## 4. FIsCAL MULTIpLIERs: HOW WILL CONsOLIdATION AFFECT LATIN AMERICA ANd THE CARIbbEAN?

## Source details

**Canonical URL:** [4. FIsCAL MULTIpLIERs: HOW WILL CONsOLIdATION AFFECT LATIN AMERICA ANd THE CARIbbEAN?](https://www.imf.org/-/media/files/publications/reo/whd/2018/may/wreo0518-chp4.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/reo/whd/2018/may/wreo0518-chp4.pdf.md)
- [Structured JSON version](/-/media/files/publications/reo/whd/2018/may/wreo0518-chp4.pdf.json)

---

### Fiscal multipliers: definition, challenges, and identification
- The fiscal multiplier measures the change in output caused by a fiscal action, measured in the same units.
- Size depends on private agents’ responses, existing distortions, and slack; in extreme circumstances multipliers "perhaps reaching levels as high as 3 or 4."
- Two main empirical challenges:
  - Omitted variable problem: fiscal revenue and spending are affected by the same factors that drive growth.
  - Reverse causality: fiscal policy often responds to economic developments, complicating causal identification.
- Identification strategies used in the chapter (within a common local projections specification):
  - Narrative method (Romer and Romer 2010; Guajardo, Leigh, and Pescatori 2014; Alesina and others 2017): contemporaneous documents used to assess motives; reduces endogeneity but depends on researcher judgment and may capture anticipated actions.
  - Forecast error approach: identifies fiscal actions using forecast errors for public expenditure; actions are, by construction, unanticipated but subject to forecast quality issues and alternative interpretations.
  - Structural vector autoregressions (SVAR) following Blanchard and Perotti (2002): widely used but imposes strong identification assumptions and may identify anticipated actions.
- Data and samples:
  - Narrative database: 14 LAC economies, 1989–2016: Argentina, Bolivia, Brazil, Chile, Colombia, Costa Rica, Dominican Republic, Ecuador, Guatemala, Jamaica, Mexico, Paraguay, Peru, and Uruguay.
  - Forecast errors sample: 19 LAC countries (adds El Salvador, Haiti, Honduras, Nicaragua, Panama, Venezuela).
  - SVAR exercise: eight countries — Brazil, Chile, Colombia, Dominican Republic, Mexico, Paraguay, Peru, and Uruguay.
- Cumulative fiscal multipliers estimated as the cumulative change in GDP per unit of fiscal action over h years (integral multiplier, following Ramey and Zubairy 2018).

### Empirical estimates and comparisons
- Meta-review of 132 recently published estimates (multiplier defined over a two-year horizon per 1 percent of GDP fiscal adjustment):
  - Average fiscal multiplier in LAC: 0.3
  - Average for other emerging market and developing economies: 0.6
  - Average for advanced economies: 0.9
  - Studies using the narrative approach yield larger multipliers (median 1.1) than VAR-based or alternative approaches (median 0.4).
- Chapter estimates (using SVAR, forecast errors, narrative) for LAC:
  - Expenditure multiplier in LAC estimated to lie between 0.5 and 1.1 after two years.
  - Narrative identification tends to yield somewhat larger multipliers; confidence intervals are wide.
  - Tax-hike multipliers (narrative) allow for possibility of larger multipliers than expenditure measures, but no compelling evidence of a systematic difference between spending-cut and tax-hike multipliers in LAC.
  - Public consumption multiplier reported as equal to −0.6 after two years (implying an expansionary effect from cutting this type of spending), but forecast-error identification for public consumption is problematic in LAC.
- Disaggregated spending multipliers:
  - Public investment multiplier reaches almost 1.5 after two years.
  - Multiplier for primary expenditures in general is about 0.5 after two years.
  - Public investment multiplier is much larger than the public consumption multiplier (public consumption appears close to zero in other estimates).

