Fiscal Federalism: Efficient or unfair?

Fiscal Federalism: Efficient or unfair?

Debra Hevenstone with Ben Jann

University of Bern

Rational Choice Seminar, VeniceNovember 27, 2012

Debra Hevenstone with Ben Jann (University of Bern)Fiscal Federalism: Efficient or unfair?Rational Choice Seminar, VeniceNovember 27, 2012 1

/ 65

Background

Fiscal Federalism

(OECD data for the most recent year available 1997-2000)

010203040506070%

Bel

gium

Irel

and

UK

Por

tuga

lN

ethe

rland

sN

. Zea

land

Fra

nce

Italy

Spa

inN

orw

ayA

ustr

iaIc

elan

dS

wed

enF

inla

ndG

erm

any

Aus

tral

iaD

enm

ark

US

Sw

itzer

land

Can

ada

Sub-national revenues (% of total)Sub-national expenditures (% of total)


/ 65

Background Literature

Literature on Effects

Efficient public goods (“Vote with your feet!”)I Tailored to local preferences

Unfair public goodsI Economic segregationI Low taxes for the richI Fewer public goods for the poor

Public expendituresI Fiscal constraint (with Leviathan government)I Fiscal indiscipline (with vertical equalization)

(Tiebout, 1956; Musgrave, 1959; Grueber, 2006; Rodden, 2003; Brennan &Buchanan, 1980)


/ 65


Dynamics

Simple dynamics might determine effects (Hirschman 1970)I Exit: (Public) Goods improve when consumers exitI Voice: (Public) Goods improve when consumers contributeI Loyalty: the ratio of exit & voice

Plus more complex dynamicsI Individual preferences for wealthy neighborsI Politicians competing to attract the richI The wealthy like less public goodsI Pork barrel politicsI Veto playersI ...


/ 65


Project

QuestionI Can exit and voice generate characteristics of federal states?

F Lower less progressive taxes in rich areasF Lower spending in devolved statesF Rich want/consume fewer public goods

I Can policy impact whether public goods are fair or efficient?

Method: Agent based modeling (microsimulation)I Feedback between exit and voiceI Interaction between micro behaviors and macro conditions


/ 65


Similar Work: Kollman, 1997

How can politics yield efficient public goods allocation?

Match agents to geography via moving and voting on public goodsWhy? Tiebout matching can yield non-optimal equilibrium

I Political platforms1 Random policy permutations2 Random permutations with sample polling3 Genetic algorithm

I Voting1 Majority rule by issue2 Agents vote for single party platform3 Agents rank party platforms

FindingsI Party platforms yield more efficient public goods under federalism

Sorting can only yield gains


/ 65


Similar Work: Penn, 2003

Does secession encourage inequality?

Match people to political units through moving and votingPoint? Progressive taxation means match is also about fairness

I Voting1 Tax rates2 Secession

I Politics1 Local tax discretion2 Majority vs supermajority for secession

FindingsI More discretion & easier secession→ lower utility & more inequality

Sorting can only yield losses


/ 65

The Research Project Model Design

Model Design INITIALIZATION

1. Landscape: 100* 100 grid (10,000 parcels)2. Jurisdictions: 16 groups of 625 parcels3. Households: 9000 households with: incomes preferences for public goods

EXIT (MOVE)1. Household: Pick a random vacant parcel O�er bid that marginally improving utility2. Parcels of Land: Look through list of bids Accept highest bidder

VOICE (VOTE)1. Jurisdictions Propose high/low change in maximum tax rate Propose high/low change in tax progression People vote Local taxes are set2. Federal government Propose high/low change in maximum tax rate Propose high/low change in tax progression People vote Federal taxes are set Equalization grants processed


/ 65


Model Design

jurisdiction border

rich households

poorhouseholds


/ 65


Model Details

Agent CharacteristicsI IncomeI Public Goods PreferenceI Utility

F Cobb DouglasF Constant returns to scaleF Public and private goods consumption

Individual DynamicsI VotingI Moving

F Pick random vacant plotsF Offer a price that would increase utilityF Highest bidder moves

MacroI Tax FormI Federal Equalization


/ 65


Assumed Income DistributionCurrently using a lognormal

Simulated Distribution (lognormal µ =11,σ=1)Swiss Income Distribution (2009)

Testing: Pareto, Weibull, generalized beta, gamma ...


