the impact of property tax appeals on vertical equity in cook … · 2017-06-07 · center for...
TRANSCRIPT
Center for Municipal Finance1155 East 60th StreetChicago IL 60637T 773.834.1468
The Impact of Property Tax Appeals on VerticalEquity in Cook County, IL
Robert Ross
May, 2017
0This paper began in a practicum lead by Professor Chris Berry during the fall of 2016. Chicago Tribune Reporter Jason Grottohelped obtain the public records necessary for this analysis. Thanks also to the students who participated in the practicum organisingdata and producing preliminary results: Sarah Guminski, Michael Harvey, Dani Litovsky Alcala, Megan Maxwell, Hector SalvadorLopez, Alex Sarabia, Daniel Truesdale, and Alvaro Valdes Mena.
Abstract
Cook County has one of the highest rates of property tax appeals in thecountry. Do appeals make property taxes more fair and accurate? Usingadministrative property tax data from for 2003-2015, I quantify the levelof regressiveness in property taxes. I find property taxes to be highlyregressive. I then investigate whether the appeals process mitigates orexacerbates regressiveness. I find that post-appeal assessment ratios aremore regressive than pre-appeal ratios, providing evidence that the prop-erty tax appeal process decreases vertical equity. I further find that ef-fective tax rates are higher in neighborhoods with a higher percentage ofminorities, and lower in wealthier and better educated neighborhoods.
1 Introduction
In his survey of taxation during the first 100 years ofAmerican independence, Richard T. Ely wrote:
The fundamental idea of our tax systems is ademocratic one. It is... that men are boundto serve the state in the degree in which theyhave the ability to serve themselves. The rea-sons why these earliest methods were aban-doned are sufficiently evident... When oneacre of land was often worth ten or twentytimes, or even fifty times, as much as an-other situated in the same commonwealth,there could not fail to arise a demand for asystem of taxation which would adjust theburdens of the government more accuratelyand make them bear upon each individualmore nearly in proportion to his ability.1
Property tax assessment is a politically contentiousexercise. As assets have become more complex and lessvisible, estimating an individual’s “capacity to serve her-self” has become highly technical and difficult. Modernsystems of property tax assessment evolved out of a se-ries of technical and legal innovations to become quasi-judicial apparatus with both administrative and legallevels of review. Local governments nevertheless con-tinue to rely heavily on property taxes, and fairness intaxation is important to the political and economic sta-bility of those governments. However, while economistsare generally in favor of property taxes, the public’smind is well set against them; the property tax is re-garded as the “least fair” tax by the average American.2
The greater the perception that property taxes areunfair, the more likely taxpayers are to appeal theirtaxes. Indeed, property tax appeals are becoming morenumerous nationwide.3 Cook County, IL (the County),enthusiastically follows this trend. In 2015, about 20percent of residential taxpayers filed more than 360,000appeals. Figure 3 shows statistics for appeals between2003 and 2015. Over that period, the number of appealsincreased by about 20% with every reassessment cycle,as has the percentage of appeals resulting in a reductionin taxable value.
The increase in appeals has been actively encour-aged by various County officials and offices. County of-ficials actively encourage taxpayers to appeal their prop-erty tax assessments, and the County Board of Review
(BOR) trains elected officials, including Chicago legisla-tors, to help constituents appeal their taxes. In addition,the County recently instituted a new online property taxappeal system which allows property owners to easily ap-peal their own taxes without legal representation. Thesefeatures of the County’s property tax system result in avery large number of appeals. Do the large and growingnumber of property tax appeals encourage fairness andaccuracy in Cook County’s property tax system?
The focus of this paper is the impact of the prop-erty tax assessment system, and property tax appealsin particular, on vertical equity in property taxation.I investigate the prima facia case for vertical equityin assessment ratios, property taxes, and tax appeals.The assessment ratio of a property is the ratio of theCook County Assessors’ Office’s (CCAO) estimate ofthat property’s market value to the actual sale price ofthe property, in the year the property sold.4 The effec-tive tax rate on housing wealth is the ratio of taxes paidin the year a property sold to the sale price of that prop-erty, in the year the property sold. If effective tax ratesare strongly negatively correlated with home values, Itake this as evidence of regressive property taxes.
Using administrative data from the Cook County As-sessors’ Office (CCAO), I find that property taxes inCook County became extremely regressive following thecollapse of the housing bubble in 2008. Figure 4 showsthe depressive effect of the housing bubble on sales vol-umes and median home prices in the County. Figure 7displays effective property tax rates by home value forthe period following the housing bubble. That figureclearly shows a strong negative relationship between ef-fective tax rates and property values. This is prima facievidence of regressiveness in property taxes.
I also find that appeals make taxes more regressive.Figure 1 shows the change in assessment ratios for allproperties for the period 2009-2015. Though appealsreduces taxable values, on average, across the range ofhome values, they tend to decrease them more for morevaluable properties. The appeals process as a whole de-creases vertical equity by granting proportionately largerreductions in value to more valuable properties.
