## wp17192

## Source details

**Canonical URL:** [wp17192](https://www.imf.org/-/media/files/publications/wp/2017/wp17192.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2017/wp17192.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2017/wp17192.pdf.json)

---

### Model features and structure
- DSGE model with household heterogeneity driven by both permanent and transitory differences in labor productivity, and an endogenous labor supply.
- Four added sources of heterogeneity:
  - Three different goods: one manufacturing tradable good and two non-tradable services, with services more labor intensive than manufactured goods.
  - Cross-sectoral linkages through intermediate inputs (manufactured goods require services and vice-versa), calibrated to match the U.S. economy.
  - Sectoral differences in worker types: low-skilled, middle-skilled, and high-skilled; low-skill services produced by people closer to the bottom of the income distribution, high-skill services produced by people closer to the top.
  - Heterogeneity in consumption composition: wealthier groups dedicate a higher share of consumption expenditures to non-tradable services.
- Three aggregate commodities and production functions:
  - Manufactured M (highly tradable; capital intensive; capital complements high-skill labor h; substitutes with medium x and low l).
  - Low-skill service L (non-tradable; produced using medium x and low l).
  - High-skill service S (partly tradable; capital complements high-skill labor).
- Households: heterogeneity from permanent skill type (l, x, h) and transitory idiosyncratic productivity shocks ζ_i ∈ Ζ_i; borrowing constraint a_i ≥ −κ_i; utility u(c, ψ) specified; idiosyncratic shock processes are finite-state Markov chains Q_i.
- Government: consumes G_M, G_S, G_L; invests I_G; levies taxes (τ_s, τ_L, τ_M, τ_*, τ_M,M, τ_S,M, τ_M,S, τ_S,S, T(.)); issues foreign debt B* at interest rate r*; budget constraint specified with Γ_i and Ξ terms for consumption and intermediate input tax revenues.
- Stationary Competitive Equilibrium defined by household policy functions, firm production plans, prices {w_l, w_x, w_h, p_L, p_S, r}, measure ϑ ∈ M, and law of motion ϑ′ = H(ϑ).

### Calibration and benchmark economy
- Calibration targets and parameter choices:
  - Risk aversion σ = 2.
  - Discount factor β chosen so benchmark equilibrium capital-output ratio K/Y = 3 with β = 0.96.
  - Average hours worked correspond to 1/3 of time; labor elasticity to wages ≈ 1/3.
  - Elasticity of substitution between consumption goods assumed to be 1 (so ρ = ρ_T = ρ_N, and ρ → 0).
  - Consumption aggregator includes c̄_N > 0 so income elasticity of demand for services increases with income and to match consumption share differences across quintiles.
  - Share of services on total consumption 1−γ and low-skill services in total services consumption γ_N calibrated so low-skill and high-skill services production correspond respectively to 30 percent and 50 percent of total GDP.
  - Share of domestically produced goods in tradable goods consumption γ_T calibrated to match imports ≈ 10 percent of total consumption expenditure.
- Government and tax calibration:
  - Consumption (and intermediate consumption) tax rates set to 7.5%.
  - Government consumption chosen to be 16% of GDP (average for 2009-2015).
  - Total government consumption concentrated on manufacturing goods (G_L = G_S = I_G = 0).
  - Income tax function is negative at lower incomes (captures EITC) and peaks around 28% for people approaching multiples of median income (10 and above).
- Household labor productivity and idiosyncratic shock parameters:
  - Population shares: μ_l = 0.39, μ_x = 0.48, μ_h = 0.13.
  - Skill-level productivity: η_l = 0.7, η_x = 1, η_h = 1.1.
  - Markov persistence: φ_l = 0.76, φ_x = 0.76, φ_h = 0.73.
  - Idiosyncratic shock variances: σ_ε,l² = 0.01, σ_ε,x² = 0.0289, σ_ε,h² = 0.0484.

### Experiments and policy scenarios
- Simulated tax policy changes:
  - (i) Middle-class tax cut: reduces effective tax rates for households earning between 0.5 to 4 times the median income; initially financed by cutting wasteful government spending (lump-sum reduction).
  - (ii) Middle-class tax cut plus an EITC expansion: same middle-class cut plus EITC expansion for lower income groups; fully financed by an increase in consumption taxes; EITC calibrated so a representative household earning one-half of median income is statically fully compensated for higher consumption tax outlays.
  - (iii) High-income tax cut with EITC expansion: tax reductions accrue to those in the top quintile; funded by higher consumption taxes and same EITC expansion as in (ii).
- Additional experiment: sensitivity to openness (closed vs small open economy) and to labor supply elasticity for high-income workers.