/ 65


Public Goods PreferencesRandomly assigned from a normal distribution (µ = .2, σ = .05)Private goods preference: (1- public goods preference)

local 0 to 19% of incomefederal 0 to 21% of incometotal 17 to 21% of income


/ 65


Key Point

Income and preferences are set exogenouslyAll other critical variables are set endogenously

I Results depend on income and preference distributions


/ 65


Assumed tax functional form

0

5

10

15

20

25

30%

0

Zurich

ZugGeneva

1million

aver

age

tax

Average tax rates by income (couple with no children)

income

tax = S(1− e−y/k )

y income

S parameter for tax level

k parameter for tax progression


/ 65


Assumed tax functional form

0

5

10

15

20

25

30%

0

Zurich

ZugGeneva

1million CHF

aver

age

tax

Average tax rates by income (couple with no children)

Geneva

05

101520253035%

data

Zurich Zug

k = 270,270S=.29p.c. income = 62,839

income

05

101520253035%

05

101520253035%

0 1 mil. 0 1 mil. 0 1 mil.k = 232,558S=.23p.c. income = 68,804

k = 192,308S=.12p.c. income = 93,753

tax = S(1− e−y/k )

y income

S parameter for tax level

k parameter for tax progression


/ 65


Voting and Voice

Government proposes a random + & − deviation inI Maximum tax rateI Progression

People vote1 “Average Man Voice”

A median voter model

2 “Rich Voice”Maximize constituent utility


/ 65


Swiss Case: VotingMaximum tax rate (“Steuerfuss”)

I Can be changed by people or politiciansProgression

I Primarily through politiciansI Though deductions can be set by people


/ 65


Model Design: Equalization

Federal RevenuesLocal Revenues

high taxcapacity area

low taxcapacity area

no equalization

verticalequalization

+ =

+ =

+ =

+ =

horizontalequalization (F)

= =

horizontalequalization (NF)

HorizontalI Between jurisdictionsI Based on the difference

between tax capacities

VerticalI From central governmentI Based on the ratio between

tax capacities


/ 65

Experiments

Experiments

Voice: Rich or Average

None

Horizontal w/ fed

Vertical w/ fed

Horizontal no fed

EqualizationParameter

0

Redistribution

0 .1 .2 .3 .4 .5 .6, .7.8 .9 1.0

0 .1 .2 .3 .4 .5 .6, .7.8 .9 1.0

0 .1 .2 .3 .4 .5 .6, .7.8 .9 1.0

68 settings, 10 runs per setting, 5000 time steps


/ 65

Emergence

Simple model, realistic outcomes

Model AssumptionsRealistic Outcomes (emergence)Unnecessary Assumptions


/ 65

Emergence Segregation

Realistic Outcomes: Segregation

Strictly local taxationno redistribution

Strictly local taxation horizontal equalization =.1

RICH VOICE

EQUAL VOICE

Strictly local taxationno redistribution

Strictly local taxation horizontal equalization =.1


/ 65

Emergence Constrained Spending

Realistic Outcomes: Federalism constrains spending


/ 65

Emergence Taxation and Public Goods

Realistic Outcomes: Rich have lower flatter taxes

Tax curves for richest and poorest jurisdictions, across all experiments


/ 65


Realistic Outcomes: Rich have fewer public goods

Public goods by jurisdiction’s income rank, across all experiments


/ 65


Realistic Outcomes: Applied taxes are regressive


/ 65


Simple model, realistic outcomes

Model AssumptionsI Dynamics: Move & VoteI Parameters: Income & Preference Distributions

