## wp1857

## Source details

**Canonical URL:** [wp1857](https://www.imf.org/-/media/files/publications/wp/2018/wp1857.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2018/wp1857.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2018/wp1857.pdf.json)

---

### Introduction and research question
- Focus: effect of government spending shocks on income distribution in emerging and developing economies (EMDEs).
- Sample and scope:
  - Panel of 103 EMDEs.
  - Period covered: 1990-2016.
- Primary empirical question: What is the effect of government spending shocks on income distribution?
- Motivations:
  - Inequality has risen in many EMDEs and remains high in regions such as Latin America and Sub-Saharan Africa.
  - Many EMDEs have recently experienced rising public debt-to-GDP ratios.
  - Policy trade-off: addressing high/rising inequality while maintaining/regaining fiscal sustainability.

### Methodology and identification
- Baseline identification:
  - Adopted Auerbach and Gorodnichenko (2013a, 2013b) (AG) approach: identify government spending shocks as forecast errors in government spending to align econometrician’s information set with that of economic agents and mitigate “fiscal foresight” bias.
  - Fiscal shocks (FE) constructed as forecast errors using IMF WEO forecasts made in October of the same year:
    - FE_{i,t} = ∆lnG_{i,t} − ∆lnG^E_{i,t}
    - FE_{i,t} = (ln G_{i,t}^{apr,2017} − ln G_{i,t−1}^{apr,2017}) − (ln G_{i,t}^{oct,t} − ln G_{i,t−1}^{oct,t})
  - Justification for October forecasts: minimizes likelihood that unanticipated changes in government spending reflect endogenous responses to contemporaneous economic developments.
- Estimation technique:
  - Local projections approach of Jordà (1995) to trace short- and medium-run responses of inequality measures to identified fiscal shocks.
  - Baseline regression (for each k = 0,1,...5):
    - y_{i,t+k} = α_{i}^{k} + θ_{t}^{k} + β^{k} FE_{i,t} + θ^{k} X_{i,t} + ε_{i,t}^{k}
    - y is the Gini coefficient; X_{i,t} includes controls, two lags of the shocks, and two lags of the Gini coefficient.
- Inequality multiplier computation:
  - Inequality multiplier at horizon h: (∑_{k=0}^{h} y_{t+k}) / (∑_{k=0}^{h} g_{t+k})
  - Three-step approach (Ramey and Zubairy (2018)): estimate impulse responses for inequality and for government spending-to-GDP ratio, then compute ratio.

### Data and inequality measures
- Inequality data:
  - Source: Standardized World Income Inequality Database (SWIID 5.1).
  - Measures available: net (post-tax, post-transfers) and market (pre-tax, pre-transfers) Gini indices.
  - SWIID income definition: sum of monetary and non-monetary income from labor, monetary income from capital, monetary social security transfers (including work-related insurance transfers, universal transfers, and assistance transfers), non-monetary social assistance transfers, monetary and non-monetary private transfers, less income taxes and social contributions.
- Distributional properties of shocks (EMDEs):
  - Total government spending shocks (1st to 99th percentile): between -41 and 45 percent.
  - Government consumption shocks: between -42 and 49 percent.
  - Government investment shocks: between -101 and 95 percent.
  - These shocks translate into approximately symmetric distribution of changes in government expenditures, with the 1st and 99th percentiles of changes in government expenditures being -5 and 5 percent of GDP, respectively.
- Additional data notes:
  - Data sources standardized include: United Nations University’s World Income Inequality Database (SWIID), the OECD Income Distribution Database, World Bank, Eurostat, and the Luxembourg Income Study.
  - Gini ranges in the sample:
    - net measures: 18 to 54
    - gross measures: 30 to 57

### Main empirical findings
- Direction and magnitude:
  - An unanticipated cut in government expenditure leads to a long-lasting increase in net income inequality (net Gini).
  - Medium-term inequality multiplier is large (about 1) and statistically significant.
  - Quantified effects:
    - An unanticipated decrease in government total expenditures of 10 percent makes the net Gini index increase by more than 0.3 percentage points five years later.
    - A cumulative decrease in government spending of 1 percent of GDP over 5 years is associated with a cumulative increase in the net Gini coefficient over the same period of about 1 percentage point.
    - The magnitude corresponds to about 1 standard deviation of the average change in the Gini coefficient in the sample.
- Distributional consequences beyond Gini:
  - Fiscal contraction leads to:
    - A decrease in the share of income held by the poorest 60 percent.
    - A rise in the share of income for the top 20 percent.
    - An increase in poverty measured by the poverty headcount ratio.
- Spending-category heterogeneity:
  - Total government expenditure shocks have a larger effect on inequality than public investment or consumption shocks.
  - Likely channel: redistributive role of transfers (data limitations prevent computing distributional effect of transfers in isolation).

### Robustness checks and alternative identification
- Robustness exercises:
  - Alternative measures of income distribution.
  - Alternative measures of fiscal shocks (including Kraay’s loan-based approach as a robustness check).
  - Controls for endogeneity and omitted variable bias, including:
    - current and lagged output growth innovations (actual GDP growth minus analysts' October forecast);
    - contemporaneous revenue surprises;
    - alternative forecast timing (October of previous year);
    - changes in trade openness, changes in financial depth, political crises, changes in terms of trade.
  - Results are robust: estimates remain statistically significant and close to baseline across these checks.
- Alternative identification (Kraay 2014):
  - Inequality multiplier lower under Kraay’s identification than baseline due to a stronger response of government spending to the spending shock itself, but the two multipliers are not statistically different.