### Key quantitative results — aggregate effects
- General:
  - The model generates positive effects on growth, consumption and investment broadly in line with the empirical literature on PIT multipliers.
  - Supply side effects are never strong enough to prevent cuts from being revenue losing (tax cuts do not “pay for themselves”).
- Middle-class tax cut (lump-sum spending cut financing):
  - Loss of revenues of 0.8 percent of GDP.
  - Raises the steady state level of GDP by just under 1 percent after 5 years.
  - Implied personal income tax multiplier of 1.1.
  - Most expansion occurs in low-skill and high-skill services; manufacturing sector shrinks in absolute terms and as a share of total output.
- Middle-class tax cut funded by VAT plus EITC expansion:
  - Growth effects are smaller than when financed via lump-sum spending cuts (muted consumption response) but still positive.
- High-income tax cut (closed economy):
  - Tends to have stronger aggregate impact than middle-class tax cuts in closed economy due to higher savings by top quintile increasing capital formation and lowering equilibrium real interest rate, and higher supply of high-skilled labor raising output.
- Open economy effects:
  - Growth response in a small open economy subject to a world interest rate is around one-half that of a closed economy where higher savings automatically generate lower interest rates and higher investment.
  - In a small open economy, high-income tax cuts yield negligible effect on investment compared to closed economy; instead increase current account surplus.

### Key quantitative results — distributional and sectoral effects
- Distributional dynamics:
  - Middle-class tax cut:
    - Both middle and low income households profit from a tax cut targeted at middle income groups due to “trickle down” demand for non-tradable services and higher wages for low-skilled labor.
    - Lowest quintile (no direct tax cut) benefits from higher demand for non-tradable services and higher wages for low-skilled labor.
    - Top quintile slightly worse off due to higher non-tradable prices and a lower interest rate.
    - Middle class (agents receiving between 50% and 150% of median income) would expand by roughly 4%.
  - Middle-class tax cut with VAT/EITC:
    - Bottom 60 percent of the income distribution better off in steady state.
    - Highest earners (top 20%) worse off in steady state consumption due to higher after-tax cost of non-tradables and lower interest rate.
    - Both middle-class cut variants reduce polarization and have progressive impacts on income distribution.
  - High-income tax cut:
    - Despite EITC expansion and some second-round supportive effects, net result is a significant decline in consumption of low and middle-income households and a significant reduction of the middle class.
    - Tax cut for the wealthy funded by higher consumption taxes significantly worsens polarization; “trickle-down” insufficient to raise welfare of the bulk of the population.
- Sectoral dynamics:
  - A middle-class tax cut increases demand for non-tradable services, raising demand for—and wages of—low-skilled labor; services expand more than manufacturing following tax reforms.
  - On average, out of a dollar spent on primary/manufactured goods in 2015, 35 cents went to labor; for a dollar spent on services, roughly 60 cents went to labor.
  - Roughly 20% of intermediate inputs used in the production of primary and manufactured goods come from services.
- Stylized facts used for aggregation:
  - After aggregation, shares of U.S. consumption (PCE) are roughly: primary and manufactured goods 20% of PCE; low-skill services 30% of PCE; high-skill services 50% of PCE.
  - The share of services in consumption rises rapidly with household income; manufactured-good expenditure share roughly 35% at the bottom of the income distribution and roughly 25% at the top quintile.

### Sensitivity analysis and robustness
- Labor supply elasticity for high-income workers:
  - Calibrated at 0.3 in baseline; lowering it from 0.3 to 0.15 reduces the growth effect by around one half for both middle and high income tax cuts.
  - Under lower elasticity, implied PIT multipliers: around 0.5 for middle-class tax cuts, and 1.5 for high-income tax cuts.
  - Qualitative result that high income tax cuts have larger macroeconomic effect preserved across parameterizations.
- Openness:
  - In a small open economy (real interest rate pinned by foreign rate) with revenue-neutral government spending offset, changes in domestic savings do not translate one-for-one into investment → high-income tax cut yields negligible effect on investment and GDP multiplier around one-half of closed economy version.
  - Even in open economy, tax cut targeted at higher earners worsens income polarization.

