Voting over Selfishly Optimal Income Tax Schedules with Tax-Driven - - PowerPoint PPT Presentation

Voting over Selfishly Optimal Income Tax Schedules with Tax-Driven Migrations

Darong Dai

Department of Economics Texas A&M University

Darong Dai (TAMU) Voting over Income Taxes 11/28/2017 1 / 27

Outline

1 Introduction 2 Model 3 The Voting Equilibrium 4 Three Characteristics of Equilibrium Tax Schedule 5 Identifying the Effect of Migrations on Equilibrium Taxes 6 Conclusion 7 The End

Darong Dai (TAMU) Voting over Income Taxes 11/28/2017 2 / 27

Introduction

Motivation

Labor income tax design: work incentive and participation incentive. Large income tax-induced mobility elasticities:

Highly paid football players: Kleven et al. (2013, AER). High income foreigners in Danish: Kleven et al. (2014, QJE). Foreign superstar inventors (employees): Akcigit et al. (2016, AER).

Geographic mobility in designing redistributive taxation:

Conventional wisdom (Stigler, 1957): limits redistribution. The classic paper: Mirrlees (1971, RES). Mirrlees (1982), Simula and Trannoy (2010, 2012), Piketty and Saez (2013) and Lehmann et al. (2014). They focus on the normative approach. Lack of research from the positive (political-economy) approach.

Darong Dai (TAMU) Voting over Income Taxes 11/28/2017 3 / 27

Introduction

The Combination of Labor Mobility and Majority Voting

Political process:

A direct democracy with citizen candidates. Pairwise majority voting. Each worker has one vote.

Voice: Equilibrium tax schedule is selected by majority voting. Exit: Workers can move between alternative jurisdictions. A real example: Switzerland (26 cantons), top tax rate: 22.5%-46%. This combination generates a complex interaction whereby:

The taxation policies chosen determine whom they attract. Whom they attract determine their choices of taxation policies.

Darong Dai (TAMU) Voting over Income Taxes 11/28/2017 4 / 27

Introduction

Question

How would labor mobility affect income redistribution in a voting equilibrium? To my knowledge, the answer is not yet well established. To answer this question:

Establish a voting equilibrium. Compare to autarky (without migrations), qualitatively and quantitatively.

Darong Dai (TAMU) Voting over Income Taxes 11/28/2017 5 / 27

Introduction

Related Literature

Combine migrations and voting:

Cremer and Pestieau (1998), Hindriks (2001), Hamilton and Pestieau (2005), Brett (2016).

Selfishly optimal nonlinear taxation determined by the majority rule:

R¨

• ell (2012), Bohn and Stuart (2013), Brett and Weymark (2016,

2017).

How the change of skill distribution affects equilibrium tax:

Leite-Monteiro (1997), Hamilton and Pestieau (2005), Brett and Weymark (2011), Lehmann et al. (2014).

Darong Dai (TAMU) Voting over Income Taxes 11/28/2017 6 / 27

Model

Environment

Two jurisdictions: A and B, not necessarily symmetry. Both jurisdictions adopt majority voting. Workers differ in

Skill/labor productivity, w ∈ [w, w] and w > 0. Migration cost/foot-voting capability, m ∈ R+.

Information structure:

Distributions F(w) = w

w f (t)dt and G(m|w) =

m

0 g(x|w)dx are

common knowledge. Values of w and m are the private information of each worker.

Quasilinear-in-consumption preferences (Gruber and Saez, 2002). Labor markets are perfectly competitive. Labor income taxes are levied according to the residence principle.

Darong Dai (TAMU) Voting over Income Taxes 11/28/2017 7 / 27

Model

Individual Choices: Behavior Responses

For a worker of type (w, m) born in jurisdiction A: How much to work and consume?: U(w) ≡ max

{c,l} c − h(l)

s.t. c = y − T(y) and y = wl.

FOC: T ′(y(w)) ≡ τ(w) = 1 − 1

w h′ y w

• , i.e., MTR = 1 − MRSc,y.

Allowed: τ(w) < 0 for some w.