### Heterogeneity: time periods, country groups, shock sign and size, business cycle
- Time periods:
  - Sample split at 2003 (pre-2003 vs post-2003); responses of inequality to government spending shocks are rather stable across time; multipliers do not differ significantly.
- Country groups:
  - Emerging Markets (EMs) versus Low-Income Countries (LICs):
    - Responses in EMs tend to be larger than in LICs but differences not statistically significant.
    - Point estimate of impulse response larger for EMs; point estimate of medium-term multiplier larger in LICs.
    - Average effect of fiscal shocks on government spending-to-GDP ratio is smaller in LICs than in EMs.
- Sign and size of shocks:
  - Responses of net inequality to positive (expansions) and negative (consolidations) shocks are not statistically different.
  - Inequality multiplier virtually the same for positive and negative shocks.
  - Large positive shocks (greater than the 75th percentile = 5%) and large negative shocks (smaller than the 25th percentile = -6%) show point estimates of multipliers not statistically different from each other.
- Business cycle state:
  - Smooth-transition-function (STAR) approach with γ set to 1.5; z is real GDP growth normalized to zero mean and unit variance.
  - Point estimates of impulse responses larger in expansions than in recessions.
  - Inequality multiplier does not significantly differ across recessions and expansions because the effect of fiscal shock on government-spending ratio is larger in expansions.

### Components of government expenditures and other distributional measures
- Government consumption and investment shocks:
  - Unexpected decrease in government consumption by 10 percent → about 0.08 percentage point reduction in Gini over the medium-term.
  - Unexpected decrease in government investment by 10 percent → about 0.15 percentage point reduction in Gini over the medium-term.
  - Medium-term inequality multiplier:
    - Government consumption: around 0.2 and not statistically significant.
    - Government investment: about 0.7.
  - Both multipliers are lower than that of total government expenditures (likely because total expenditures include transfers).
- Income shares (WDI) and poverty:
  - A cumulative decrease in government spending of 1 percent of GDP is associated with:
    - Cumulative reduction in income share held by the lowest 60 percent of about −0.2 percent.
    - Cumulative increase in income share held by the top 20 percent of about +0.2 percent.
    - Cumulative increase in poverty headcount ratio of about +0.7 percent.
  - These effects correspond to about 1 standard deviation of the average change in these income shares within countries.
- Measurement caveats:
  - SWIID contains model-based imputations and subject to measurement errors (Solt, 2016); Gini is more sensitive to middle-class income.

### Key quantitative summary findings and selected statistics
- Medium-term inequality multiplier (baseline estimate): about 1 — a decrease (increase) in government expenditures of 1 percent of GDP over five years leads to an increase (decrease) in the net Gini index of about 1 percentage point.
- Specific quantitative results:
  - Unanticipated decrease in government total expenditures of 10 percent → net Gini increases by more than 0.3 percentage points five years later.
  - Cumulative decrease in government spending of 1 percent of GDP over 5 years → net Gini increases by about 1 percentage point.
  - Government consumption: 10 percent unexpected decrease → about 0.08 percentage point reduction in Gini over the medium-term.
  - Government investment: 10 percent unexpected decrease → about 0.15 percentage point reduction in Gini over the medium-term.
  - Income share changes for a cumulative 1 percent of GDP decrease in spending:
    - Lowest 60 percent: about −0.2 percent (cumulative).
    - Top 20 percent: about +0.2 percent (cumulative).
    - Poverty headcount ratio: about +0.7 percent (cumulative).
- Selected descriptive statistics (EMDE full sample, Table 1):
  - Net Gini: Mean 40.26; Std.dev. 8.97; Min 14.76; Max 67.21.
  - Market Gini: Mean 46.23; Std.dev. 8.69; Min 18.53; Max 76.89.
  - ∆ Market Gini: Mean 0.07; Std.dev. 1.68; Min -11.6; Max 16.43.
  - ∆ Net Gini: Mean 0.03; Std.dev. 1.33; Min -7.05; Max 12.67.
- Inequality multipliers (Table 4, Multiplier (h=5)):
  - Baseline: -0.958*** (0.318).
  - Growth Forecasts: -0.997*** (0.362).
  - Revenue Shocks: -0.877*** (0.291).
  - FS in previous year: -0.461** (0.197).
  - Additional controls: -0.922*** (0.315).
  - Before 2003: -0.949*** (0.303).
  - After 2003: -0.961*** (0.316).
  - EMs: -0.645* (0.355).
  - LICs: -0.160*** (0.417).
  - Positive Shocks: -1.006*** (0.379).
  - Negative Shocks: -1.008*** (0.382).
  - Large Positive Shocks: -1.005*** (0.340).
  - Large Negative Shocks: -0.923** (0.428).
  - Expansions: -0.970*** (0.321).
  - Recessions: -0.959*** (0.319).
  - Government Investment: -0.691** (0.315).
  - Government Consumption: -0.184 (0.173).
  - Kray: -0.775* (0.431).

### Policy implications and recommendations
- Total government expenditures have a stronger distributional impact than government consumption or investment alone, likely because total expenditures include transfers with direct redistributive effects.
- In contexts of fiscal consolidation, prioritizing spending items is key to mitigate distributional impacts.
- A budget-neutral shift of government expenditure composition away from consumption toward public investment could be output-enhancing and could help reduce inequality in the medium-term.
- Progressive taxation and targeted social benefits can offset adverse distributional impacts of spending-based adjustments (consistent with evidence for OECD countries).

*Source — Content unit: wp1857 - References (PDF chapter/section)*

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

### wp1857 - References

### Introduction and research question
- Focus: effect of government spending shocks on income distribution in emerging and developing economies (EMDEs).
- Sample and scope:
  - Panel of 103 EMDEs.
  - Period covered: 1990-2016.
- Motivations:
  - Inequality has risen in many EMDEs and remains high in regions such as Latin America and Sub-Saharan Africa.
  - Many EMDEs have recently experienced rising public debt-to-GDP ratios.
  - Policy trade-off: addressing high/rising inequality while maintaining/regaining fiscal sustainability.
- Primary empirical question: What is the effect of government spending shocks on income distribution?