### Policy implications and conclusions
- Design of personal income tax changes should consider both aggregate multipliers and general equilibrium distributional effects across sectors and household types.
- Targeting tax cuts to the middle class can produce both positive macroeconomic effects and progressive distributional outcomes via sectoral demand linkages and wage effects for low-skilled labor.
- Financing choices matter:
  - Lump-sum spending cuts versus consumption taxes/VAT materially affect the magnitude of growth and distributional outcomes and should be explicitly considered when evaluating PIT reforms.
  - Shifting from direct to indirect taxation reduces but does not eliminate positive growth effects of PIT cuts; it tends to make high earners worse off relative to lower earners.
- Trade-offs:
  - High-income tax cuts can generate larger macro multipliers (especially in closed economy calibrations) but at the cost of increased income polarization and reduced welfare for low/middle-income groups when financed by higher consumption taxes.
- Broad takeaway:
  - Tax cuts generate positive effects on growth, consumption, and investment consistent with empirical literature, but are never strong enough to prevent cuts from being revenue losing; where cuts are targeted and how they are financed crucially determine both macro and distributional outcomes.

*Italic: Source — IMF Working Paper (wp17192), chapter/section excerpts from the provided PDF content.*

### References __________________________________________________________________________ 29

### wp17192 - References __________________________________________________________________________ 29

### Model features
- DSGE model with household heterogeneity driven by both permanent and transitory differences in labor productivity, and an endogenous labor supply.
- Four added sources of heterogeneity:
  - Three different goods: one manufacturing tradable good and two non-tradable services, with services more labor intensive than manufactured goods.
  - Cross-sectoral linkages through intermediate inputs (manufactured goods require services and vice-versa), calibrated to match the U.S. economy.
  - Sectoral differences in worker types: low-skilled, middle-skilled, and high-skilled; low-skill services produced by people closer to the bottom of the income distribution, high-skill services produced by people closer to the top.
  - Heterogeneity in consumption composition: wealthier groups dedicate a higher share of consumption expenditures to non-tradable services.

### Calibration and fiscal policy scenarios
- Benchmark economy calibrated to match aggregate and cross-sectional features of the U.S. economy, including:
  - Macro ratios: investment to GDP, consumption to GDP.
  - Sectoral ratios: sectoral shares of output, input-output structure.
  - Distributional statistics: tax incidence by income level, consumption by decile, composition of consumption by income group.
- Federal income taxes introduced via a parametric tax function; tax function is negative at lower incomes to capture the EITC and is progressive, particularly for households earning less than 50 percent of the median income.
- Simulated tax policy changes:
  - (i) “Middle-class tax cut”: reduces effective tax rates for households earning between 0.5 to 4 times the median income and is offset by lower government spending.
  - (ii) “Middle-class tax cut” plus an EITC expansion fully financed by an increase in consumption taxes.
  - (iii) Tax cut for high income groups combined with an EITC expansion and financed by a higher consumption tax.

### Key results and quantitative findings
- Tax-cut effects on aggregate activity:
  - The model generates positive effects on growth, consumption and investment that are broadly in line with the recent empirical literature on PIT multipliers.
  - Supply side effects are never strong enough to prevent cuts from being revenue losing (tax cuts do not “pay for themselves”).
- Distributional and sectoral dynamics:
  - A middle-class tax cut, financed by a lump-sum reduction in government spending, results in a loss of revenues of 0.8 percent of GDP but raises the steady state GDP by just under 1 percent after 5 years (i.e., a personal income tax multiplier of 1.1).
  - PIT cuts have important “trickle down” effects: both middle and low income households profit from a tax cut targeted at middle income groups.
  - The tax cut increases demand for non-tradable services, raising demand for—and wages of—low-skilled labor; this trickle-down mechanism can produce a progressive impact in general equilibrium and help reduce income polarization.
- Financing method matters:
  - Growth effects are smaller when lower personal income taxes are funded with a VAT, though shifting from direct to indirect tax still has a positive impact on growth.
- Top-end tax cuts:
  - Tax cuts for higher income groups tend to have a stronger aggregate impact than tax cuts for the middle class.
  - In the simple case where the tax cuts are paid for by lump sum cuts in government spending, the personal income tax multiplier is around [value not provided in this excerpt].

