Political Economy of Parcel Tax
in California School Districts
Soomi Lee
Assistant Professor
University of La Verne
January 9, 2016
Southern Political Science Association Annual Meeting
San Juan, Puerto Rico
California Tax Foundation, 2014
36
25
15
7
7
3
2
2
2
1
Education (K14)
Infrastructure
Health Care
Park and Recreation
Fire Protection District
Library
Public Safety
Unavailable
Emergency Medical Services
Other
Parcel Tax Revenue (2013-14):
Percentage Distribution by Function
$1.9 billion in total
From 1970 parcel taxes
in 754 local governments
1
Parcel tax elections in school
districts
• First school parcel tax was adopted in 1983.
• 644 parcel tax measures since then.
– 23% of school districts (222) held at least one
parcel tax election.
– 13% (124) passed at least one parcel tax
measure.
2
57
68 70
84
93
108
2003-04 2005-06 2007-08 2009-10 2011-12 2012-13
Number of School Districts with
Parcel Taxes in California
Ed Source 2013.
Data: SACS unaudited data files, California Department of Education 3
$373 million
$138 million
Previous studies
• Why isn’t the use of parcel tax more pervasive in
California school districts?
• Focus on characteristics of school districts with
parcel tax
1. Brunner (2001): marginal price too high
2. Lang and Sonstelie (2015): tax price for per
pupil spending, district median income, Bay
Area
3. Kiewiet and Hill (2015): political ideology,
election strategies, district median income, Bay
Area 4
• Not surprising that rich districts are more
willing to pay additional taxes for public
schools. Why do some rich districts adopt a
parcel tax but not others?
• Given the same income, what determines the
likelihood of parcel tax adoption?
• Regressivity and the supermajority
requirement would make distribution of
property values more important than the
median household income.
5
Illustrative model: Community with
two home owners, A and B
• Assumptions for simplicity
– Same demand function for public schooling for A and
B
– Same income for A and B
– All property taxes are spent on public schools
– Total enrollment = 100
6
Total parcel tax revenue goal = $400
A B
Property value ($) 500,000 300,000
1% ad valorem property tax ($) 5,000 3,000
A B
Property value ($) 500,000 300,000
1% ad valorem
property tax ($)
5,000 3,000
Ad Valorem Property Tax Regime
Proposed tax increase = 0.05%
New property tax rate 1.05% 1.05%
Tax increase ($) 250 150
Total taxes 5,000+250=5,250 3,000+150=3,150
Additional cost per pupil
for each household
250/100=$2.5 150/100=$1.5
7
A B
Property value ($) 500,000 300,000
1% ad valorem
property tax ($)
5,000 3,000
Ad Valorem Property Tax Regime
Total parcel tax revenue goal = $400
Proposed tax increase = 0.05%
New property tax rate 1.05% 1.05%
Tax increase ($) 250 150
Total taxes 5,000+250=5,250 3,000+150=3,150
Additional cost per pupil
for each household
250/100=$2.5 150/100=$1.5
8
1. Different income effects (thus, income
distribution may play a bigger role)
2. Marginal cost of schooling per pupil is higher
for A  less likely to support parcel tax
measures
A B
Property value ($) 500,000 300,000
1% ad valorem
property tax ($)
5,000 3,000
Parcel Tax Regime
Proposed parcel tax per parcel = $200
Additional cost per pupil
for each household
200/100 = $2 200/100=$2
Taxes after parcel tax 5,200 3,200
New property tax rate 1%+0.04% = 1.04% 1%+0.07% = 1.07%
9
Same income effect for A and B
A B
Property value ($) 500,000 300,000
1% ad valorem
property tax ($)
5,000 3,000
Parcel Tax Regime
Total parcel tax revenue goal = $400
Proposed parcel tax per parcel = $200
Additional cost per pupil
for each household
200/100 = $2 200/100=$2
Taxes after parcel tax 5,200 3,200
New property tax rate 1%+0.04% = 1.04% 1%+0.07% = 1.07%
10
1. 0.03%pt higher property tax rate on B  less
likely to support parcel tax
2. Districts with a large difference in property
values would have a larger property tax rate
differences.