### Methodology and identification
- Baseline identification:
  - Adopted Auerbach and Gorodnichenko (2013a, 2013b) (AG) approach: identify government spending shocks as forecast errors in government spending.
  - Rationale: aligns econometrician’s information set with that of economic agents and mitigates “fiscal foresight” bias.
- Alternative identification approaches discussed (not used as baseline due to data constraints):
  - Natural experiment approach using military spending (Barro, Ramey).
  - Structural VAR (SVAR) approach (Blanchard and Perotti).
  - Kraay (2012, 2014) method using loans from official creditors (used as robustness check).
- Estimation technique:
  - Local projections approach of Jordà (1995) to trace short- and medium-run responses of inequality measures to identified fiscal shocks.
  - Computation of cumulative “inequality multiplier” following analogue of three-step approach of Ramey and Zubairy (2018): integral of the inequality response divided by integral of the government spending response.

### Construction of fiscal shocks and key data sources
- Fiscal shocks (FE) constructed as forecast errors using IMF WEO forecasts made in October of the same year:
  - FE_{i,t} = ∆lnG_{i,t} − ∆lnG^E_{i,t}
  - Formula explicitly: FE_{i,t} = (ln G_{i,t}^{apr,2017} − ln G_{i,t−1}^{apr,2017}) − (ln G_{i,t}^{oct,t} − ln G_{i,t−1}^{oct,t})
- Justification for October forecasts:
  - Minimizes likelihood that unanticipated changes in government spending reflect endogenous responses to contemporaneous economic developments; legislative amendment timing makes within-quarter endogenous adjustments unlikely.
- Distributional properties of shocks (EMDEs):
  - For total government spending: bulk of shocks (1st to 99th percentile) lie between -41 and 45 percent.
  - For government consumption: bulk between -42 and 49 percent.
  - For government investment: bulk between -101 and 95 percent.
  - These shocks translate into an approximately symmetric distribution of changes in government expenditures, with the 1st and 99th percentiles of changes in government expenditures being -5 and 5 percent of GDP, respectively.
- Inequality data:
  - Source: Standardized World Income Inequality Database (SWIID 5.1).
  - Measures available: net (post-tax, post-transfers) and market (pre-tax, pre-transfers) Gini indices.
  - Income definition per SWIID: sum of monetary and non-monetary income from labor, monetary income from capital, monetary social security transfers (including work-related insurance transfers, universal transfers, and assistance transfers), non-monetary social assistance transfers, monetary and non-monetary private transfers, less income taxes and social contributions.

### Main empirical findings
- Direction and magnitude:
  - An unanticipated cut in government expenditure leads to a long-lasting increase in net income inequality (net Gini).
  - The medium-term inequality multiplier is large (about 1) and statistically significant.
  - Quantified effect: a cumulative decrease in government spending of 1 percent of GDP over 5 years is associated with a cumulative increase in the Gini coefficient over the same period of about 1 percentage point.
  - Interpretation: the above increase corresponds to about 1 standard deviation of the average change in the Gini coefficient in the sample.
- Distributional consequences beyond Gini:
  - Fiscal contraction leads to:
    - A decrease in the share of income held by the poorest 60 percent of the population.
    - A rise in the share of income for the top 20 percent.
    - An increase in poverty measured by the poverty headcount ratio.
- Spending-category heterogeneity:
  - Total government expenditure shocks have a larger effect on inequality than public investment or consumption shocks.
  - Likely channel: redistributive role of transfers (data limitations prevent computing distributional effect of transfers in isolation).

### Robustness checks and heterogeneity analysis
- Robustness exercises:
  - Alternative measures of income distribution.
  - Alternative measures of fiscal shocks (including Kraay’s loan-based approach as a robustness check).
  - Checks for endogeneity and omitted variable bias.
  - Controls for size and sign of fiscal shocks (positive versus negative; large positive versus large negative).
  - Controls for state of the business cycle (expansions versus recessions).
  - Subsample analysis by level of development (low-income countries versus emerging markets).
- Findings are robust across these checks.

### Relation to existing literature and contributions
- Novelty:
  - Focus on EMDEs while most previous studies concentrate on advanced economies.
  - Introduction and use of an “inequality multiplier” concept to compare effects across states.
- Comparison with prior results:
  - Consistent with findings for OECD countries that spending-based adjustments tend to worsen inequality, whereas progressive taxation and targeted social benefits can offset adverse distributional impacts.
  - Complements Furceri and Li (2017) by comparing effects of various types of government spending shocks on multiple inequality measures, rather than focusing only on determinants of public investment multiplier.

_Italic: Source — Content unit: wp1857 - References (PDF chapter/section)_

### 2013. It incorporates data from several sources (United Nations University’s World  Income

### wp1857 - 2013. It incorporates data from several sources (United Nations University’s World  Income

### Data and inequality measures
- Data sources standardized include: United Nations University’s World Income Inequality Database (SWIID), the OECD Income Distribution Database, World Bank, Eurostat, and the Luxembourg Income Study.
- As a robustness check, income shares from the WDI are used.
- Gini coefficients: theoretically bounded between 0 and 100.
- Observed Gini ranges in the sample:
  - net measures: 18 to 54
  - gross measures: 30 to 57
- Higher levels of inequality are typically recorded for developing countries.

### Empirical methodology
- Primary objective: estimate average impact of government spending shocks on:
  - net Gini (net of transfers and taxes)
  - income shares by population percentiles
  - poverty
- Estimation approach: local projection method (LPM) (Jordà, 2005) to estimate impulse-response functions.
- Baseline regression specification (equation (2)):
  - y_{i,t+k} = α_{i}^{k} + θ_{t}^{k} + β^{k} FE_{i,t} + θ^{k} X_{i,t} + ε_{i,t}^{k}
  - y is the Gini coefficient; α_{i} are country fixed effects; θ_{t} are time fixed effects; FE_{i,t} is the government spending shock; X_{i,t} includes controls, two lags of the shocks, and two lags of the Gini coefficient; ε is the error term.
  - Equation is estimated for each k = 0,1,...5 (k = 0 is the year when the shock takes place).
- Impulse-response functions computed from estimated β^{k}; confidence bands from clustered robust standard errors at the country level.