### Policy implications
- Design of personal income tax changes should consider both aggregate multipliers and general equilibrium distributional effects across sectors and household types.
- Targeting tax cuts to the middle class can produce both positive macroeconomic effects and progressive distributional outcomes via sectoral demand linkages and wage effects for low-skilled labor.
- Financing choices (lump-sum spending cuts versus consumption taxes/VAT) materially affect the magnitude of growth and distributional outcomes and should be explicitly considered when evaluating PIT reforms.

*Source: wp17192 - References __________________________________________________________________________ 29*

### 3. However, such a tax reduction generates negative distributional effects. Even

### wp17192 - 3. However, such a tax reduction generates negative distributional effects. Even

### Distributional effects of top-income tax cuts
- Tax cuts incident on high income households increase income polarization.
- Even accounting for trickle-down effects and allowing for an increase in the EITC to protect the poor, income polarization rises.
- Model findings support Mertens (2015) that top marginal rate cuts have sizeable real economic effects and spill over to lower income groups.
- Empirical references cited: Barro and Redlick (2011) and Mertens and Ravn (2013) estimated tax multipliers between 1.1 and 2.5; model generates growth effects in that range.
- Mertens (2015) documented that a top marginal rate cut for the top 1% implies a short run taxable income elasticity for the top 1% around 1.5, a rise in real GDP, a lower aggregate unemployment and positive effect on incomes outside of the top 1%.

### Sensitivity of the tax multiplier
- The size of the tax multiplier is very sensitive to the general equilibrium feed through to interest rates.
- Growth response in a small (financially) open economy subject to a world interest rate is around one-half that of a closed economy where higher savings automatically generate lower interest rates and higher investment.

### Stylized facts about the U.S. economy
- The U.S. economy can be thought of as producing and consuming three baskets of goods and services:
  - (i) primary and manufactured goods produced with little labor and highly tradable;
  - (ii) low-skilled services produced using a lot of labor and are mostly not tradable;
  - (iii) high-skill services, which are partly tradable and use a combination of high and low skilled labor.
- The share of services in consumption rises rapidly with household income.

### Heterogeneity in goods and production (metrics and findings)
- Metrics constructed using 2015 BEA NIPA Input-Output “use table” (after redefinitions, evaluated at producer prices):
  - (i) share of each commodity in US domestic demand (absorption);
  - (ii) tradability of each commodity (ratio of exports and imports of commodity i to final US demand of commodity i);
  - (iii) labor intensity approximated by network-adjusted labor share of each commodity (NALI).
- NALI accounts for labor compensation along the whole supply chain; computed as NALI_i = L w (where L is vector of labor compensation as a fraction of output by industry, and w is industry-by-commodity domestic requirement matrix).
- Two stylized facts from Figure 1:
  - Heterogeneity in labor intensity across commodities is significant.
  - Primary products (agriculture, mining and utilities) and manufactured goods are produced with much less labor than services, even after accounting for interlinkages.
  - On average, out of a dollar spent on primary/manufactured goods in 2015, 35 cents went to labor; for a dollar spent on services, roughly 60 cents went to labor.
  - Except for transportation and warehousing services, primary and manufactured goods are much more traded than services.
- Roughly 20% of intermediate inputs used in the production of primary and manufactured goods come from services.

### Heterogeneity in labor skill intensity across sectors
- Skill proxy computed from IPUMS-CPS employment data by educational attainment categories:
  - (i) high school diploma or less;
  - (ii) college or some college;
  - (iii) above college.
- Examples:
  - More than half of the labor force employed in construction, recreation, accommodation and food services has high school diploma or less.
  - Education, health care, and finance sectors employ mostly workers with at least some college background.
- Aggregate representation into three “aggregate” commodities:
  - (i) highly tradable manufactured good (incorporates manufacturing, mining and agriculture), produced mostly with capital;
  - (ii) non-tradable low-skill service (including retail trade, accommodation and food services);
  - (iii) partly-tradable high-skill service (including finance, healthcare, education).
- Appendix A Table 1 provides aggregation method details and characteristics (shares in consumption, production, total employment, NALI).

### Heterogeneity in consumption across income
- After aggregation, shares of U.S. consumption (PCE) are:
  - primary and manufactured goods: roughly 20% of PCE;
  - low-skill services: 30% of PCE;
  - high-skill services: 50% of PCE.
- Using 2015 Consumer Expenditure Survey (CEX) by income quintiles, the share of spending on manufactured goods decreases with income:
  - roughly 35% at the bottom of the income distribution;
  - roughly 25% at the top quintile.
- Method for manufactured-good expenditure share: sum of average annual expenditures on specified CEX categories (food at home; alcoholic beverages; tobacco products and smoking supplies; utilities minus telephone services; housekeeping supplies and household equipment; apparel; gasoline and motor oil; car maintenance and repair; drugs and medical supplies; audio and visual equipment; toys, hobbies and playground equipment).