11
Stylized fact 1: median income & support for
parcel tax adoption
12
Stylized fact 2: housing value gap & support
for parcel tax adoption
Empirical test:
Pooled logistic regression
• Data: California school districts with 200 or more
students from 2003 to 2013. N=4063.
• Dependent variable: presence of parcel tax
revenue (binary)
• Main explanatory variables:
– Income gap: mean to median household
income ratio
– Home value gap: upper 25% to lower 25%
average single family home value ratio
13
• Control variables
– Median household income
– Median home value
– Racial heterogeneity (Herfindahl index)
– State aid (% of total revenue)
– Total population
– Population of age 65 or older (%)
– Renters in occupied housing units (%)
– People with BA degree (%)
– Bay Area indicator (1=Bay Area, 0=Otherwise)
14
15
(1) (2) (3) (4)
Median household
income (log)
4.086***
(.157)
Median housing
price (log)
3.082***
(.137)
Income gap
1.201***
(.306)
Housing value gap
-2.542***
(.189)
Control variables No No No No
N 4062 4062 4062 3884
Notes: Robust standard errors are in parentheses.; * p<0.1; ** p<.0.05; *** p<0.01
Preliminary results 1:
Bivariate logistic regressions
16
(1) (2) (3)
Median household
income (log)
-.440 -.187
(.460) (.399)
Median housing price
(log)
.329 .226
(.266) (.227)
Income gap
-.705 -.613 -.573
(.468) (.470) (.470)
Home price gap
-.212** -.247** -.210*
(.107) (.106) (.113)
Control variables Yes Yes Yes
N 3884 3884 3884
Notes: Robust standard errors are in parentheses.; * p<0.1; ** p<.0.05; *** p<0.01
Preliminary results 2:
Multiple logistic regressions
17
(1) (2) (3)
Median household
income (log)
-.585 -.331
(.468) (.420)
Median housing
price (log)
.331 .200
(.267) (.235)
Income gap
-.577 -.487 -.427
(.489) (.486) (.495)
Home price gap
-.205** -.241** -.204*
(.103) (.102) (.110)
Racial diversity
.864 .862 .745
(.562) (.563) (.555)
Control variables Yes Yes Yes
N 3884 3884 3884
Notes: Robust standard errors are in parentheses.; * p<0.1; ** p<.0.05; *** p<0.01
Preliminary results 3:
Effect of racial heterogeneity
Conclusion
• Within-variations in property values in school
districts, not the median household income,
have robust effects on parcel tax adoption.
• The empirical model needs to be improved to
address potential selection bias and
endogeneity.
18
Implications & future research
• Research on parcel tax is about how
Californians cope with local revenue constraints.
• Heterogeneous school districts in many
Southern California communities will have less
opportunity to bring additional resources such as
parcel taxes.
• The impact of parcel tax on public school
funding is unknown, however.
– E.g. Its impact on financial disparities across
school districts and on school performance
19
Ed Source 2013.
Data: SACS unaudited data files, California Department of Education
77
23
87
13
Never Held an
Election (736
districts)
Held at Least
One Election (222
districts)
Never passed a
Parcel Tax (834
districts)
Passed at Least
One Parcel Tax
(124 districts)
School Districts and Parcel Tax
Elections (%), 1983-2012
20
2.4
3.1 3.1
3.7 3.7
2003-04 2005-06 2007-08 2009-10 2011-12
57 districts 68 districts 70 districts 84 districts 93 districts
District Average Parcel Tax
Revenue
($millions, 2011 constant)
Ed Source 2013.
Data: SACS unaudited data files, California Department of Education 21
Parcel Tax Elections by California School Districts
1983-2008
Lang and Sonstelie, 2015. 22
Characteristics of Parcel Tax Ballot Measures, 1983-2013
Effort to lower political costs
Ballot Measure Provisions Percent (# elections)
Senior Exemption 78.6% (439/559)
SSI Disability Exemption 11.4% (64/559)
Other Exemption 6.8% (25/559)
Ad Valorem Correlates 17.0% (95/559)
COLA 20.6% (115/559)
No Sunset Clause 6.2% (40/648)
Kiewiet and Hill, 2015
Strategies to reduce political
costs
23

SPSA 2016

  • 1.