### Computing the inequality multiplier
- Inequality multiplier at horizon h defined as:
  - (∑_{k=0}^{h} y_{t+k}) / (∑_{k=0}^{h} g_{t+k})
  - ∑_{k=0}^{h} y_{t+k} is the sum of inequality from t to t+h.
  - ∑_{k=0}^{h} g_{t+k} is the sum of the government spending variable (government spending-to-GDP ratio) from t to t+h.
- Adopted three-step approach (Ramey and Zubairy (2018)):
  1. Estimate impulse response of inequality to fiscal shocks using equation (2).
  2. Estimate impulse response of the government spending-to-GDP ratio to fiscal shocks using the analogue of equation (2).
  3. Compute h-period ahead multipliers as the ratio of these two responses.
- Rationale: avoids using a time- and cross-country-varying ex-post conversion factor (GDP-to-government expenditure ratio) that could bias multipliers.

### Results — Baseline
- Figures and tables use effects of an anticipated fiscal consolidation by changing the sign of estimated LPM coefficients; methodology treats positive and negative fiscal shocks symmetrically.
- Main finding: an unanticipated fiscal contraction leads to a long-lasting and statistically significant increase in income inequality.
- Quantitative results:
  - An unanticipated decrease in government total expenditures of 10 percent makes the net Gini index increase by more than 0.3 percentage points five years later.
  - A cumulative decrease in government spending of 1 percent of GDP over 5 years is associated with a cumulative increase in the net Gini coefficient over the same period of about 1 percentage point (Table 4, column I).
  - The magnitude corresponds to about 1 standard deviation of the average change in the Gini coefficient in the sample.
- Comparable magnitude to findings for advanced economies (Ball et al. (2013); Agnello and Sousa (2014)).

### Robustness checks
- Endogeneity to output growth:
  - Control: add current and lagged output growth innovations (actual GDP growth minus analysts' October forecast).
  - Results: similar to baseline; medium-term multiplier very close to baseline (Figure 3; Table 3, column II; Table 4, column II).
  - Practical implementation: two-stage approach—first regress government spending forecast errors on GDP growth forecast errors; use residuals as measure of government spending shocks.
- Contemporaneous revenue shocks:
  - Control: include revenue surprises.
  - Results: estimates remain statistically significant and close to baseline; medium-term multiplier slightly smaller but not statistically different (Figure 3; Table 3, column III; Table 4, column III).
- Alternative forecast timing:
  - Use forecasts made in October of the previous year (rather than October of the same year).
  - Results: response of inequality not statistically different from baseline (Figure 3; Table 3, column IV).
  - Point estimate of implied inequality multiplier is lower (Table 3, column IV), but confidence bands overlap with baseline.
- Additional controls for other drivers of inequality:
  - Controls considered: (i) changes in trade openness (exports + imports as share of GDP); (ii) changes in financial depth (private credit-to-GDP ratio); (iii) political crises; (iv) changes in terms of trade.
  - Results: similar to baseline, supporting the exogeneity of identified fiscal policy shocks (Figure 3; Table 3, column V; Table 4, column V).

### Effect across time periods
- Sample span: 1990 to 2016.
- Sample split: two equal parts using 2003 as cut-off to compare pre-2003 and post-2003 responses.
- Rationale: many LICs experienced structural change in mid-2000s with higher average real GDP growth rates and lower output and inflation volatility.
- Modified specification (equation (3)) to allow different responses pre- and post-2003:
  - y_{i,t+k} = α_{i}^{k} + θ_{t}^{k} + β_{1}^{k} D_{it} FE_{i,t} + β_{2}^{k} (1−D_{it}) FE_{i,t} + θ^{k} X_{i,t} + ε_{i,t}^{k}
  - D is a dummy equal to one for pre-2003 observations and zero for post-2003 observations.

*Source: wp1857 - 2013. It incorporates data from several sources (United Nations University’s World  Income*

### 2003. The results reported in Figure 4 suggest that the responses of inequality to government

### wp1857 - 2003. The results reported in Figure 4 suggest that the responses of inequality to government

### Responses across time and income groups
- Responses of inequality to government spending shocks are rather stable across time; subsample estimates and multiplier effects (Table 4, column VI-VII) do not differ significantly from those for the full sample.
- Distinction between Emerging Markets (EMs) and Low-Income Countries (LICs):
  - Responses of inequality to government spending shocks in EMs tend to be larger than in LICs, but the differences are not statistically significantly different from each other or from the entire sample (Figure 5).
  - Point estimate of the impulse response is larger for EMs than for LICs, while the point estimate of the medium-term multiplier is larger in LICs (Table 4, column VIII-IX).
  - The average effect of fiscal shocks on the government spending-to-GDP ratio is smaller in LICs than in EMs; failing to compute the inequality multiplier could lead to incorrect conclusions about magnitude.

### Sign and magnitude of fiscal shocks
- Sign of shocks:
  - Estimated responses of net inequality to positive (expansions) and negative (consolidations) government spending shocks are not statistically different (Figure 6).
  - Inequality multiplier is virtually the same for positive and negative shocks (Table 4, column X and XI).
  - Note: sign of estimated coefficients and implied multiplier is the same (negative) because negative spending shocks deliver an increase in net inequality and vice versa.
- Size of shocks:
  - Equation (4) uses D1 = 1 for large positive government spending shocks—greater than the 75th percentile (5%)—and D2 = 1 for large negative shocks—smaller than the 25th percentile (-6%)—with shocks ranked unconditionally across countries and years.
  - Responses of inequality to large positive and large negative shocks are reported in Figure 7.
  - Point estimate of the inequality multiplier for large negative shocks is slightly smaller than for large positive shocks, but the two are not statistically different (Table 4, columns XII and XIII).
- Overall: neither the size nor the sign of the shocks significantly affect the response of inequality to government spending shocks.