### Model structure (key ingredients)
- Two new sources of heterogeneity:
  - Three goods differing in tradability, labor and skill intensity (Manufactured M, Low-skill service L, High-skill service S).
  - Non-homothetic preferences: increasing share of expenditures on services as income grows.
- Production:
  - Manufacturing: M = F_M(h, x, l, k, m_M,M, m_S,M). Manufacturing is capital intensive; capital and high-skill labor h are complements; capital is substitute with medium skill x and low skill l. Medium and low skill labor are imperfect substitutes.
  - High-skill services: S = F_S(h, x, k, m_M,S, m_S,S). Capital and high-skill labor are complements; capital and medium-skill labor are substitutes.
  - Low-skill services: L = F_L(x, l). Produced using only low skill and medium skill labor.
  - Manufactured good M is tradable (price determined exogenously); both services are assumed non-tradable.
- Households:
  - Heterogeneity from permanent skill type (l, x, h) and transitory idiosyncratic productivity shocks ζ_i ∈ Ζ_i.
  - Households maximize expected lifetime utility over consumption vector c_i = (c_s,i, c_L,i, c_M,i, c_*,i) and leisure ψ_i, accumulate assets a_i with market return r, receive market wage w_i for labor supply 1−ψ_i, face income and consumption taxes and borrowing constraint a_i ≥ −κ_i.
  - Utility specification and constraints given (policy functions v, c, ψ, a′).
  - Idiosyncratic shock process ζ_i follows finite-state Markov chain Q_i.
- Government:
  - Consumes G_M, G_S, G_L; invests I_G; levies taxes (τ_s, τ_L, τ_M, τ_*, τ_M,M, τ_S,M, τ_M,S, τ_S,S, T(.)); issues foreign debt B* at interest rate r*.
  - Budget constraint specified: ∑∫Γ_i Q_i(ζ_i,Ζ_i) i + Ξ + ∑∫T(...)Q_i(...) i + (1+r*)B* = G_M + G_S + G_L + I_G + B*′, with definitions for Γ_i and Ξ in terms of consumption and intermediate input tax revenues.
- Stationary Competitive Equilibrium:
  - Defines household policy functions, firm production plans, prices {w_l, w_x, w_h, p_L, p_S, r}, measure ϑ ∈ M, satisfying household optimization, factor price conditions, government budget balance, market clearing conditions, and law of motion ϑ′ = H(ϑ).
  - Aggregation sets and spaces defined: a_i ∈ A_i = [−κ_i, ∞), ζ_i ∈ Ζ_i = {ζ_1^i, ζ_2^i, …, ζ_n^i}, A = A_l × A_x × A_h, Ζ = Ζ_l × Ζ_x × Ζ_h, Ψ = A × Ζ.
  - Equilibrium conditions enumerated 1–5 (policy solving, factor prices equal marginal products, government budget, market clearing, law of motion for measure).

### Benchmark economy calibration
- Steady state (“benchmark US economy”) chosen to match key macro ratios and distributional statistics.
- Preferences: period utility
  - u(c, ψ) = [1/(1−σ)](c − ω(1−ψ)^(θ+1)/(θ+1))^(1−σ).
- Consumption aggregator:
  - c = [γ(c_T(c_M, c_*))^ρ + (1−γ)(c_N(c_s, c_L) + c̄_N)^ρ]^(1/ρ)
  - c_T = [γ_T (c_M)^ρ_T + (1−γ_T)(c_*)^ρ_T]^(1/ρ_T)
  - c_N = [γ_N (c_L)^ρ_N + (1−γ_N)(c_S)^ρ_N]^(1/ρ_N)
- Calibration choices and targets:
  - Risk aversion σ fixed at 2.
  - Discount factor β chosen so benchmark equilibrium capital-output ratio K/Y = 3 with β = 0.96.
  - ω and θ chosen so average hours worked correspond to 1/3 of time and labor elasticity to wages ≈ 1/3.
  - Elasticity of substitution between consumption goods assumed to be 1 (so ρ = ρ_T = ρ_N, and ρ → 0).
  - Parameter c̄_N > 0 chosen so income elasticity of demand for services increases with income and such that consumption share of manufacturing and imports is approximately eight percentage points larger for the top quintile than for the bottom quintile.
  - Share of services on total consumption 1−γ and low-skill services in total services consumption γ_N calibrated to match production shares: low-skill and high-skill services production correspond respectively to 30 percent and 50 percent of total GDP.
  - Share of domestically produced goods in tradable goods consumption γ_T calibrated to match imports ≈ 10 percent of total consumption expenditure.
- Table 1. Preferences Parameters (excerpt):
  - β = 0.96 — Target K/Y = 3
  - σ = 2 — Fixed: in the range of common values used for this parameter.