    Political Economy ofParcel Tax in California School Districts Soomi Lee Assistant Professor University of La Verne January 9, 2016 Southern Political Science Association Annual Meeting San Juan, Puerto Rico
  • 2.
    California Tax Foundation,2014 36 25 15 7 7 3 2 2 2 1 Education (K14) Infrastructure Health Care Park and Recreation Fire Protection District Library Public Safety Unavailable Emergency Medical Services Other Parcel Tax Revenue (2013-14): Percentage Distribution by Function $1.9 billion in total From 1970 parcel taxes in 754 local governments 1
  • 3.
    Parcel tax electionsin school districts • First school parcel tax was adopted in 1983. • 644 parcel tax measures since then. – 23% of school districts (222) held at least one parcel tax election. – 13% (124) passed at least one parcel tax measure. 2
  • 4.
    57 68 70 84 93 108 2003-04 2005-062007-08 2009-10 2011-12 2012-13 Number of School Districts with Parcel Taxes in California Ed Source 2013. Data: SACS unaudited data files, California Department of Education 3 $373 million $138 million
  • 5.
    Previous studies • Whyisn’t the use of parcel tax more pervasive in California school districts? • Focus on characteristics of school districts with parcel tax 1. Brunner (2001): marginal price too high 2. Lang and Sonstelie (2015): tax price for per pupil spending, district median income, Bay Area 3. Kiewiet and Hill (2015): political ideology, election strategies, district median income, Bay Area 4
  • 6.
    • Not surprisingthat rich districts are more willing to pay additional taxes for public schools. Why do some rich districts adopt a parcel tax but not others? • Given the same income, what determines the likelihood of parcel tax adoption? • Regressivity and the supermajority requirement would make distribution of property values more important than the median household income. 5
  • 7.
    Illustrative model: Communitywith two home owners, A and B • Assumptions for simplicity – Same demand function for public schooling for A and B – Same income for A and B – All property taxes are spent on public schools – Total enrollment = 100 6 Total parcel tax revenue goal = $400 A B Property value ($) 500,000 300,000 1% ad valorem property tax ($) 5,000 3,000
  • 8.
    A B Property value($) 500,000 300,000 1% ad valorem property tax ($) 5,000 3,000 Ad Valorem Property Tax Regime Proposed tax increase = 0.05% New property tax rate 1.05% 1.05% Tax increase ($) 250 150 Total taxes 5,000+250=5,250 3,000+150=3,150 Additional cost per pupil for each household 250/100=$2.5 150/100=$1.5 7
  • 9.
    A B Property value($) 500,000 300,000 1% ad valorem property tax ($) 5,000 3,000 Ad Valorem Property Tax Regime Total parcel tax revenue goal = $400 Proposed tax increase = 0.05% New property tax rate 1.05% 1.05% Tax increase ($) 250 150 Total taxes 5,000+250=5,250 3,000+150=3,150 Additional cost per pupil for each household 250/100=$2.5 150/100=$1.5 8 1. Different income effects (thus, income distribution may play a bigger role) 2. Marginal cost of schooling per pupil is higher for A  less likely to support parcel tax measures
  • 10.
    A B Property value($) 500,000 300,000 1% ad valorem property tax ($) 5,000 3,000 Parcel Tax Regime Proposed parcel tax per parcel = $200 Additional cost per pupil for each household 200/100 = $2 200/100=$2 Taxes after parcel tax 5,200 3,200 New property tax rate 1%+0.04% = 1.04% 1%+0.07% = 1.07% 9 Same income effect for A and B
  • 11.
    A B Property value($) 500,000 300,000 1% ad valorem property tax ($) 5,000 3,000 Parcel Tax Regime Total parcel tax revenue goal = $400 Proposed parcel tax per parcel = $200 Additional cost per pupil for each household 200/100 = $2 200/100=$2 Taxes after parcel tax 5,200 3,200 New property tax rate 1%+0.04% = 1.04% 1%+0.07% = 1.07% 10 1. 0.03%pt higher property tax rate on B  less likely to support parcel tax 2. Districts with a large difference in property values would have a larger property tax rate differences.
  • 12.