### State-dependent multipliers (business cycle)
- Method: smooth-transition-function (STAR) approach modifying equation (2) with G(z_it) = exp(−γ z_it) / (1+exp(−γ z_it)), γ>0; z is real GDP growth normalized to zero mean and unit variance; γ set to 1.5 as in Abiad et al (2015).
- Findings:
  - Point estimates of impulse responses are larger in expansions than in recessions (Figure 8).
  - Inequality multiplier does not significantly differ across recessions and expansions (Table 4, columns XIV and XV), because the effect of the fiscal shock on the government-spending ratio is larger in expansions than recessions.
  - Again, failure to compute the inequality multiplier could produce wrong conclusions.

### Components of government expenditures
- Constructed unexpected government consumption shocks (FE_C) and investment shocks (FE_I) analogous to total government expenditure shocks; replaced FE in equation (2) with FE_C and FE_I one at a time.
- Effects:
  - A reduction in both government consumption and government investment has a persistent and negative effect on inequality.
  - An unexpected decrease in government consumption by 10 percent reduces the Gini index over the medium-term by about 0.08 percentage point.
  - An unexpected decrease in government investment by 10 percent reduces the Gini index over the medium-term by about 0.15 percentage point.
  - While point estimates are larger for consumption than investment, the inequality multiplier is larger for investment than for consumption because the response of total government spending-to-GDP is larger for government consumption than investment (Table 4, columns XVI-XVII).
  - Medium-term inequality multiplier associated with government consumption is around 0.2 and not statistically significant; multiplier associated with investment is about 0.7.
  - Both multipliers are lower than that of total government expenditures, likely because total expenditures include transfers, which directly affect net inequality.

### Other distributional measures
- Concerns about SWIID: some observations obtained by model-based imputations and subject to measurement errors (Solt, 2016); Gini is also more sensitive to middle-class income.
- Re-examination using income shares from WDI:
  - Equation (2) estimated with dependent variable as average income share from t to t+5 (data sparse).
  - A cumulative decrease in government spending of 1 percent of GDP is associated with a cumulative reduction (increase) in the income share held by the lowest 60 percent (top 20 percent) of about 0.2 percent (Table 5, columns I-V).
  - Effect corresponds to about 1 standard deviation of the average change in these income shares within countries.
- Poverty:
  - Using poverty headcount ratio from WDI averaged from t to t+5, a cumulative decrease in government spending of 1 percent of GDP is associated with a cumulative increase in the headcount ratio of about 0.7 percent (Table 5, column VI).

### Alternative identification and robustness
- Kray (2014) alternative identification focuses on predictable fiscal shocks using a dataset of lending by official creditors to governments in developing countries to construct an instrument for government spending; exploits lags between approval and disbursement.
- Using Kray’s identification for government spending shocks:
  - Inequality multiplier is lower under Kray’s identification than under the baseline due to a stronger response of government spending to the spending shock itself.
  - The two multipliers are not statistically different from each other.
  - Kray’s instrument covers 102 countries over 1970-2010; estimation here starts from 1990 for comparability.

### Key quantitative summary findings
- Medium-term inequality multiplier (baseline estimate): about 1 — a decrease (increase) in government expenditures of 1 percent of GDP over five years leads to an increase (decrease) in the net Gini index of about 1 percentage point.
- Government consumption: 10 percent unexpected decrease → about 0.08 percentage point reduction in Gini over the medium-term.
- Government investment: 10 percent unexpected decrease → about 0.15 percentage point reduction in Gini over the medium-term.
- Cumulative decrease in government spending of 1 percent of GDP:
  - Income share change for lowest 60 percent (cumulative): about −0.2 percent.
  - Income share change for top 20 percent (cumulative): about +0.2 percent.
  - Poverty headcount ratio (cumulative): about +0.7 percent.

### Policy implications and recommendations
- Total government expenditures have a stronger distributional impact than government consumption or investment alone, likely because total expenditures include transfers with direct redistributive effects.
- A budget-neutral shift of government expenditure composition away from consumption toward public investment could be output-enhancing and could help reduce inequality in the medium-term.
- In contexts of fiscal consolidation, prioritizing spending items is key to mitigate distributional impacts.