Where to work and consume?: not to migrate ⇐ ⇒ U(w)

stay

≥ U−(w) − m

• move

, (participation constraint) following Lehmann et al. (2014, QJE).

Darong Dai (TAMU) Voting over Income Taxes 11/28/2017 8 / 27

Model

Migration Elasticity

The ex post density of residents of skill w in jurisdiction A: φ(∆(w); w) ≡          f (w) + G−(∆(w)|w)f−(w)n−

• inflow

for ∆(w) ≥ 0, f (w) − G(−∆(w)|w)f (w)

• utflow

for ∆(w) ≤ 0 with ∆(w) ≡ U(w) − U−(w). Ex post skill density: ˜ f (w) ≡ φ(∆(w); w). Following Lehmann et al. (2014, QJE), migration elasticity is: θ(∆(w); w) ≡ ∂˜ f (w) ∂∆(w) c(w) ˜ f (w) ≡ ˜ θ(w).

Darong Dai (TAMU) Voting over Income Taxes 11/28/2017 9 / 27

Model

Government Budget Constraint

Government budget constraint (GBC) under pure redistribution: w

w

T(y(w))˜ f (w)dw ≥ 0. Ex post tax base, denoted by Γ(w, w) ≡ w

w

˜ f (w)dw, is endogenous.

Darong Dai (TAMU) Voting over Income Taxes 11/28/2017 10 / 27

Model

Selfishly Optimal Nonlinear Tax Schedules

By Taxation Principle, a worker of type k ∈ [w, w] proposes an income tax schedule by solving the problem: max

{c(w),y(w)}w∈[w,w]

U(k) s.t. U(k) = c(k) − h y(k) k

• (IU);

w

w

[y(w) − c(w)]˜ f (w)dw ≥ 0 (GBC); U′(w) = h′ y(w) w y(w) w2 , ∀w ∈ [w, w] (FOIC); y′(w) ≥ 0, ∀w ∈ [w, w] (SOIC).

Darong Dai (TAMU) Voting over Income Taxes 11/28/2017 11 / 27

The Voting Equilibrium

Theorem By imposing pairwise majority-voting rule to the continuum of selfishly

• ptimal income tax schedules, the one for the median skill type turns out

to be a Condorcet winner. By establishing weak single-peakedness of preferences (inverted U-shape): U(w, w) ≥ U(w, k1) ≥ U(w, k2), ∀

proposers’ type

• w > k1 > k2

U(w, w) ≥ U(w, k1) ≥ U(w, k2), ∀w < k1 < k2 then appealing to Black’s Median Voter Theorem (1948, JPE). Recent empirical evidences: Agranov and Palfrey (2015), Corneo and Neher (2015) and Gr¨ undler and K¨

• llner (2016).

Darong Dai (TAMU) Voting over Income Taxes 11/28/2017 12 / 27

Three Characteristics of Equilibrium Tax Schedule

The First Characteristic

Proposition It coincides with the maximax tax schedule for types below the median skill level and coincides with the maximin tax schedule for types above the median skill level. It redistributes incomes from the poor and rich toward the middle. Intuition: Median voter is selfish. Autarky equilibrium: Brett and Weymark (2017, GEB). Director’s Law (Stigler, 1970): labor mobility and inter-jurisdictional tax competition.

Darong Dai (TAMU) Voting over Income Taxes 11/28/2017 13 / 27

Three Characteristics of Equilibrium Tax Schedule

The Second Characteristic

Proposition For incomes below the median skill type, marginal tax rates are negative. For incomes above the median skill level, marginal tax rates are positive if their migration elasticities are endogenously bounded above (this upper bound is in general different for different types). Intuition: The tradeoff between:

Maximizing resources extracted from other types. Maximizing resources available for extraction.

Tax higher skills because:

High wage rate implies a high opportunity cost of leisure.

Subsidize lower skills because:

Low wage rate implies a low opportunity cost of leisure. Taxing them strengthens their motive to mimic high types.