Property assessment is a complex, constantly evolv-ing field. While the data for this analysis comes fromCook County, IL, I do not believe the County is uniquein terms of veritcal equity or property tax appeals. Thisanalysis suggests a number of policy changes to encour-age more equity in the property tax system. Lessons
1Ely (1888),[8] p. 1312 Gallup (2005)[11]. Also see Chapter 11 of Fisher (1996)[9].3Doerner and Ihlanfeldt (2012)[7].4For simplicity, I calculate the assessment ratio as 10*(assessed value)/(market value), so that an assessment ratio of 1 means that
the assessor has perfectly predicted a property’s value.
1
learned in the revision and execution of assessment prac-tices in the County can be duplicated by other countiesnationwide.
2 The “least bad” tax
The property tax as administered in the United Statesis a paradox: on the one hand, voters regularly reportthat they believe they get the “best bang for the buck”from property taxes, 5 local governments rely heavily onproperty taxes to finance local services,6 and economistscredit the property tax with a number of positive at-tributes. On the other hand, voters regularly report thatthe property tax is the “most unfair,” “worst” tax ac-cording to repeated opinion surveys,7 there are frequentproperty tax revolts and other initiatives to limit prop-erty taxation, and property tax revenues as a share ofGDP are declining in the US even while other tax rev-enues are increasing. Indeed, the property tax revolts ofthe late 1970s were defining events in the political econ-omy and local public finance of the United States, andyet there is no consensus to date on the causes of theserevolts.8
Canonical public finance textbooks take an opti-mistic view of the property tax with respect to verticalequity. Fisher (2007) sums up a discussion of the eco-nomic effects of the property tax writing “these factorscombined do not support a conclusion of general prop-erty tax regressivity.”9 Rosen (2002) argues that theproperty tax is broadly hated because it is perceived asbeing regressive. This perception he attributes to thedominance of the “traditional view” of property taxesas excise taxes on a bundle of land and capital, and thedisjoint between housing wealth and liquidy assets forsome subgroups of taxpayers.10 He equivocates as towhether the tax is, in fact, regressive.
Other economists have argued as well that the prop-erty tax is theoretically progressive.[30] In an environ-ment where the Tiebout Hypothesis holds, propertytaxes are fully capitalized into home values and behaveslike a user fee rather than a tax. In such an environment,
Rosen (2002) writes that “the notion of progressivenessis meaningless.” It is unclear, however, whether ver-tically inequitable property taxes might still behave asuse taxes; do lower income residents derive greater pro-portional benefits from local government than wealthierresidents? Most recently, Youngman (2016)[29] writes“The easy use of the term regressive to describe the prop-erty tax in popular debate is not justified on economicgrounds,” arguing that the ambiguity of the incidence ofthe property tax makes it difficult to ascertain its trueregressiveness.
Most of the aforementioned work deals with the inci-dence of a uniform tax on real estate. Recent empiricalresearch, on the other hand, has focused more on the ad-ministration of the property tax. In that body of work,there is an emerging consensus that property taxes areadminister in such a way as to be regressive on theirface. This seems particularly true when conditions forproperty appraisal are unfavorable, like the period af-ter a major real estate bubble. Suits (1977)[25], Phares(1980)[20], Metcalf (1994)[18], and Plummer (2003)[21]use different approximations of income to compute asuites index and determine whether property taxes areregressive.11 In these studies, estimated Suites Indicesrange from .23 (fairly progressive) to -.13 (a bit regres-sive). More recently, Cornia and Slade (2006)[6] andMcMillen and Weber (2008)[17] find that property taxesare regressive in Cook County specifically, although theirdata are limited in size and cover periods from before thecollapse of the housing bubble.
McMillen (2011)[15], Hodge et. al. (2016)[12], andKrupa (2012)[14] find that post-housing bubble assess-ments became highly regressive in Cook County, Detroit,and Indiana respectively. McMillen (2013)[16] exam-ines the impact of property tax appeals in Chicago in2006, finding that they decrease vertical equity slightly.Doerner and Ihlandfeldt (2012) [7] examine the effectof appeals on assessment ratios in Miami-Dade Countyand find that they disproportionately benefit white, richneighborhoods. This paper most resembles these papersin scope, methods, and focus.
This paper contributes to the literature on local
5Cabral and Hoxby (2012)[5]6Anderson and Ross (2013)[3]7Gallup (2005)[11]8Anderson and Pape (2011)[2]9Fisher (2007)[10], p. 370
10Rosen (2002)[22], p. 49411The Suites Index was proposed by Suites (1977)[25] as a standard measure of verhtical equity in taxation. I calcualte the index in
Appendix B. For a good description of the Suites Index, and a bootstrapping method for hypothesis testing, see Anderson (2003)[1].For a summary of recent empirical work on progressivness in property taxes, see Stanahan et. al. (2014)[26]. The suites index followsthe logic of the Gini Coefficient calculated from the Lorenze Curve. To calculate the Suites Index, plot the cumulative tax paymentsagainst the cumulative wealth of a population, and calculate one minus the ratio of the area under this curve to the area under a 45degree line. Like the Gini Coefficient, the resulting index shows the degree to which taxpayers at lower levels of wealth bear largerproportions of the total tax burden.