### Macroeconomic effects of a 1 percent of GDP fiscal consolidation (baseline experiment)
- Baseline: fiscal adjustment that raises the primary balance by 1 percent of GDP, implemented via any combination of expenditure and revenue measures.
- Output:
  - After two years, output in the 14-country sample contracts by an average of 0.9 percent.
  - Confidence interval: between 0.6 and 1.1 at a 90 percent level.
- Unemployment:
  - Each percentage point of GDP in fiscal consolidation raises the unemployment rate by about 0.3 percentage point after two years.
  - Comparable estimate for advanced economies: an increase of over 0.5 of a percentage point.
- Income inequality (Box 4.1):
  - No impact of fiscal consolidation on different measures of the Gini coefficient in the main results.
  - Point estimates small and not statistically significant: market Gini increases by 0.03 units after two years.
  - Focusing on expenditure cuts shows a moderate increase in inequality; disposable Gini increase larger than market Gini in years 0 and 1, with partial recovery by year 2.
- External sector and exchange rate:
  - Fiscal consolidation improves the external current account balance by approximately one-to-one (twin deficits view).
  - Real effective exchange rate depreciates by close to 3 percent, a stronger response than observed among advanced economies.
- State dependence:
  - Conditioning on the sign of the output gap one year prior to the fiscal shock shows "no significant differences in the multiplier" between slack and non-slack periods.

### Confidence effects and perceived sovereign risk
- Hypothesis: consolidation may trigger confidence effects that ease financing conditions and stimulate private demand, especially when perceived fiscal stress is high.
- Using an Institutional Investor LLC perceived sovereign risk index, sample split into low- and high-risk bins:
  - In higher perceived sovereign default risk economies, fiscal consolidation places a smaller drag on growth: multiplier of 0.6 after two years.
  - In relatively low default risk economies, the multiplier is 1.1 after two years.
  - Even in high-risk economies, consolidation "remains contractionary."
  - Suggestive evidence of confidence channels: Emerging Markets Bond Index sovereign spread falls by about 100 basis points after two years on average across narrative episodes in the 14 economies.

### Aggregate impact of planned consolidation (2016–20) and debt dynamics
- Average planned consolidation in LAC: 1.4 percent of GDP between 2016 and 2020.
- Mapped local-projection estimates imply:
  - Impact on the level of regional output between 0.7 and 1.5 percent during 2016–20.
  - If adjustment implemented exclusively through public investment cuts, short-term impact on regional output would reach 2.1 percent over the same period.
- Debt-accumulation exercise using the estimated multiplier range indicates that, for the levels of debt and the size of government observed on average in the region, the multiplier range is not large enough to trigger unfavorable debt dynamics in the short term.

### Composition of adjustment and policy implications
- Composition matters: preserving public investment mitigates short-term growth harm and accelerates debt-to-GDP stabilization.
- Four policy implications:
  - First: Fiscal consolidation in the region will hurt, placing a drag on economic growth and employment in the short term; short-term output level impact estimated between 0.7 and 1.5 percent during 2016–20.
  - Second: Policymakers can design the composition of plans to mitigate growth drag and accelerate debt stabilization; governments should favor plans that preserve public investment where possible.
  - Third: Fiscal consolidation in LAC is likely to help stabilize debt even in the short term—sustained fiscal effort over time will outweigh the one-off contractionary effect on output.
  - Fourth: Multipliers in LAC appear broadly constant over the business cycle; periods of high perceived sovereign default risk are a particularly opportune moment to undertake consolidation, as contractionary effects tend to be mitigated to some extent.

