls: current and lagged output growth innovations added to address endogeneity concerns; main results remain robust.
  - Sign of shocks: introducing a dummy = 1 for positive shocks and 0 otherwise shows "no statistically significant differences between positive and negative shocks."
- Control variables:
  - Different combinations and lag lengths considered (examples: public sector debt, inflation, fiscal variables, revenue shocks, and monetary policy measure).
  - Main results "broadly remain unchanged" when using these variables.
- Sample periods:
  - Model re-estimated using the longest available sample period for each country instead of starting in 1995.
  - Results are "broadly the same."
- Business cycle measure:
  - Tested slack measured by output gap instead of growth rate.
  - Main results "remain unchanged."
- Footnote: "The results of these robustness checks are available upon request." (22)

### Main empirical conclusions
- Cross-group comparison:
  - "The output effects of government spending shocks in LICs are notably smaller than those in AEs and slightly smaller than those in EMs."
  - "On average, the fiscal multipliers in LICs are less than half of those in AEs."
- Fiscal instrument heterogeneity:
  - Government consumption shocks: effects are temporary.
  - Public investment (capital spending) shocks: "have a larger and long-lasting effect."
- State dependence and structural factors:
  - Larger output effects of fiscal policy shocks are found:
    - (i) during recessions;
    - (ii) under fixed change rate regimes;
    - (iii) with higher institution quality.
  - Under floating exchange rate regimes, "positive output effects are not found even during recessions."
  - Of the three factors, "the state of the economy is most important in determining the size of the output effects."

### Policy implications and recommendations
- Targeting and estimation:
  - When estimating output effects in LICs, "the state of the economy and the country’s structural characteristics, such as exchange rate regimes and quality of institutions, should be considered, preferably with different weights."
- Fiscal adjustment costs:
  - Because output effects of fiscal spending are low in LICs, "the output costs of fiscal adjustment appear not very high."
  - Implication: output costs of fiscal adjustment in LICs "may not be as large as previously thought," especially if adjustments are:
    - adopted outside of a recession;
    - under a floating exchange rate regime;
    - based on cutting public consumption;
    - and/or accompanied by reform to enhance institutions.
- Countercyclical policy and exchange rate regime:
  - "State dependent output effects" imply that undertaking countercyclical fiscal policy is essential.
  - Under floating exchange rate regimes, "the output effects of fiscal spending appear non-existent."
- Institutional quality:
  - LICs "should enhance institutional quality to promote growth stimulus of fiscal spending."

### Annex notes and figure interpretation
- Annex I–IV figures:
  - Note repeated across annexes: "t=0 is the year of the shock. Solid lines present the responses (in percent) to an unanticipated shock to government spending of 1 percentage point of GDP. Dashed lines denote 90 percent confidence bands."
  - Figures compare responses for AEs, EMs, LICs; separate panels examine booms vs recessions, floating exchange rate regimes, and possible determinants of the size of the fiscal multiplier (examples listed in figures: Trade openness, Public debt level, Labor market rigidity, Automatic stabilizer).
- Visual scale markers shown in annexes include numeric tick labels such as "-0.4", "0", "0.4", "0.8", "1.2", "1.6", "-2", "-1", "1", "2", "-3" etc., and "Years after the shock" with axis range markers like "-1 0 1 2 3" indicating multi-year impulse responses.

*Source: wpiea2020012-print-pdf (internal content provided).*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2020/english/wpiea2020012-print-pdf.pdf_