The Impact of the Tax Cut and Jobs Act on the Spatial Distribution of High Productivity Households and Economic Welfare Daniele Coen-Pirani University of Pittsburgh Holger Sieg University of Pennsylvania and NBER December 17, 2018
Motivation The Tax Cut and Jobs Act (TCJA) of 2017 represents the most comprehensive reform of the federal tax code since 1986. We focus on its cap on state and local tax (SALT) deductions in federal tax returns. The cap is $10,000 for a married couple filing jointly. Main objective: clarify and quantify impact of TCJA on the spatial distribution and aggregate level of economic activity.
Fact 1: Importance of SALT deductions increases with AGI AGI Tax Itemized SALT/ τ f (SALT)/ AMT range returns deductions AGI AGI ($1,000) % % % % % < 75 74.63 15.69 2.52 0.38 0.04 75 100 8.63 53.59 4.29 1.07 0.61 100 200 12.25 75.99 6.04 1.69 3.44 200 500 3.62 93.56 7.35 2.42 59.38 500 1,000 0.58 93.19 7.43 2.94 46.36 > 1,000 0.29 91.48 7.51 2.97 19.20 Source: IRS data for the US (tax year 2015).
Fact 2: Uneven distribution of SALT across states SALT/AGI (%) AGI CA TX range ($1,000) < 200 5.30 2.51 200 500 10.00 3.79 500 1,000 11.08 2.89 > 1,000 12.83 1.56 Source: IRS data (tax year 2015)
Fact 3: Effect of SALT on top earners in CA relative to TX Federal income taxes/agi (%) AGI CA TX TX-CA range ($1,000) < 200 10.26 10.06-0.20 200 500 19.93 20.81 0.88 500 1,000 26.00 27.86 1.86 > 1,000 27.12 30.24 3.12
SALT deductions and average taxes if AGI > $500K average tax liability/agi (%) 24 26 28 30 32 AK TN TX SD WA FL NV WY ND LA NH IN NJ MS IL VA MI OH WV WI CT AL KS NMPA MO MA IA RI AZ KY MD COSCNC NE UT DE DC MN OK GA ID ME MT HIOR AR 0 5 10 15 SALT deduction/agi (%) VT NY CA The slope of the regression line is 0.184 (s.e. 0.062).
Should We Be Concerned About Relocations? Moretti and Wilson s (2017) empirical findings predict a significant relocation of top productivity households. We develop a new spatial equilibrium life-cycle model with exogenous and endogenous productive amenities. Empirical fact: high tax locations tend to have higher productive amenities. If productive amenities are mostly exogenous, the relocation of top types hurts high-tax location and benefits low-tax one, but has only small aggregate effects. If productive amenities are mostly endogenous aggregate earnings may fall by 1 4 percent.
Spatial Dynamic Model of Local Labor Markets OLG version of a Rosen-Roback model with two heterogeneous locations, j = S, D, that differ by: local taxes housing supply exogenous consumption amenities exogenous productive amenities
Productivity, Mobility, and Learning Households differ by age a and productivity type e and have earnings profiles, denoted by µ (a, e). Only high productivity types are mobile, but face mobility costs, denoted by κ (a, e). Young mobile household choose an initial location without knowing their exact productivity type. They learn their specific type e at end of a = 1 period. Keep on relocating in response to changes in income and local preference shocks.
earnings, $1,000 Lifetime Earnings Profiles 900 800 700 600 age 25 age 30 age 35 age 40 age 45 age 50 age 55 age 60 500 400 300 200 100 0 80 82 84 86 88 90 92 94 96 98 100 e Based on tabulations of SSA data by Guvenen et al (2016).
Productive Amenities and Agglomeration Effects Productive amenity depend on older top productivity types: x j = (1 α) x j + α E A n j (a, E) e=e a=2 Probability of type e in j: exp (x j e) f (e x j ) = z M exp (x jz) Agglomeration effects depend on the interaction between the mass of young top types, denoted by s j and productive amenity, x j : s j f (e x j )
Optimization Each period households choose consumption c and housing h in each period to maximize utility: U j (c, h) = (1 λ) ln c + λ ln h + ζ j + χ ln g j subject to a budget constraint: w j µ (a, e) = c + p j h + T (w j µ (a, e), p j h, c; a, e, j) where g j are local public goods and ζ j local consumption amenities. Mobile types also solve a dynamic discrete choice locational problem.
Production and Housing Supply Production function in location j is given by Y j = E e=1 a=1 A n j (a, e) µ (a, e). Housing supply function in location j is given by H j = Φ j p θ j j. Housing owned by absentee landowners.