### SVAR framework, methodological notes, and key statistics
- SVAR country-by-country framework (Blanchard and Perotti 2002) uses quarterly government consumption and investment, tax revenue (minus transfers and interest payments), and GDP in real per capita terms.
- Timing and identifying assumptions include:
  - Government does not change spending as a reaction to what happens to GDP within the quarter (b1 = 0).
  - Decisions on spending are taken before those on taxation (b2 = 0).
  - Tax revenue elasticity to GDP set to 2 (Blanchard and Perotti estimate) for comparability across countries; a reasonable LAC range could be between 1 and 2.
- Data constraints limit SVAR identification to eight countries with quarterly data.
- Summary statistics and correlations (selected):
  - Narrative packages: Countries N 1439; Mean 20.2; Median 0.0; Standard Deviation 0.6; Range [20.9, 4.1]
  - Forecast errors primary expenditures: Countries N 1953; Mean 2; Median 20.1; Standard Deviation 0.1; Range 1.6 [27.2, 7.8]
  - SVAR primary expenditures: Countries N 8; Mean 1540.0; Median 0.0; Standard Deviation 1.4; Range [22.7, 4.2]
  - CAPB Change correlations with narrative packages: 0.38; with narrative spending: 0.31; with narrative tax: 0.32; with forecast errors primary expenditures: 0.13; with SVAR primary expenditures: 0.04.
- Notes:
  - Forecast-error and SVAR shock series are trimmed at the 2nd and 98th percentiles.
  - Forecast-error construction uses October WEO forecasts and first-release outcomes for public consumption (ncg) and public investment (nfig); forecast errors deflated by GDP deflator growth (ngdp_d) and scaled by ngdp; sign inverted so positive corresponds to an unexpected cut to spending.

*International Monetary Fund | April 2018 — Chapter excerpt: "4. Fiscal Multipliers: How Will Consolidation Affect Latin America and the Caribbean?"*

### 1. Latin America and the Caribbean: Gross Public Debt

### 1. Latin America and the Caribbean: Gross Public Debt

### Fiscal multipliers and empirical challenges
- The fiscal multiplier measures the change in output caused by a fiscal action, measured in the same units.
- The size of multipliers depends on many factors, including private agents’ responses, existing distortions, and slack in the economy; in extreme circumstances multipliers "perhaps reaching levels as high as 3 or 4."
- Two main empirical challenges:
  - Omitted variable problem: fiscal revenue and spending are affected by the same factors that drive growth (example: fall in commodity prices reduces revenues and hurts growth).
  - Reverse causality: fiscal policy often responds to economic developments (example: expansion of public expenditure during a recession), making it hard to isolate causal effects.
- Estimating fiscal effects requires identifying movements of fiscal variables that are exogenous to current economic developments.

### Existing evidence on multipliers in Latin America and the Caribbean (LAC)
- Review of 132 recently published estimates:
  - Average fiscal multiplier in LAC: 0.3
  - Average for other emerging market and developing economies: 0.6
  - Average for advanced economies: 0.9
- Some studies report multipliers "very close to zero" in the region, implying "fiscal consolidation imposes almost no pain."
- Theoretical considerations are mixed as to whether LAC multipliers should be smaller or larger than elsewhere (factors that dampen multipliers: potential confidence effects, high import-to-GDP ratios; factors that amplify multipliers: tight credit constraints, less flexible exchange rates).

### Identification strategies and data used in this chapter
- Three identification strategies are used within a common local projections specification:
  - Narrative method (Romer and Romer 2010; Guajardo, Leigh, and Pescatori 2014; Alesina and others 2017): contemporaneous documents used to assess motives; reduces endogeneity but depends on researcher judgment and may capture anticipated actions.
  - Forecast error approach: identifies fiscal actions using forecast errors for public expenditure; actions are, by construction, unanticipated but subject to forecast quality issues and alternative interpretations.
  - Structural vector autoregressions (SVAR) following Blanchard and Perotti (2002): widely used but imposes strong identification assumptions and may identify anticipated actions.
- Data and samples:
  - New narrative database for 14 LAC economies between 1989 and 2016: Argentina, Bolivia, Brazil, Chile, Colombia, Costa Rica, Dominican Republic, Ecuador, Guatemala, Jamaica, Mexico, Paraguay, Peru, and Uruguay.
  - Forecast errors sample includes 19 LAC countries (adds: El Salvador, Haiti, Honduras, Nicaragua, Panama, Venezuela).
  - SVAR exercise carried out for eight countries: Brazil, Chile, Colombia, Dominican Republic, Mexico, Paraguay, Peru, and Uruguay.
- The chapter estimates cumulative fiscal multipliers defined as: for each unit of fiscal action over h years, by how many units does GDP change? (definition follows Ramey and Zubairy 2018, termed the integral multiplier).