Realistic Outcomes (emergence)I Economic SegregationI Rich prefer lower flatter taxes, few public goodsI Devolved federalism constrains spending

Unnecessary AssumptionsI Preferences for wealthy neighborsI The wealthy like less public goodsI Veto-playersI Pork barrel politics


/ 65

Three Interesting Cases

Three interesting casesEqual Voice Rich Voice

Pure Federalism Ayn Rand ParadiseHorizontal (F) Optimal Hybrid (CH)Vertical Ayn Rand NightmareHorizontal (NF)


/ 65

Three Interesting Cases

Segregation

Economic Segregation Preference Segregation


/ 65

Three Interesting Cases Ayn Rand Paradise

Ayn Rand Paradise: Outline

TaxesPopulationPublic GoodsUtilityEfficiency


/ 65


Ayn Rand Paradise: Taxes


/ 65


Ayn Rand Paradise: Population


/ 65


Ayn Rand Paradise: Public goods


/ 65


Ayn Rand Paradise: Utility


/ 65


Ayn Rand Paradise: Efficiency


/ 65


Ayn Rand Paradise: SummarySegregation

I High by income & preferencesTaxes

I About equal ratesPopulation

I People want to live near the richPublic Goods

I Rich have somewhat moreUtility

I Rich jurisdictions have high utilities, the poor middlingEfficiency

I Some gains

The rich self-segregate, provide their own public goods, and maintainhigh utilities


/ 65

Three Interesting Cases Ayn Rand Nightmare

Ayn Rand Nightmare: Overview



/ 65


Ayn Rand Nightmare: Taxes


/ 65


Ayn Rand Nightmare: Population


/ 65


Ayn Rand Nightmare: Public goods


/ 65


Ayn Rand Nightmare: Utility

Policy inverted the correlation between income and utilityIt is a plague to be rich or to live near a rich person!


/ 65


Ayn Rand Nightmare: Efficiency


/ 65


Ayn Rand Nightmare: Summary

SegregationI Neither by income nor preference

PopulationI Agents run away from the rich

TaxesI Largely federal and mildly progressive

Public GoodsI About equal, and sporadic

UtilityI Good for the poorest, bad for the rest

EfficiencyI No gains from preference sorting

The tyranny of the majority


/ 65

Three Interesting Cases Optimal Hybrid

Optimal Hybrid: Overview



/ 65


Optimal Hybrid: Taxes


/ 65


Optimal Hybrid: Population


/ 65


Optimal Hybrid: Public goods


/ 65


Optimal Hybrid: Utility

Policy maximizes utility in all jurisdictionsThe rich have higher utilities at all equalization levels


/ 65


Optimal Hybrid: Efficiency


/ 65


Optimal Hybrid: Summary

TaxesI Primarily federal, local are regressive

PopulationI Shifts with equalization from rich to poor areas

Public GoodsI Rich consume less at all levels of equalization

UtilityI All groups achieve optimal utility

EfficiencyI Significant gains from preference sorting

A moderately segregated society with regressive taxation and efficientpublic goods provision, yielding Pareto and Rawlsian optimal utilities


/ 65

Conclusion

Conclusion

A simple model of using exit and voice might explainI Economic segregationI Constrained spending under federalismI Lower flatter taxes in rich areasI Less public goods consumption among the wealthyI Applied regressive taxation

Policy can generateI Tyranny of the majorityI Wealthy flightI Efficient and (somewhat) fair public goods provision


/ 65

Conclusion

Thank You


/ 65

Appendix Caveats

Appendix: Caveats

1 Model AssumptionsI Results stem from the exogenous income and preferenceI Different distributions might yield different results

2 Real WorldI “Loyalty”, or a lack of mobility, would temper these effectsI People are not that rational


/ 65

Appendix Appendix: Future Work

Future Work

Framing current results, targeting an audienceVaried income distributions and preference distributionsModel validation with empirical data