*Source: wp1857 - 2003 (excerpts).*

### REFERENCES

### REFERENCES

### References
- An,  Zidong., Joao  Tovar Jalles, Prakash  Loungani, and  Ricardo  Sousa. “Do  IMF  Fiscal Forecasts add Value?”, Journal of Forecasting, 2018, forthcoming.
- Abiad, Abdul, Davide Furceri and Petia Topalova. “The Macroeconomic Effects of Public Investment: Evidence from Advanced Economies,” IMF Working Papers, no. 15/95, 2015.
- Agnello, Luca and Ricardo M Sousa. “How Does Fiscal Consolidation Impact on Income Inequality?,” Review of Income and Wealth, Vol. 60, 2014, pp. 702–726.
- Auerbach, Alan J and Yuriy Gorodnichenko. "Output Spillovers from Fiscal Policy," American Economic Review, vol. 103, no. 3, 2013a, pp. 141-146.
- Auerbach, Alan J and Yuriy Gorodnichenko. "Fiscal Multipliers in Recession and Expansion," in Alberto Alesina and Francesco Giavazzi (editors), Fiscal Policy after the Financial Crisis, National Bureau of Economic Research Conference Report, 2013b, Chicago IL, University of Chicago Press, pp. 63-98.
- Ball, Laurence M, Davide Furceri, Daniel Leigh and Prakash Loungani. "The Distributional Effects of Fiscal Consolidation," IMF Working Papers, No. 13/151, International Monetary Fund, 2013.
- Barro, Robert. “Output Effects of Government Purchases,” Journal of Political Economy, vol. 89, no. 6, 1981, pp. 1086-1121.
- Batini, Nicoletta, Giovanni Callegari and Giovanni Melina. "Successful Austerity in the United States, Europe and Japan," IMF Working Papers 12/190, International Monetary Fund, 2012.
- Ben Zeev, Nadav and Evi Pappa. "Multipliers of unexpected increases in defense spending: An empirical investigation," Journal of Economic Dynamics and Control, vol. 57(C), 2015, pages 205-226.
- Berg, Andrew G. and Jonathan D. Ostry. “Inequality and Unsustainable Growth: Two Sides of the Same Coin?,” IMF Economic Review, 2017, in press.
- Blanchard, Olivier, and Roberto Perotti. “An Empirical Characterization of the Dynamic Effects of Changes in Government Spending and Taxes on Output,” The Quarterly Journal of Economics, vol. 117, no. 4, 2002, pp. 1329–1368.
- Blanchard, Olivier, & Daniel Leigh, 2013. "Growth Forecast Errors and Fiscal Multipliers," American Economic Review, American Economic Association, vol. 103(3), pages 117-120, May.
- Forni, Mario, and Luca Gambetti. “Fiscal Foresight and the Effects of Government Spending,” CEPR Discussion Paper, 2010.
- Furceri, Davide and Bin G. Li. “The Macroeconomic (and Distributional) Effects of Public Investment in Developing Economies,” IMF Working Papers, No. 17/217, International Monetary Fund, 2017.
- Granger, C W J and Timo Teräsvirta. Modelling Nonlinear Economic Relationships. Oxford, Oxford University Press, 1993.
- Ilzetzki, Ethan, Enrique G Mendoza and Carlos A Vegh. “How Big (Small?) are Fiscal Multipliers?,” Journal of Monetary Economics, vol. 60, no. 2, 2013, pp. 239-254.
- IMF. “Fiscal Policy and Income Inequality,” IMF Policy Paper. Washington DC, International Monetary Fund, 2014.
- Jordà, Òscar. “Estimation and Inference of Impulse Responses by Local Projections,” American Economic Review, vol. 95, no. 1, 2005, pp. 161–182.
- Kraay, Aart. “How Large Is the Government Spending Multiplier? Evidence from World Bank Lending,” The Quarterly Journal of Economics, vol. 127, no. 2, 2012, pp. 829–887.
- Kraay, Aart. 2014. "Government Spending Multipliers in Developing Countries: Evidence from Lending by Official Creditors," American Economic Journal: Macroeconomics, vol. 6, no. 4, pp. 170-208.
- Lane, Philip R. and Gian Maria Milesi-Ferretti. “The external wealth of nations mark II: Revised and extended estimates of foreign assets and liabilities, 1970–2004,” Journal of International Economics, vol. 73, 2007, pp. 223-250.
- Leeper, Eric M, Alexander W Richter and Todd B. Walker. “Corrigendum: Quantitative Effects of Fiscal Foresight,” American Economic Journal: Economic Policy, vol. 4, no. 3, 2012, pp. 283–283.
- Leeper, Eric M, Todd B Walker and Shu‐Chun S Yang. “Fiscal Foresight and Information Flows,” Econometrica, vol. 81, no. 3, 2013, pp. 1115–1145.
- Melina, Giovanni and Rafael Portillo. "Economic Fluctuations in Sub-Saharan Africa," IMF Working Papers 17/XXX, International Monetary Fund, Forthcoming, 2017.
- Owyang Michael T, Valerie A Ramey and Sarah Zubairy, 2013. "Are Government Spending Multipliers Greater during Periods of Slack? Evidence from Twentieth-Century Historical Data," American Economic Review, vol. 103, no. 3, 2013, pp. 129-134.
- Ramey, Valerie, A “Identifying Government Spending Shocks: It’s all in the Timing,” The Quarterly Journal of Economics, vol. 126, no. 1, 2011a, pp. 1-50.
- Ramey, Valerie, A “Can Government Purchases Stimulate the Economy?,” Journal of Economic Literature, vol. 49, no. 3, 2011b, pp. 673-685.
- Ramey, Valerie, A, and Matthew D Shapiro. “Costly capital reallocation and the effects of government spending,” Carnegie-Rochester Conference Series on Public Policy, vol. 48, no. 1, 1998, pp. 145-194.
- Ramey, Valerie, A, and Sarah Zubairy. “Government Spending Multipliers in Good Times and in Bad: Evidence from U.S. Historical Data,” Journal of Political Economy, 2018, forthcoming.
- Ricco, G. “A New Identification of Fiscal Shocks Based on the Information Flow,” ECB Working Paper 1813, European Central Bank, 2015.
- Solt, Frederick. “The Standardized World Income Inequality Database,” Social Science Quarterly, vol. 97, 2016, 1267–1281.
- Stock, James H, and Mark W Watson. “Why Has U.S. Inflation Become Harder to Forecast?” Journal of Money, Credit and Banking, vol. 39, no. 1, 2007, pp. 3-33.
- Valencia, Fabian and Luc Laeven. “Systemic Banking Crises Database; An Update,” IMF Working Papers, no. 12/163, International Monetary Fund, 2012.
- Woo, Jaejoon, Elva Bova, Tidiane Kinda, and Yuanyan S Zhang. “Distributional Consequences of Fiscal Adjustments: What Do the Data Say?,” IMF Economic Review, vol. 65, no. 2, 2017, pp. 273-307.