### Macroeconomic effects of a 1 percent of GDP fiscal consolidation
- Baseline experiment: a fiscal adjustment package that raises the primary balance by 1 percent of GDP, implemented via any combination of expenditure and revenue measures.
- Output:
  - After two years, output in the sample of 14 LAC economies contracts by an average of 0.9 percent.
  - Confidence interval: between 0.6 and 1.1 at a 90 percent level.
  - This range is "consistent with an estimate for a sample of 17 advanced economies" using comparable narrative consolidations.
- Unemployment:
  - Each percentage point of GDP in fiscal consolidation raises the unemployment rate by about 0.3 percentage point after two years.
  - Comparable estimate for advanced economies: an increase of over 0.5 of a percentage point.
  - The smaller response in LAC may reflect a large informal sector acting as an adjustment margin.
- Income inequality:
  - Box 4.1 finds "no impact of fiscal consolidation on different measures of the Gini coefficient."
- External sector and exchange rate:
  - Fiscal consolidation improves the external current account balance by approximately one-to-one (in line with the twin deficits view).
  - Real effective exchange rate depreciates by close to 3 percent, a stronger response than observed among advanced economies.
- State dependence (slack vs non-slack):
  - Conditioning the multiplier on the sign of the output gap one year prior to the fiscal shock shows "no significant differences in the multiplier" depending on whether the economy is in a period of slack.

### Confidence effects and perceived sovereign risk
- Hypothesis: fiscal consolidation may trigger confidence effects that ease financing conditions and stimulate private demand, especially when perceived fiscal stress is high.
- Using an index of perceived sovereign risk from Institutional Investor LLC, the sample is split into low- and high-risk bins.
- Findings:
  - In economies perceived to be subject to higher sovereign default risk, fiscal consolidation places a smaller drag on growth: multiplier of 0.6 after two years.
  - In economies with relatively low default risk, the multiplier is 1.1 after two years.
  - Even in high-risk economies where consolidation has a smaller drag, the consolidation "remains contractionary."
  - Suggestive evidence of confidence channels: Emerging Markets Bond Index sovereign spread falls by about 100 basis points after two years on average across the narrative episodes in the 14 economies.

### Composition of adjustment and policy implications (preview of approach)
- To minimize short-term harm while stabilizing debt, the composition of fiscal adjustment matters.
- Public spending multipliers are estimated using all three identification strategies (SVAR, forecast errors, narrative) where data permit.
- The chapter distinguishes adjustments by composition to assess how policy design can protect growth while tightening fiscal belts.

*International Monetary Fund | April 2018 — Chapter excerpt: "4. Fiscal Multipliers: How Will Consolidation Affect Latin America and the Caribbean?"*

### 4. FIsCAL MULTIpLIERs: HOW WILL CONsOLIdATION AFFECT LATIN AMERICA ANd THE CARIbbEAN?

### 4. FIsCAL MULTIpLIERs: HOW WILL CONsOLIdATION AFFECT LATIN AMERICA ANd THE CARIbbEAN?

### Empirical Estimates of Fiscal Multipliers in LAC
- Using information from three identification methods (SVAR, forecast errors, narrative), the expenditure multiplier in LAC is estimated to lie between 0.5 and 1.1 after two years.
- The narrative identification approach tends to yield somewhat larger multipliers than the other two methods, but confidence intervals are wide and do not allow for inference about their relative size.
- The estimated multiplier range for tax hikes (narrative approach) allows for the possibility of larger multipliers than expenditure measures, but no compelling evidence is found for a difference between spending cut and tax hike multipliers in LAC.
- The public consumption multiplier is noted as equal to –0.6 after two years (implying an expansionary effect from cutting this type of spending), but the forecast-error identification for public consumption is problematic in LAC for at least three reasons described in the source.