Darong Dai (TAMU) Voting over Income Taxes 11/28/2017 14 / 27

Three Characteristics of Equilibrium Tax Schedule

The Third Characteristic

Proposition Truth-telling implies that workers with heterogenous skills may face the same marginal tax rate. w y

yR∗(·) yM∗(·)

w w k yM∗(w) wη wα wβ wγ yR∗(w)

Figure: Income schedule with three bridges.

Darong Dai (TAMU) Voting over Income Taxes 11/28/2017 15 / 27

Three Characteristics of Equilibrium Tax Schedule

The Third Characteristic (Cont’d)

w y

yR∗(·) yM∗(·)

w w k yM∗(w) wα wβ wγ yR∗(w) w y

yR∗(·) yM∗(·)

w w k yM∗(w) wη wα wβ yR∗(w)

Figure: Income schedule with two bridges and an upward discontinuity.

Darong Dai (TAMU) Voting over Income Taxes 11/28/2017 16 / 27

Identifying the Effect of Migrations on Equilibrium Taxes

Qualitative Characterization: Case I

Proposition Suppose ΘM(w) < ˜ θ(w) < ΘMR(w) for ∀w ∈ (w, wm] and ΘR(w) < ˜ θ(w) < ΘMR(w) for ∀w ∈ (wm, w), then we have: (i) If ˜ wm ≥ wm, then distortions and redistribution for ∀w ∈ (w, w) are smaller than in autarky. (ii) If ˜ wm < wm, then (i) still holds for ∀w ∈ (w, ˜ wm] ∪ (wm, w), whereas redistribution and distortions for ∀w ∈ (˜ wm, wm] turn out to be greater than in autarky.

Darong Dai (TAMU) Voting over Income Taxes 11/28/2017 17 / 27

Identifying the Effect of Migrations on Equilibrium Taxes

Graphic Illustration: Case I

Black ones represent the autarky equilibrium. w y

ˆ yR(·) ˆ yM(·) yM(·) yR(·)

w w wm˜ wm w y

ˆ yR(·) ˆ yM(·) yM(·) yR(·)

w w wm ˜ wm

Darong Dai (TAMU) Voting over Income Taxes 11/28/2017 18 / 27

Identifying the Effect of Migrations on Equilibrium Taxes

Contribution and Intuition: Case I

Contribution: Derive endogenous lower bounds ΘM(w) and ΘR(w). Intuition: For high skills:

Large migration elasticity implies strong threat. Lower MTRs to prevent the shrink of tax base.

For low skills:

Lower MTRs on high skills provide more incentives to mimic high types. High migration possibility strengthens this. More transfers to prevent mimicking (more information rents).

Types between median types:

The income class they belong to totally changes! From the group receiving transfers to that paying taxes, or vice versa.

Darong Dai (TAMU) Voting over Income Taxes 11/28/2017 19 / 27

Identifying the Effect of Migrations on Equilibrium Taxes

Qualitative Characterization: Case II

Proposition Suppose ˜ θ(w) < ΘM(w),

inflow

• ˜

f (w)/Γ(w, w) > f (w)/F(w) for ∀w ∈ (w, wm], and ˜ θ(w) < ΘR(w),

• utflow
• Γ(w, w) − Γ(w, w)

˜ f (w) > 1 − F(w) f (w) for ∀w ∈ (wm, w), then we have: (i) If ˜ wm ≤ wm, then distortions and redistribution for ∀w ∈ (w, w) are greater than in autarky. (ii) If ˜ wm > wm, then (i) still holds for ∀w ∈ (w, wm] ∪ (˜ wm, w), whereas redistribution and distortions for ∀w ∈ (wm, ˜ wm] turn out to be smaller than in autarky.