Steady State Equilibrium 1. Individuals solve static and dynamic optimization. 2. Labor and housing markets clear in both locations. 3. Local governments budgets balance in both location. 4. Agglomeration effects are consistent with agents location choices.
Comments Incentives to relocate: Importance of idiosyncratic mobility shocks: σ Mobility cost function: κ (a, e) General equilibrium effects: Taste for public goods: χ Housing supply elasticities: θ j Share of endogenous productive amenities: α
The Tax Function We use linear approximations of the income tax functions to keep the model tractable. Tax functions depend on type (a, e) and location. State and federal income tax functions are approximated using TAXSIM. We also use IRS data for the combined San Jose-San Francisco-Oakland and Dallas-Fort Worth CSAs to attribute to each age-type combination a marital status for tax purposes, a number of dependents, non-salt deductions for itemizers, such as mortgage interest and charitable contributions, and SALT deductions.
Parameters A number of parameters can be set a-priori. Parameter Value Meaning a 8 one period 5 years β 0.99 5 discount factor λ 0.35 housing share χ 0.18 public goods θ S 2.50 housing supply elasticity in S θ D 10.20 housing supply elasticity in D τ p S 0.223 housing tax rate, location S τ p D 0.234 housing tax rate, location D τs c 0.024 sales tax rate, location S τd c 0.035 sales tax rate, location D We determine key parameters {x D, x S, κ, γ a, γ e, σ, } using a methods of moments algorithm subject to some normalizations.
Method of Moments Estimation MODEL DATA DESCRIPTION 1.72 1.72 ratio of housing rents S/D 1.68 1.80 Moretti-Wilson (2017) s spatial elasticity 18.52 24.83 migration rate of top type (e = E) 7.91 6.79 average migration rate (college) 4.57 4.45 ratio of migration rates young/old 1.06 1.09 % pop. with AGI > 500 in S 0.59 0.57 % pop. with AGI > 500 in D 4.09 4.50 % pop. with 200 <AGI< 500 in S 1.93 2.30 % pop. with 200 <AGI< 500 in D 10.81 9.58 % pop. with 100 <AGI< 200 in S 3.21 5.72 % pop. with 100 < AGI< 200 in D 55.39 55.39 % pop. in S
The Effect of the TCJA The policy experiment of interest is to change the federal tax code according to the TCJA, keeping constant local sales and property tax rates. We consider San Francisco and Dallas as the two metropolitan areas. We use TAXSIM to calibrate the income tax functions. TCJA increased the federal income tax rate faced by households in the top percentile residing in San Francisco by about 3 percentage points at ages 40 and above relative to Dallas. The effect on all other percentiles does not exceed 1 percentage point.
population measure of e=e Location of Top Types: Exogenous Productive Amenities 0.25 benchmark counterfactual 0.2 Location S 0.15 0.1 Location D 0.05 0 25 30 35 40 45 50 55 60 age
Effect of TCJA: Exogenous Productive Amenities (α = 0) Benchmark Counterfactual Spatial effects Population share S 1.06 0.06 ppt with AGI > 500 D 0.59 +0.05 ppt Population share S 4.10 0.06 ppt with AGI (200, 500) D 1.93 +0.06 ppt Public goods S 4.96 +0.58 % D 2.14 +3.37 % Total earnings S 28.19 1.91 % D 14.51 +3.55 % Aggregate effects Income 44.63 0.14 %
Mechanism with Endogenous Productive Amenities 1. Relocation of top types from S to D causes productive amenities to fall in S and increase in D: x S + x D = 0 2. Effect on the measure of new (a = 1) top types by location: s S x S < 0 s D x D > 0 3. Net effect is negative if s S > s D
population measure of e=e Location of Top Types: Endogenous Productive Amenities 0.25 Benchmark Counterfactual 0.2 Location S 0.15 0.1 0.05 Location D 0 25 30 35 40 45 50 55 60 age
Effect of TCJA: Endogenous Productive Amenities share of x S x D that is endogenous 25% 50% Spatial effects Population share S 0.09 ppt 0.17 ppt with AGI > 500 D +0.04 ppt 0.02 ppt Total earnings S 3.08 % 6.29 % D +2.49 % 0.94 % Aggregate effects Income 1.27 % 4.54 %
Conclusions With exogenous productive amenities, TCJA mostly reallocates economic activity away from high-tax to low-tax cities. With endogenous productive amenities, low-tax cities might gain less than what high-tax cities lose. We need more empirical work on sorting of and spillovers from top productivity types. It would be interesting to incorporate reactions of state and local governments.