    11 Stylized fact 1:median income & support for parcel tax adoption
  • 13.
    12 Stylized fact 2:housing value gap & support for parcel tax adoption
  • 14.
    Empirical test: Pooled logisticregression • Data: California school districts with 200 or more students from 2003 to 2013. N=4063. • Dependent variable: presence of parcel tax revenue (binary) • Main explanatory variables: – Income gap: mean to median household income ratio – Home value gap: upper 25% to lower 25% average single family home value ratio 13
  • 15.
    • Control variables –Median household income – Median home value – Racial heterogeneity (Herfindahl index) – State aid (% of total revenue) – Total population – Population of age 65 or older (%) – Renters in occupied housing units (%) – People with BA degree (%) – Bay Area indicator (1=Bay Area, 0=Otherwise) 14
  • 16.
    15 (1) (2) (3)(4) Median household income (log) 4.086*** (.157) Median housing price (log) 3.082*** (.137) Income gap 1.201*** (.306) Housing value gap -2.542*** (.189) Control variables No No No No N 4062 4062 4062 3884 Notes: Robust standard errors are in parentheses.; * p<0.1; ** p<.0.05; *** p<0.01 Preliminary results 1: Bivariate logistic regressions
  • 17.
    16 (1) (2) (3) Medianhousehold income (log) -.440 -.187 (.460) (.399) Median housing price (log) .329 .226 (.266) (.227) Income gap -.705 -.613 -.573 (.468) (.470) (.470) Home price gap -.212** -.247** -.210* (.107) (.106) (.113) Control variables Yes Yes Yes N 3884 3884 3884 Notes: Robust standard errors are in parentheses.; * p<0.1; ** p<.0.05; *** p<0.01 Preliminary results 2: Multiple logistic regressions
  • 18.
    17 (1) (2) (3) Medianhousehold income (log) -.585 -.331 (.468) (.420) Median housing price (log) .331 .200 (.267) (.235) Income gap -.577 -.487 -.427 (.489) (.486) (.495) Home price gap -.205** -.241** -.204* (.103) (.102) (.110) Racial diversity .864 .862 .745 (.562) (.563) (.555) Control variables Yes Yes Yes N 3884 3884 3884 Notes: Robust standard errors are in parentheses.; * p<0.1; ** p<.0.05; *** p<0.01 Preliminary results 3: Effect of racial heterogeneity
  • 19.
    Conclusion • Within-variations inproperty values in school districts, not the median household income, have robust effects on parcel tax adoption. • The empirical model needs to be improved to address potential selection bias and endogeneity. 18
  • 20.
    Implications & futureresearch • Research on parcel tax is about how Californians cope with local revenue constraints. • Heterogeneous school districts in many Southern California communities will have less opportunity to bring additional resources such as parcel taxes. • The impact of parcel tax on public school funding is unknown, however. – E.g. Its impact on financial disparities across school districts and on school performance 19
  • 21.
    Ed Source 2013. Data:SACS unaudited data files, California Department of Education 77 23 87 13 Never Held an Election (736 districts) Held at Least One Election (222 districts) Never passed a Parcel Tax (834 districts) Passed at Least One Parcel Tax (124 districts) School Districts and Parcel Tax Elections (%), 1983-2012 20
  • 22.
    2.4 3.1 3.1 3.7 3.7 2003-042005-06 2007-08 2009-10 2011-12 57 districts 68 districts 70 districts 84 districts 93 districts District Average Parcel Tax Revenue ($millions, 2011 constant) Ed Source 2013. Data: SACS unaudited data files, California Department of Education 21
  • 23.
    Parcel Tax Electionsby California School Districts 1983-2008 Lang and Sonstelie, 2015. 22
  • 24.
    Characteristics of ParcelTax Ballot Measures, 1983-2013 Effort to lower political costs Ballot Measure Provisions Percent (# elections) Senior Exemption 78.6% (439/559) SSI Disability Exemption 11.4% (64/559) Other Exemption 6.8% (25/559) Ad Valorem Correlates 17.0% (95/559) COLA 20.6% (115/559) No Sunset Clause 6.2% (40/648) Kiewiet and Hill, 2015 Strategies to reduce political costs 23