### Key Tables and Selected Statistics
- Table 1. Gini Coefficients: Descriptive Statistics — EMDEs (full sample)
  - Net Gini: Mean 40.26; Std.dev. 8.97; Min 14.76; Max 67.21
  - Market Gini: Mean 46.23; Std.dev. 8.69; Min 18.53; Max 76.89
  - ∆ Market Gini: Mean 0.07; Std.dev. 1.68; Min -11.6; Max 16.43
  - ∆ Net Gini: Mean 0.03; Std.dev. 1.33; Min -7.05; Max 12.67
- Table 1. EMs
  - Net Gini: Mean 39.76; Std.dev. 9.47; Min 14.76; Max 67.21
  - Market Gini: Mean 45.66; Std.dev. 8.8; Min 18.53; Max 76.89
  - ∆ Market Gini: Mean 0.1; Std.dev. 1.7; Min -11.6; Max 16.43
  - ∆ Net Gini: Mean 0.04; Std.dev. 1.29; Min -7.05; Max 12.67
- Table 1. LICSs
  - Net Gini: Mean 41.24; Std.dev. 7.81; Min 20.38; Max 61.84
  - Market Gini: Mean 47.35; Std.dev. 8.37; Min 24.58; Max 75.36
  - ∆ Market Gini: Mean 0.01; Std.dev. 1.64; Min -7.09; Max 9.45
  - ∆ Net Gini: Mean -0.01; Std.dev. 1.41; Min -4.61; Max 9.16

- Table 2. LMP Regression Results at Selected Time Horizons — Dependent Variable: Net Inequality
  - Public Expenditure Shocks (k=0, k=1, k=5)
    - FE(t): -0.231, -0.444, -3.381*** (standard errors: (0.328), (0.722), (1.088))
    - FE(t-1): -0.141, -0.236, -1.724* (0.248), (0.491), (0.958)
    - FE(t-2): -0.199, -0.330, -2.091** (0.198), (0.436), (0.895)
    - Gini(t-1): 1.228***, 1.346***, 0.015 (0.065), (0.116), (0.136)
    - Gini(t-2): -0.202***, -0.621***, 0.243** (0.069), (0.126), (0.102)
    - Gini(t-3): -0.169***, -0.070, -0.292** (0.034), (0.076), (0.146)
    - Constant: 4.831***, 12.544***, 39.783*** (1.000), (2.271), (5.038)
    - Observations: 993, 896, 553; R-squared: 0.991, 0.977, 0.964
  - Public Consumption Shocks (k=0, k=1, k=5)
    - FE(t): -0.051, -0.255, -1.393* (0.241), (0.402), (0.829)
    - FE(t-1): 0.094, -0.126, -0.187 (0.199), (0.313), (0.454)
    - FE(t-2): -0.393**, -0.575**, -0.990 (0.156), (0.244), (0.794)
    - Gini(t-1): 1.284***, 1.444***, 0.306*** (0.043), (0.086), (0.090)
    - Gini(t-2): -0.288***, -0.737***, -0.189** (0.066), (0.107), (0.091)
    - Gini(t-3): -0.131***, -0.019, -0.095 (0.038), (0.060), (0.116)
    - Constant: 5.237***, 12.415***, 39.473*** (0.534), (1.077), (3.056)
    - Observations: 1,399, 1,296, 914; R-squared: 0.990, 0.974, 0.935
  - Public Investment Shocks (k=0, k=1, k=5)
    - FE(t): -0.160, -0.241, -0.779** (0.110), (0.200), (0.378)
    - FE(t-1): 0.036, 0.028, 0.020 (0.093), (0.173), (0.245)
    - FE(t-2): 0.043, 0.075, 0.131 (0.103), (0.166), (0.215)
    - Gini(t-1): 1.276***, 1.442***, 0.231* (0.048), (0.083), (0.127)
    - Gini(t-2): -0.247***, -0.695***, -0.108 (0.071), (0.108), (0.112)
    - Gini(t-3): -0.183***, -0.104, -0.193 (0.039), (0.067), (0.134)
    - Constant: 6.114***, 14.191***, 43.922*** (0.662), (1.447), (3.748)
    - Observations: 1,303, 1,201, 826; R-squared: 0.990, 0.973, 0.938
  - Notes: Clustered robust standard errors in parentheses. ***, **, * denote p<0.01, p<0.05, p<0.1, respectively.

- Table 3. LMP Regression Results: Robustness Checks (5-Year Horizon)
  - Baseline (k=5): FE(t) -3.381*** (1.088); FE(t-1) -1.724* (0.958); FE(t-2) -2.091** (0.895); Gini(t-1) 0.015 (0.136); Gini(t-2) 0.243** (0.102); Gini(t-3) -0.292** (0.146); Constant 39.783*** (5.038); Observations 553; R-squared 0.964.
  - Growth forecasts (k=5): FE(t) -3.102*** (1.042); Observations 554; R-squared 0.964.
  - Revenue shocks (k=5): FE(t) -2.694** (1.051); Observations 543; R-squared 0.964.
  - FS in previous year (k=5): FE(t) -2.926** (1.334); Observations 508; R-squared 0.966.
  - Additional controls (k=5): FE(t) -2.691*** (0.962); Observations 528; R-squared 0.968.
  - Notes: Clustered robust standard errors in parentheses. ***, **, * denote p<0.01, p<0.05, p<0.1, respectively.

- Table 4. Inequality Multipliers (Multiplier (h=5))
  - Baseline: -0.958*** (0.318); Observations 553; R-squared 0.984.
  - Growth Forecasts: -0.997*** (0.362); Observations 554; R-squared 0.984.
  - Revenue Shocks: -0.877*** (0.291); Observations 543; R-squared 0.985.
  - FS in previous year: -0.461** (0.197); Observations 508; R-squared 0.985.
  - Additional controls: -0.922*** (0.315); Observations 528; R-squared 0.985.
  - Before 2003: -0.949*** (0.303); Observations 553; R-squared 0.985.
  - After 2003: -0.961*** (0.316); Observations 553; R-squared 0.985.
  - EMs: -0.645* (0.355); Observations 553; R-squared 0.985.
  - LICs: -0.160*** (0.417); Observations 553; R-squared 0.985.
  - Positive Shocks: -1.006*** (0.379); Observations 553; R-squared 0.984.
  - Negative Shocks: -1.008*** (0.382); Observations 553; R-squared 0.984.
  - Large Positive Shocks: -1.005*** (0.340); Observations 553; R-squared 0.984.
  - Large Negative Shocks: -0.923** (0.428); Observations 553; R-squared 0.984.
  - Expansions: -0.970*** (0.321); Observations 553; R-squared 0.985.
  - Recessions: -0.959*** (0.319); Observations 553; R-squared 0.985.
  - Government Investment: -0.691** (0.315); Observations 826; R-squared 0.975.
  - Government Consumption: -0.184 (0.173); Observations 914; R-squared 0.974.
  - Kray: -0.775* (0.431); Observations 996; R-squared 0.953.
  - Notes: Clustered robust standard errors in parentheses. ***, **, * denote p<0.01, p<0.05, p<0.1, respectively.