### Composition of Spending Matters
- When disaggregating public expenditure:
  - The multiplier for public investment reaches almost 1.5 after two years.
  - The multiplier for primary expenditures in general is about 0.5 after two years.
  - The public investment multiplier is much larger than the public consumption multiplier (which appears close to zero in other estimates).
- Implication: composition of consolidation (investment vs. primary spending vs. consumption) has major implications for growth impacts.

### Aggregate Impact of Planned Consolidation (2016–20)
- On average, LAC countries are expected to undertake fiscal consolidation amounting to 1.4 percent of GDP between 2016 and 2020.
- Using local-projection estimates mapped to the 2016–20 period, the analysis suggests:
  - The impact of fiscal consolidation on the level of regional output will be between 0.7 and 1.5 percent during 2016–20.
  - If adjustment is implemented exclusively through public investment cuts, the short-term impact on regional output would be expected to reach 2.1 percent over the same period.
- A simple debt-accumulation exercise using the estimated range of fiscal multipliers indicates that, for the levels of debt and the size of government observed on average in the region, the range of multiplier estimates is not large enough to trigger unfavorable debt dynamics in the short term.

### Four Policy Implications
- First: Fiscal consolidation in the region will hurt, placing a drag on economic growth and employment in the short term; the short-term output level impact is estimated between 0.7 and 1.5 percent during 2016–20.
- Second: Policymakers can design the composition of plans to mitigate growth drag and accelerate debt-to-GDP stabilization; governments should favor plans that preserve public investment where possible.
- Third: Fiscal consolidation in LAC is likely to help stabilize debt even in the short term—sustained fiscal effort over time will outweigh the one-off contractionary effect on output.
- Fourth: Multipliers in LAC appear broadly constant over the business cycle; periods of high perceived sovereign default risk are a particularly opportune moment to undertake consolidation, as contractionary effects tend to be mitigated to some extent.

### Fiscal Consolidation and Income Inequality (Box 4.1)
- Sample and methods:
  - Uses fiscal shocks identified by the narrative approach for 14 LAC countries.
  - Employs local projections to estimate responses of market and post-tax-and-transfer (disposable) income distributions.
  - Inequality measured by market and disposable income Gini from SWIID 6.1; estimation period 1989–2016.
- Main findings:
  - Fiscal consolidations have very little effects on income inequality in LAC. Point estimates are positive but very small—the market Gini increases by 0.03 units after two years—and are not statistically significant.
  - Focusing on disposable income does not materially change results; the Gini coefficient is relatively insensitive to fiscal consolidation shocks.
  - Focusing on expenditure cuts: a moderate increase in income inequality is observed. The market Gini increase is more persistent; the disposable Gini increase is larger than the market Gini in years 0 and 1, implying a decrease in fiscal redistribution—suggesting expenditure cuts might have worsened inequality by decreasing transfers. By year 2, fiscal redistribution recovers somewhat and partially offsets the increase in market Gini.
- Overall: Fiscal consolidation shocks have very little effects on disposable income in the sample; results contrast with stronger effects found for advanced economies but align with subnational evidence for Brazil.

### Existing Empirical Evidence (Box 4.2)
- Meta-summary of 132 recent estimates (multiplier defined as change in GDP over a two-year horizon per 1 percent of GDP fiscal adjustment):
  - Estimates for LAC typically smaller than in other regions—averaging less than 0.3—with a high proportion concentrated just above zero.
  - Estimates for other EMDEs average 0.6; for advanced economies average 0.9.
  - Methodological note: studies using the narrative approach yield larger multipliers (median 1.1) than VAR-based or alternative approaches (median 0.4), indicating methodological choices affect estimated magnitudes.

