Subnational Redistribution and Economic Growth

1

• Subnational Redistribution and Economic Growth
• Sept. 06 Version
• Howard Chernick and Paul Sturm
• Hunter College and the Graduate Center, CUNY
• NYC Dept. of Finance

2
Questions?

• IS Sub-National Redistribution Inefficient
• Does a relatively progressive tax structure
  and/or a relatively pro-poor expenditure pattern
  affect State Economic Growth?

3
HYPOTHESES

• Standard HYP. Redistribution as a Negative
  Fiscal Amenity
• In more redistributive states Negative
  fiscal residual for high-income individuals
• B(yhi) T(yhi) lt 0 Negative Migration
  Incentive, Negative Labor Supply Effect
• Positive fiscal residual for low-skilled
  individuals B(ylo) T(ylo) gt 0 Positive
  migration incentive, Negative Labor Supply
  Effect.

4
Alternative Hypotheses

• Preferences for Redistribution Vary Sorting by
  Preferences. Variation in Redistribution is
  therefore welfare enhancing. Hence no effect on
  Growth.
• Redistribution as Investment in Human Capital
  Education, Health, Housing. Positive Return
  positive effect on growth

5
Factors Affecting Sub-National Growth

• 1) National Economic Growth. Elast. Of Growth
  w.r.t. nat. growth 1 (Blanchard and Katz)
• 2) Long-run Convergence in productivity, per
  capita income. (Barro and Sala-Martin) but
  convergence is slow.
• 3) Density Effect. Differential rates of
  productivity growth in large cities. (Hall and
  Ciconne, 1996) Influence on statewide economic
  growth.

6
Factors Affecting Sub-Nat. Growth (cont)

•  4) Importance of State or Region specific
  shocks, mainly on the Demand Side differences in
  industrial structures.
•  Shocks lead to wide swings in unemployment,
  migration among the unemployed. Long-run
  persistence of shocks to jobs,population, and
  aggregate income

7
ANNUAL PER CAPITA INCOME GROWTHAverage of 5 Year
Range 1977 to 1982
8
ANNUAL PER CAPITA INCOME GROWTHAverage of 5 Year
Range 1985 to 1990
9
ANNUAL PER CAPITA INCOME GROWTHAverage of 5 Year
Range 1991 to 1996
10
ANNUAL PER CAPITA INCOME GROWTHAverage of 5 Year
Range 1995 to 2000
11
ANNUAL PER CAPITA INCOME GROWTHAverage of 20
Year Range 1977 to 1997
12
What do the Maps Tell Us

• Differences across 5 yr periods in regional
  patterns of growth.
• Long-run growth. Convergence in the south.
  Northeast grows rapidly as well.
• Long-run range much smaller than SR range
• Long-run aggreg. Growth persistent differences.

13
Growth Effect Labor Supply depends on p, the
fiscal residual delta is an agglomeration
effect.
14
Comp. Diffs. And PO Redistribution

• Indirect Utility Fctn of high and low skilled
  worker in rents, gross wages, taxes, and
  altruistic component
• VH VvH(rH,wH,tH),VL V0,H (2a)
• VL(rL,wL,tLTR) V0,L (2b)
• Relative Tax Rates Determined by State Budget
• Constraint
• tH tH (G,TR,tL) (4a)
• tL tL (G,TR,tH) (4b)

15
Comp. Statics

• MVH/MwH -MVH/MtH
• MVH/MTR)(MTR/MtH) (7)
• MVL/MwL -MVL/MtL MVL/MTR/ (1
  MwL/MTR) (9)
• Hence the compensating effect on gross wages of
  a tax shift or a change in transfers depends on
  marginal utility to the donor, and the marginal
  productivity to the recipients, from additional
  transfers.

16
Empirical Evidence

• Labor Supply Effect.
•  a)   Capitalization of taxes in gross wages.
  Negative effect on growth, through increase in
  costs, decrease in productivity. Note conflict
  if demand inelastic
• Wallace (1993), - 25 of cases. Not more in
  high-skilled occupations. (where supply is more
  elastic)
• F-V (1998) full capitalization of changes in tax
  incidence.  
• Chernick (2004) no evidence of tax
  capitalization in aggregate data.

17
Migration, Labor Supply, Income Effects

• Taxes Pema (2005) small negative effect of
  income and sales taxes increases with education.
  Similar to wage differential effect
• Welfare
• Some Migration Effect (Gelbach, 2004), McKinnish
  (2005) migration low among those at risk, hence
  overall flows are small

18
Labor Supply, Income Effects

• Labor supply effects of redistribution. In-state
  effects bigger than migration effects. Neumark
  and Powers (2005), Meyer (2000)
• Effect of Welfare Spending on Personal Income
  Negative Effect Helms (1985) Welfare as Excluded
  Category

19
Expenditure Categories

• 1) Public goods (1/3 of spending police, fire,
  sanitation, transportation, gen. Admin., interest
  on debt
• 2) Redistributive (1/4 of spending) Pub.
  Assistance, Medicaid, child welfare services
• 3) Education (over 1/3 of spending) primary and
  secondary, higher ed