- Table 5. Income Shares and Poverty Multipliers (Multiplier (h=5))
  - 1st 20 percent: -0.092** (0.036); Observations 215; R-squared 0.97.
  - 2nd 20 percent: -0.074*** (0.023); Observations 215; R-squared 0.99.
  - 3rd 20 percent: -0.050*** (0.015); Observations 215; R-squared 0.99.
  - 4th 20 percent: -0.005 (0.019); Observations 215; R-squared 0.96.
  - 5th 20 percent: 0.221*** (0.062); Observations 215; R-squared 0.99.
  - Head count ratio: 0.738** (0.3380; Observations 176; R-squared 0.92.
  - Notes: Clustered robust standard errors in parentheses. ***, **, * denote p<0.01, p<0.05, p<0.1, respectively.

### Figures (captions and notes)
- Figure 1. Share of Emerging Market and Developing Economies with rising inequality since 1990
  - Source: Standardized World Income Inequality Database (Solt, 2016), WDI.
- Figure 2. Effect of an Unexpected Decrease in Total Government Expenditures on Net Income Inequality (Net Gini Coefficients)
  - Note: x-axes denote years; t=0 is the year of the shock; the solid blue line denotes responses to an unanticipated 10 percent decrease in government expenditures; dashed lines denote 90 percent confidence bands. Estimates based on equation (2).
- Figure 3. Effect of an Unexpected Decrease in Total Government Expenditures on Net Income Inequality (Net Gini)—Alternative Shocks
  - Note: x-axes denote years; t=0 is the year of the shock; the solid blue line denotes responses to an unanticipated 10 percent decrease in government expenditures in the baseline model; dashed lines denote 90 percent confidence bands in the baseline model; solid lines in other colors denote alternative models. Estimates based on equation (2) modified appropriately for each case.
- Figure 4. Effect of an Unexpected Decrease in Total Government Expenditures on Net Income Inequality (Net Gini Coefficient)—Time Subsamples
  - Note: x-axes denots years; t=0 is the year of the shock; the solid blue line denotes responses to an unanticipated 10 percent decrease in government expenditure in the baseline sample; dashed lines denote 90 percent confidence bands in the baseline sample; solid red and green lines denote alternative subsamples. Estimates based on equation (3).
- Figure 5. Effect of an Unexpected Decrease in Total Government Expenditures on Net Income Inequality (Net Gini Coefficient)—Country Subsamples
  - Note: x-axes denote years; t=0 is the year of the shock; the solid blue line denotes percent responses to an unanticipated 10 percent decrease in government expenditures in the baseline sample; dashed lines denote 90 percent confidence bands in the baseline sample; solid red and green lines denote alternative subsamples. Estimates based on equation (3).
- Figure 6. Effect of an Unexpected Change in Government Expenditures on Net Income Inequality (Net Gini Coefficient)—Positive versus Negative Shocks
  - Note: x-axes denots years; t=0 is the year of the shock; the solid blue lines denotes percent responses to an unanticipated 10 percent decrease in government expenditures in the baseline model; dashed lines denote 90 percent confidence bands in the baseline model; solid red and green lines denote the response to negative and positive shocks, respectively. The response to positive shocks is reported with inverted sign to allow comparability. Estimates based on equation (3).
- Figure 7. Effect of an Unexpected Change in Government Expenditures on Net Income Inequality (Net Gini)—Large Positive versus Large Negative Shocks
  - Note: x-axes denote years; t=0 is the year of the shock; the solid blue line denotes percent responses to an unanticipated 10 percent decrease in government expenditures in the baseline model; dashed lines denote 90 percent confidence bands in the baseline model; solid red and green lines denote the response to negative and positive shocks, respectively. The response to positive shocks is reported with inverted sign to allow comparability. Estimates based on equation (4).
- Figure 8. Effect of an Unexpected Decrease in Total Government Expenditures on Net Income Inequality (Net Gini Coefficient)—The role of the Business Cycle
  - Note: x-axes denots years; t=0 is the year of the shock; the solid blue line denotes percent responses to an unanticipated 10 percent decrease in government expenditures in the baseline model; dashed lines denote 90 percent confidence bands in the baseline model; solid red and green lines denote alternative models. Estimates based on equation (5).
- Figure 9. Effect of an Unexpected Decrease in Government Consumption and Investment — Net Income Inequality (Net Gini)
  - Panel A: Government Consumption; Panel B: Government Investment
  - Note: x-axes denote years; t=0 is the year of the shock; solid blue lines denote percent responses to an unanticipated 1 percent decrease in government consumption (investment); dashed lines denote 90 percent confidence bands. Estimates based on equation (2).

### Appendix: Country Coverage and Distributions
- Table A1. Country Coverage — lists country, EM/LIC classification, and coverage years (1990-2015 and other country-specific ranges). Notes: EM = emerging market; LIC = low-income country. Classification based on IMF WEO.
- Figures A1–A3. Distribution of Government Expenditure Shocks in EMDE; Distribution of Government Consumption Shocks in EMDE; Distribution of Government Investment Shocks in EMDE (Source: IMF World Economic Outlook and IMF Staff calculations).

*Source: wp1857 - REFERENCES (wp1857 - REFERENCES).*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2018/wp1857.pdf_