### Technical Approach and Identification
- Impulse responses estimated with the local projections method of Jordà (2005) in a panel setting; for real GDP the estimated parameter β̂h is interpreted as the cumulative fiscal multiplier at horizon h.
- Three identification methods for fiscal shocks:
  - Structural VAR (country-by-country Blanchard-Perotti (2002) SVAR models).
  - Narrative approach: contemporaneous policy documents used to assess motivation, size, and timing; includes only actions motivated by reducing an inherited deficit/high debt or long-run considerations unrelated to the cycle; measures rely on contemporaneous estimates of revenue/expenditure impact and exclude unimplemented announcements.
  - Forecast-error approach: uses October WEO forecasts and first-release outcomes to construct forecast errors for public consumption (series code ncg) and public investment (series code nfig); forecast errors deflated by GDP deflator growth (ngdp_d), multiplied by average ratio of nominal spending to nominal GDP (ngdp); sign inverted so positive corresponds to an unexpected cut to spending; primary expenditure forecast errors constructed as sum of consumption and investment forecast errors.
- In SVAR and forecast-error shock series, variables are truncated at the 2nd and 98th percentiles; for identification using forecast errors, concerns about endogeneity and inflation surprises are acknowledged and mitigated as described in the source.

*International Monetary Fund | April 2018*

### Chapter 4 of the October 2017 World Economic

### Chapter 4 of the October 2017 World Economic

### Identification Using Country-by-Country SVAR Models
- Framework:
  - Start from reduced-form vector autoregression Xt = A(L, q) Xt−1 + et, where Xt = [St, Tt, Yt]' includes the logarithm of quarterly spending (government consumption and investment), tax revenue (minus transfers and interest payments), and GDP, respectively, in real per capita terms, and et is the vector of estimated residuals.
  - Structural decomposition assumed: et = linear function of structural shocks ut plus contemporaneous terms:
    - et_s = b1 et_y + b2 ut_T + ut_s (A4.1.2)
    - et_T = a1 et_y + a2 ut_s + ut_T (A4.1.3)
    - et_y = c2 et_s + c1 et_T + ut_y (A4.1.4)
- Timing and identifying assumptions (following Blanchard and Perotti 2002):
  - Government does not change spending as a reaction to what happens to GDP within the quarter, so b1 = 0.
  - Decisions on spending are taken before those on taxation, thus b2 = 0.
  - a1 — the effect of GDP surprises on tax revenues — is estimated outside the system in Blanchard and Perotti; their practice often assumes the elasticity of tax revenues to GDP is 1 or slightly larger. For LAC, a reasonable level could be between 1 and 2 because economic growth is associated with formalization.
  - For comparability, the Blanchard and Perotti estimate of 2 is used for all countries in this analysis.
  - c1 and c2 can be obtained within the SVAR; in some cases one coefficient was taken from an instrumental-variable estimation when statistically significant.
- Data and sample constraints:
  - Quarterly data on fiscal variables and real output were required, restricting LAC sample to eight countries: Brazil, Chile, Colombia, Dominican Republic, Mexico, Paraguay, Peru, and Uruguay.
  - Variables used:
    - Government revenue net of interest payments and part of the subsidies and transfers (subtracted in the spirit of Blanchard and Perotti 2002).
    - Government spending, including expenditures on wages and goods and services plus investment and the remaining part of the transfers.
    - Country GDP.
  - Exogenous controls: terms-of-trade index and trade-weighted foreign partners’ GDP to control for commodity and foreign demand effects on government revenues and spending.

### Description of Fiscal Shocks across Identification Strategies
- Identification strategies compared:
  - Narrative approach (action-based narrative shocks)
  - Forecast-error approach (forecast errors)
  - Country-by-country SVARs
- General observation:
  - Narrative shocks display a smaller range and variability than those identified using SVAR and forecast-error approaches.
  - Only narrative shocks have a significant correlation with the change in the cyclically adjusted primary balance (CAPB). Forecast-error and SVAR shocks have very low correlations with all other shocks, indicating different concepts are captured by alternative identification strategies and that these are not always closely related to the overall change in the fiscal balance.