20
Measure of Expend. Progressivity

• 1) Welfare Spending per cap., Maximum welfare
  benefit
• 2) State Govt Share of public school spending
  (correlation with Inequality of Spending -.26)
• Higher Education Spending from Public Revenues
  (non-tuition sources)

21
Growth Model

• Growtht1 a0 a1(BURDENt0) a2(NEIGHBURDENt0)
  a3(EXPDISTR) a4(NEIGHEXPDISTR) a5(RELATIVE
  INCOMEt0) a6(DEMOGRt0) a7(BIGCITY)
  a8(REGION/YEAR) error term
• 5 and 10 year average growth in per capita income
  as function of Distributional Characteristics
• Estimate Tax Effect, controlling for Expend.
  Incidence

22
Hyp. For Fiscal Vbles.Variable
Expected EffectTax Progressivity
Negative Neighbor Progr.
PositiveSelf. Spend/Max Ben NegativeState Ed
Share ?Higher Ed Spend
Positive
23
Are Fiscal Vbles. Endogenous

• Welfare If growth reduces need lowers
  expenditures, negative welfare coefficient might
  be mistakenly interpreted as causing a reduction
  in the rate of economic growth.
• If growth increases redistribution (altruism as
  normal good) positive welf coeff. Would not imply
  a causal effect.
• Tests a) regressed growth on lagged growth no
  effect b) regressed fiscal vbles on lagged
  growth. Results All Insign, except neg. effect
  on welfare spending of aggreg. 10 year growth.
• Therefore use OLS. (IV estimates not different)
  

24
Data and Variables.

• Model Estimated with Pooled Cross Section of
  Continental U.S. States years are 77, 85, 91,
  and 95.
• Tax Data from Phares (1977), and Citizens for
  Tax Justice. (See Chernick, 2005 for
  description.
• BURD1, BURD3, BURD5 state quintile specific
  tax burden, for repres. Family, at that quintile,
  net of federal deductibility.
• Neighbor pop. weighted avg.of geogr. Neighbor
  tax progressivity
• Progressivity BURD5/BURD1
• Expenditure Variables. U.S. Census of Govts,
  various years.

25
Estimation Results

• Table 5. Basic Specification. (All specs.
  Include region, year, and year by region
  indicator variables.)
• Tax Effects
• Own Tax Progressivity Coeff. Positive, sign. in
  aggreg. Growth regressions. Controlling for
  Expend. Redistribution, no effect on per capita
  income, positive effect on growth in aggregate
  Income
• Positive Neighbor Effect. (New Jersey, or Conn
  benefit from NYs progressivity)

26
Expenditure Redistribution Effects

• Negative relation between welfare spending and
  aggreg. Growth. - Endogeneity?
• Negative state ed share effect on p.c. income
  growth, over 5 years.
• Higher Ed negative but insign.

27
Alternative Specifications

• 1) Table 6. Actual Quintile Tax Burdens. Agents
  more likely to respond to actual burden, than
  relative burden. Results no effect on p.c.
  income growth for any quintile negative effect
  on growth in top income. (high income response?)
  
• Weak Evidence of High Income Neighbor Effect
  

28
Robustness Checks

• Table 7.
• 1) National Income Quintiles. Tax Progr.
  Insignificant
• 2) Non-overlapping long-period growth sample No
  change
• 3) Density/Agglom. Effects only PCTURB sign.
  Localization of Agglom. Economies?

29
Additional Tax Specifications

• Table 8.
• 1) controlling for total taxes, no progr.
  Effect.
• 2) Income tax alone top quintile burden insign
  neg. effect on aggreg. Growth of low-income inc.
  tax burden
• Table 9 Education Share
• Robust to inclusion of total ed. Spending.
• Sensitive to tax specification.
• State ed share positively correlated with state
  taxes. Reduction in Tiebout Choice?
• Effect much weaker over ten years. Human K
  investment effect over longer period

30
Neighbor Redistribution

• (Table 9) No Effect on growth in per capita
  income.
• Positive Effect of Neighbor Higher Ed Spending on
  Aggreg. Growth. Positive Spatial Externality.

31
Summary of Results

• There is considerable variation in fiscal
  incidence across states and over time
• Tax progressivity does not adversely affect
  growth. States benefit from progressive taxation
  in neighbor states (NJ vs NY, Rhode Island vs.
  Mass)
• Expend. Progr
• State Ed Share some negative effect, short run.
  Less so long run (Tiebout choice vs. invest. In
  human capital)
• Welfare Spending greater in slower growing
  states. Causality question? Doesnt support
  Helms. (86)
• Higher Ed no growth effect some neighbor growth
  effect.

32
Conclusions

• Why so little effect consistent with
  capitalization and migration studies.
• Federal Cost Sharing Reduces net burden of
  redistribution. Deductibility of taxes, federal
  grants for PA, Medicaid, child care, etc.
• Human capital investment payoff?
• Pareto Optimal variation? Subnational Fiscal
  System is in distributional equilibrium.