*Italic: Source — wp17192 (excerpt) from the provided IMF PDF content.*

### 0.01 Elasticity of substitution between

### 0.01 Elasticity of substitution between different types of consumptions being 1

### Model structure and calibration
- Preferences and production
  - CES production functions assumed for the three sectors.
  - Labor intensities and intermediate output intensities jointly calibrated to match the data.
- Government parameters
  - Consumption (and intermediate consumption) tax rates set to 7.5%.
  - Government consumption chosen to be 16% of GDP (average for 2009-2015).
  - Total government consumption concentrated on manufacturing goods (i.e., 퐺퐿 = 퐺푆 = 퐼퐺 = 0).
  - Income tax function follows a Conesa and Krueger (2006) variant; effective tax rate plotted as a function of multiples of median income:
    - Tax function is negative and very progressive for households earning less than 50% of median income (EITC presence).
    - Function set to peak around 28% for people approaching multiples of median income (10 and above).
- Household labor productivity
  - Permanent component proxied by education attainment; medium skill normalized to 휂ₓ = 1.
  - Skill-level parameters and population shares (Table 2):
    - 휇ₗ = 0.39 — Population share with high school degree or less.
    - 휇ₓ = 0.48 — Population with some college, no degree, associate degree or bachelor degree.
    - 휇ₕ = 0.13 — Population share with a Bachelor’s degree or more.
    - 휂ₗ = 0.7 — Approx. 12 years of education.
    - 휂ₓ = 1 — Approx. 16 years of education.
    - 휂ₕ = 1.1 — More than 16 years of education.
  - Stochastic idiosyncratic productivity:
    - Each process governed by a 3 state Markov chain, summarized by persistence 휑ᵢ and variance 휎_휖,ᵢ² (Table 3).
    - 휑ₗ = 0.76 — Persistence parameter for low skill households.
    - 휑ₓ = 0.76 — Persistence parameter for medium skill households.
    - 휑ₕ = 0.73 — Persistence parameter for high skill households.
    - 휎_휖,ₗ² = 0.01 — Variance for low skill idiosyncratic shock.
    - 휎_휖,ₓ² = 0.0289 — Variance for medium skill idiosyncratic shock.
    - 휎_휖,ₕ² = 0.0484 — Variance for high skill idiosyncratic shock.

### Experiments: design and scenarios
- Two main tax-cut scenarios simulated (tax cuts applied to labor and investment income):
  - (i) Middle-class tax cut:
    - Reduces effective tax rates for households earning between 0.5 to 4 times the median income.
    - Revenue neutrality initially achieved by cutting wasteful government spending (no feedback into model).
  - (ii) Middle-class tax cut with EITC expansion:
    - Same middle-class tax cut plus EITC expansion for lower income groups.
    - Entirely paid for by an increase in consumption taxes.
    - EITC calibrated so a representative household earning one-half of median income is, in a static sense, fully compensated for higher consumption tax outlays.
- Additional experiment:
  - High-income tax cut:
    - Tax reductions accrue to those in the top quintile.
    - Funded by higher consumption taxes and the same EITC expansion as in (ii).

### Key results — Middle-class tax cut
- Aggregate effects
  - Loss of revenues of 0.8 percent of GDP.
  - Raises steady state level of GDP by just under 1 percent after 5 years.
  - Implied personal income tax multiplier of 1.1.
  - Most expansion occurs in low-skill and high-skill services; manufacturing sector shrinks in absolute terms and as share of total output.
- Distributional effects
  - Middle- and low-income households profit from the cut.
  - Lowest quintile (no direct tax cut) benefits from higher demand for non-tradable services and higher wages for low-skilled labor.
  - Top quintile slightly worse off due to higher non-tradable prices and a lower interest rate.
  - Middle class (agents receiving between 50% and 150% of median income) would expand by roughly 4%.
- General finding: supply-side response is not large enough to offset revenue loss (income tax cuts do not “pay for themselves”).