### Key statistics and comparison (Annex Table 4.1)
1. Summary statistics (as reported)
- Narrative
  - Packages: Countries N 1439; Mean 20.2; Median 0.0; Standard Deviation 0.6; Range [20.9, 4.1]
  - Spending measures: Countries N 1439; Mean 20.1; Median 0.0; Standard Deviation 0.2; Range [20.5, 2.0]
  - Tax measures: Countries N 1439; Mean 20.1; Median 0.0; Standard Deviation 0.4; Range [20.9, 4.1]
- Forecast errors
  - Primary expenditures: Countries N 1953; Mean 2; Median 20.1; Standard Deviation 0.1; Range 1.6 [27.2, 7.8]
  - Public consumption: Countries N 1957; Mean 0.0; Median 0.0; Standard Deviation 0.9; Range [24.1, 4.1]
  - Public investment: Countries N 1967; Mean 2; Median 20.1; Standard Deviation 0.2; Range 1.2 [24.7, 5.5]
- SVAR
  - Primary expenditures: Countries N 8; Mean 1540.0; Median 0.0; Standard Deviation 1.4; Range [22.7, 4.2]

2. Pairwise correlations (selected contemporaneous correlations reported)
- Narrative correlations:
  - Packages with Packages: 1.00
  - Packages with Spending: 0.68
  - Packages with Tax: 0.91
  - Spending with Tax: 0.33
- Forecast errors correlations:
  - Primary expenditures with Primary expenditures: 0.11
  - Primary expenditures with Consumption: 0.12
  - Primary expenditures with Investment: 0.07
  - Consumption with Investment: 0.14
  - Consumption with Primary expenditures: 0.05
  - Investment with Primary expenditures: 0.10
  - Forecast errors block internal correlation summary: Consumption with Primary expenditures 0.56 (table context)
- SVAR correlations:
  - Primary expenditures (SVAR) with Primary expenditures (SVAR): 0.04
  - SVAR block correlations include numbers: 0.07, 0.00, −0.03, −0.16, 0.09, 1.00 (as presented in the table)
- CAPB (Change) correlations with others:
  - CAPB Change with Narrative Packages: 0.38
  - CAPB Change with Narrative Spending: 0.31
  - CAPB Change with Narrative Tax: 0.32
  - CAPB Change with Forecast errors Primary expenditures: 0.13
  - CAPB Change with Forecast errors Consumption: 0.04
  - CAPB Change with Forecast errors Investment: 0.15
  - CAPB Change with SVAR Primary expenditures: 0.04
  - CAPB Change with CAPB Change: 1.00

- Notes regarding table construction:
  - In the case of forecast errors and SVAR shocks, samples have been trimmed at the 2nd and 98th percentiles.
  - CAPB = cyclically adjusted primary balance; SVAR = structural vector autoregression.

### Methodological implications and interpretation
- Elasticity choice:
  - The analysis uses an elasticity of 2 (Blanchard and Perotti estimate) for tax response to GDP for all countries for comparability, though a reasonable LAC range could be between 1 and 2.
  - Identification of shocks is reported as not very sensitive to the size of this constant.
- Instrumental-variable estimation:
  - Blanchard and Perotti estimated c1 and c2 outside the SVAR using an instrument (cyclically adjusted, reduced-form tax residual ert = et_T − a1 et_y) and only estimated a2 inside the SVAR; the annex notes c1 and c2 can also be estimated within the SVAR and were in this work, with occasional use of IV estimates when statistically significant.
- Data limitations:
  - Requirement of quarterly fiscal and output data restricts cross-country SVAR identification to the eight LAC countries listed above.

*Source: IMF staff calculations, Chapter 4 Annex Table 4.1 and accompanying text.*

---


_Source: https://www.imf.org/-/media/files/publications/reo/whd/2018/may/wreo0518-chp4.pdf_