Darong Dai (TAMU) Voting over Income Taxes 11/28/2017 20 / 27

Identifying the Effect of Migrations on Equilibrium Taxes

Graphic Illustration: Case II

Black ones represent the autarky equilibrium. w y

yR(·) yM(·) ˆ yM(·) ˆ yR(·)

w w wm ˜ wm w y

yR(·) yM(·) ˆ yM(·) ˆ yR(·)

w w wm˜ wm

Darong Dai (TAMU) Voting over Income Taxes 11/28/2017 21 / 27

Identifying the Effect of Migrations on Equilibrium Taxes

Contribution and Intuition: Case II

Contribution: Derive endogenous upper bounds ΘM(w) and ΘR(w). Reasonable counterexamples to the conventional wisdom. Intuition: For high skills:

Small migration elasticity implies weak threat. Outflow implies the shrink of tax base (brain drain) relative to autarky. Higher MTRs to make up for the shrink of tax base.

For low skills:

Higher MTRs on high skills weaken incentives to mimic high types. Low migration possibility strengthens this. Less transfers to prevent mimicking (less information rents). Inflow implies more people to share the transfers.

Types between median types: the same as Case I.

Darong Dai (TAMU) Voting over Income Taxes 11/28/2017 22 / 27

Identifying the Effect of Migrations on Equilibrium Taxes

Numerical Experiments: Functional Forms

Pareto distribution (Saez (2001) and Atkinson et al. (2011)). As in Diamond (1998), h(l) takes the form h(l) = l1+1/ε/(1 + 1/ε). The difference of MTRs is δMTR ≈

ˆ τ R(w)=

• 1

a

• 1 + 1

ε

• 1 + 1

a

• 1 + 1

ε

−

τ R(w)≈

• 1

˜ a

• 1 + 1

ε

• − ˜

θ(w)ξ(w) 1 + 1

˜ a

• 1 + 1

ε

• for top incomes, with ˆ

τ R(w) denoting autarky-equilibrium tax rate. An approximation (quasilinear-in-consumption preferences) of the elasticity of utility with respect to pre-tax income: ξ(w) ≡ ∂U(w) ∂y(w) y(w) c(w).

Darong Dai (TAMU) Voting over Income Taxes 11/28/2017 23 / 27

Identifying the Effect of Migrations on Equilibrium Taxes

Numerical Experiments: Parameter Values

Table: Parameter Values (U.S.)

Value Description Source ˜ θ 0.25 Global migration elasticity Lehmann et al. (2014) ε [0.12, 0.4] Labor-supply elasticity Saez et al. (2012) ˜ a 1.5 Pareto index Diamond & Saez (2011) τ top 42.5% Real top tax rate Diamond & Saez (2011) Case (a): ˜ a = 1.5, ε = 0.25, ˜ θ = 0.25, ξ = 5.96 ⇒ τ R = τ top. Case (b): ˜ a = 1.5, ε = 0.40, ˜ θ = 0.25, ξ = 3.64 ⇒ τ R = τ top.

Darong Dai (TAMU) Voting over Income Taxes 11/28/2017 24 / 27

Identifying the Effect of Migrations on Equilibrium Taxes

Numerical Experiments: Results

Holding everything else constant. How large the difference can be made by migrations?

Table: The difference of MTRs (%) under case (a)

a 1.35 1.4 1.5 1.7 2.0 3.0 δMTR 36.2 35.6 34.4 32.1 28.9 20.0

Table: The difference of MTRs (%) under case (b)

a 1.35 1.4 1.5 1.7 2.0 3.0 δMTR 29.6 28.9 27.5 24.8 21.1 11.4

Darong Dai (TAMU) Voting over Income Taxes 11/28/2017 25 / 27

Conclusion

Implications and Insights

Emphasize the interaction between tax rate and tax base. Avoid the emergence of extreme extraction:

No dictator: majority voting (democracy). No tyranny of middle class: informational constraint and foot-voting.

Achieve certain balance between efficiency and equity:

Tax high skills while subsidize low skills. Without resorting to a benevolent social planner. Without resorting to other-regarding social preferences. The key is institutional arrangement.

No laissez faire: democracy + skill heterogeneity. A possible low-efficiency trap: high redistribution ⇒ brain drain ⇒ poorer median voter ⇒ higher redistribution ⇒ more brain drain.

Darong Dai (TAMU) Voting over Income Taxes 11/28/2017 26 / 27

The End

Thank You!

Darong Dai (TAMU) Voting over Income Taxes 11/28/2017 27 / 27