### Key results — Consumption tax and EITC expansion (middle-class cut funded by VAT)
- Aggregate effects
  - Shift from direct to indirect tax produces a more modest growth effect (muted consumption response) but still positive on growth.
- Distributional effects
  - Bottom 60 percent of the income distribution better off.
  - Highest earners (top 20%) worse off in steady state consumption due to higher after-tax cost of non-tradables and lower interest rate.
  - Both middle-class cut variants (with and without VAT/EITC) reduce polarization and have progressive impacts on income distribution.

### Key results — High-income tax cut
- Aggregate effects
  - Larger growth effects relative to middle-class tax cut in the closed economy:
    - Two channels: higher savings by the top quintile increases capital formation and lowers equilibrium real interest rate; higher supply of high-skilled labor raises output.
    - Increased after-tax consumption of high-income households incident on non-tradables raises demand for labor of low- and middle-income groups.
- Distributional effects
  - Despite EITC expansion and some second-round supportive effects for low/middle incomes, net result is:
    - Significant decline in consumption of low and middle-income households.
    - Significant reduction of the middle class.
    - Tax cut for the wealthy funded by higher consumption taxes significantly worsens polarization; “trickle-down” insufficient to raise welfare of the bulk of the population.
- Calibration detail: elasticity of labor supply for higher earners calibrated at 0.3.

### Sensitivity analysis
- Labor supply elasticity
  - Lowering labor supply elasticity for high-income workers from 0.3 to 0.15 reduces the growth effect by around one half for both middle and high income tax cuts.
  - Implied PIT multipliers: around 0.5 for middle-class tax cuts, and 1.5 for high-income tax cuts under the lower elasticity.
  - The qualitative result that high income tax cuts have larger macroeconomic effect preserved across parameterizations.
- Savings, investment, and the real interest rate (open vs closed economy)
  - In a small open economy (real interest rate pinned by foreign rate) with revenue-neutral government spending offset:
    - Changes in domestic savings do not translate one-for-one into investment.
    - High-income tax cut yields negligible effect on investment compared to closed economy; instead current account surplus increases.
    - When interest rate does not fall, capital increases less and substitution between capital and labor is smaller → higher wages and after-tax incomes accrue to those with higher propensities to consume → aggregate consumption rises but insufficiently to offset lower investment.
    - Net GDP multiplier around one-half of that in closed economy version.
  - Even in open economy, tax cut targeted at higher earners worsens income polarization.

### Sectoral and distributional stylized facts
- Services more likely to expand than manufacturing as a result of tax reforms.
- Commodity aggregation (2015 BEA use table):
  - Manufactured Good: Agriculture, forestry, fishing, and hunting; Mining; Utilities; Manufacturing.
  - Low Skill service: Wholesale trade; Retail trade; Transportation and warehousing; Construction; Arts, entertainment, recreation, accommodation, and food services; Other services (except government).
  - High Skill Service: Information; Finance, insurance, real estate, rental, and leasing; Professional and business services; Educational services, health care, and social assistance; Government.
- Sectoral statistics used to calibrate: share in Consumption (PCE), share in GDP, share in total employment, Network Adjusted Labor Intensity (NALI), skill distribution within each industry.

### Conclusions and policy implications
- DSGE model calibrated to U.S. features finds:
  - Tax cuts generate positive effects on growth, consumption, and investment consistent with empirical literature on PIT multipliers, but are never strong enough to prevent cuts from being revenue losing.
  - Macroeconomic and distributional effects are highly sensitive to where tax cuts are targeted:
    - Middle-class tax cuts: positive effects on growth, reduce income inequality and polarization, expand middle class.
    - High-income tax cuts: larger macro multiplier in closed economy but substantial trade-off with increased polarization and reduced welfare for low/middle-income groups.
  - Services expand more than manufacturing following tax reforms.
- Policy takeaway:
  - Targeting matters: middle-class focused cuts can be both pro-growth and progressive; cuts concentrated at the top funded by indirect taxation can amplify growth in some settings but worsen inequality and shrink the middle class.

_Italic: Source — IMF Working Paper (wp17192), chapter/section "0.01 Elasticity of substitution between different types of consumptions being 1"._

---


_Source: https://www.imf.org/-/media/files/publications/wp/2017/wp17192.pdf_