/ 65

Appendix Methods Appendix

Method Appendix: Income

Currently using Testingenormal(µ=11,σ2=.1) x ∗ −ln(uniform(0to1))

Empirical literatureI Pareto- fits the high end betterI Lognormal - fits the low end betterI Weibull, hybrid exponential decay with power decay, (generalized)

beta, gamma (Singh & Maddala, Nirei, Thorow, Salem & Mount, McDonald)


/ 65


Method Appendix: Cobb Douglas Utility

u` =

(gj

nj+

1nj

∑i

(yi ∗ ti)

)α`︸︷︷︸

public goods

∗ (y` ∗ (1− t`)− h`)1−α`︸︷︷︸private goods

` householdy` income of household`t` tax of household`h` housing cost of household `i index of householdsj jurisdictiongj grant to jurisdiction jnj number of households in jurisdiction jα` preference for public goods of household `1− α` preference for private goods of household `


/ 65


Method Moving

Pick an empty parcelCalculate price that would yield current utilityOffer a bid of 1 franc less

h2 = y − t2 −

((p1p2

)α(y − t1 − h1

)1−α) 1

1−α

− 1

Parcel offered to the highest bidder at the end of the roundOnly positive bids are takenWith one jurisdiction, all bids are -1 and no one moves


/ 65


Method Appendix: Commutability

Zurich

Commutable Distances

11 cantons wtihin commuting distance of Zurich

Zug


/ 65


Method Appendix: Equalization formulae

baseline horizontal NF horizontal vertical

rj rj + θhNj (x̄ − xj )︸︷︷︸horizontal grant

NjN Rf + rj + θhNj (x̄ − xj )︸︷︷︸

horizontal grant

rj + Rf Nj (1/cj )θv∑

i Ni (1/ci )θv︸︷︷︸vertical grant

Nj population in jN total populationRf total federal revenuerj revenue collected in jθh horizontal redistribution parameterxj jurisdiction’s per capita tax capacityx̄ national per capita tax capacityθv vertical redistribution parametercj jurisdiction j ’s relative per capita revenue (cj =

xjx̄j

)


/ 65

Appendix Convergence

Appendix: Convergence (rich voice, horizontal, no federal, .3)

Federal level convergence

economic segregation tax progression maximum tax

Jurisdiction level convergence

average income for average max tax for average maximum taxrich, poor, middle rich, poor,middle by preferences(average across 10 runs, for 1 experimental setting)Debra Hevenstone with Ben Jann (University of Bern)Fiscal Federalism: Efficient or unfair?

Rational Choice Seminar, VeniceNovember 27, 2012 58/ 65

Appendix Housing Prices

Appendix: Housing Prices Problem 1, Convergence

Prices converge very slowly if at allA simulation of 30,000 ticks (vs 5,000)(equal voice, horizontal redistribution, equalization 0)


/ 65


Appendix: Housing Prices Problem 2, Housing Pricesand Jurisdiction Attractiveness

Ayn Rand ParadiseI Rich jurisdictions pay less for housingI When they have more control and rich areas are more attractive


/ 65


Appendix: Housing Prices Problem 2, cont.

Ayn Rand NightmareI Rich pay more for housingI When they have less control and their areas are unattractive!


/ 65


Appendix: Housing Prices Problem 3

Equalization increases housing prices for the rich!Even though equalization makes these areas less attractive for theaverage agent


/ 65


Appendix: Housing Prices

Optimal ScenarioI Rich pay same for housing at low equalization, more at high

equalization


/ 65

Appendix Appendix: Population swaps and equalization

Appendix: Population swaps as equalization shifts


/ 65

Appendix Appendix: Population swaps and equalization

Why population swaps as equalization shifts: Utilitycurves


/ 65

Fiscal Federalism: Efficient or unfair?

Documents

Transcript of Fiscal Federalism: Efficient or unfair?