2
property taxes in a number of ways. First, I documentthe prima facia regressiveness of the property tax sys-tem in the second most populous county in the US usinga straightforward approach: effective tax rates. This isthe only study I am aware of to clearly show how regres-sive assessments and appeals lead to regressive taxes. Iuse a very large administrative dataset covering a largeperiod of time to show patterns of assessment and tax-ation across realized property values and neighborhooddemographic attributes. The only other studies I amaware of to use such a large dataset for this purpose isKrupa (2012)[14] and Doerner and Ihlandfeldt (2012)[7].This paper builds on their analysis by offering novel, andsimpler ways of measuring regressiveness, and by exam-ining the period before, during, and after a large hous-ing bubble. This informs our understanding of how theCounty’s property tax performs under conditions whichare unfavorable to predicting property real estate values,and whether the appeals process sufficiently corrects forerrors in assessed values during a housing bubble. Myfindings add to a growing body of literature which sug-gests that the property tax is regressive in practice be-cause assessments are regressive.
3 Data and calculations
The data for this analysis comes from administrativerecords of the CCAO, the Cook County Treasurer’s Of-fice, the City of Chicago, and the US Census Bureau,and includes more than 1.5 million residential properties’taxes, assessed values, sales, appeals, and other infor-mation used for property tax purposes from 2003-2015.Unique Property Identification Numbers were matchedwith the Census Tracts which contained them. Thisallows me to identify correlations between census-tractlevel characteristics with tract-level property taxes. Myfinal dataset contains 21.4 million observations of resi-dential properties’ administrative property tax recordsfrom 2003-2015.
I selected properties which sold between 2003-2015.Consistent with standard assessment practices, I takethe sale price of a home as the realized value in theyear the home sold. My dataset of residential sales con-tains just less than 600,000 residential home and condo-minium sales from 2003-2015.12 Annual sales volumesand median prices are reported in Figure 4.
This study considers three phases in the property taxassessment process. The first step in the assessment pro-
cess is a “first-pass” made be the CCAO. In this step, theCCAO notifies taxpayers of their preliminary assessedvalues. These assessed values mostly determine a tax-payers’ property taxes for the next three years. Afterbeing notified, taxpayers may appeal their assessmentsthrough both the CCAO and the BOR, and then in courtfollowing these agencies determinations. Following theresolution of all appeals, a final assessed values is as-signed to each property. Finally, taxes are calculatedbased on that final assessed value, exemptions, and theapplicable tax rates. This paper focuses on the bench-marks of this process: first-pass assessments, final as-sessments, and taxes.
To measure regressiveness, I rely on the standardmeasure used in ratio studies: Price Related Differen-tial (PRD).13 In Appendix B, I also calculate two othermeasures of vertical equity. These other statistics areconsistent with the patterns shown by the PRD.
The PRD is given as the ratio of un-weighted meanassessment ratios to weighted mean ratios in a given timeperiod:
PRD =
∑nj=1ECCAO[Vj ]/Vj/N∑N
j=1wj ∗ ECCAO[Vj ]/Vj/N(1)
where property values are used as weights. If theweighted mean is equal to the unweighted mean, as-sessments are perfectly uniform across home values andthe PRD will equal 1. If the weighted mean is rela-tively smaller than the unweighted mean, higher valueproperties are assessed at lower rations, and the PRD isgreater than 1. PRD is negatively associated with re-gressiveness. The standard “acceptable” range for PRDis [0.97, 1.02]. This range is asymmetric around 1 be-cause there is an upward bias in the denominator whichdoes not effect the numerator.14 The variance of thisstatistic is not well defined, so I use bootstrapping toobtain an estimate of the variance.
4 Results and discussion
4.1 Vertical equity in assessments
Table 1 and figure 5 show PRD before and after appeals.Figure 2 plots the change in the PRD by major propertytype, with bootstrapped margins of error. And figure 1shows the change in assessment ratios by home value forall County properties 2009-2015.
12I dropped 68 observations where the appeals process actually increased properties’ taxable values. The omission of these observa-tions has no qualitative effect on my results, but their inclusion causes my graphs to display oddly because the amount of increase wasextreamly large for these observations.
13IIAO Standard for Ratio Studies, 2013, p. 1714IIAO (2013)[13], p. 35
3
Figure 1: Impact of appeals on assessment ratios by home value
.6.8
11.2
1.4
Ratio o
f assessm
ents
to s
ale
price
$250K $500K $1 mln $1.3 mln
Sale value
Assessment ratios
−6%
−5%
−4%
−3%
Effect of appeals
on
assessm
ent ra
tio
$250K $500K $1 mln $1.3 mln
Sale value
Difference in pre− andpost− assessment ratios
Cook County 2009−2015
Pre−appeal Post−appeal
Together, these figures demonstrate that propertytax assessments in the County are regressive, and aremade more so by appeals. Looking at homes in the 25thand 75th percentiles of sale price 2009-2015, homes inthe lower quartile are assessed during the CCAO’s first-pass at ratios 23% higher than those in the upper quar-tile. At the end of the appeals process, homes in thelower quartile are assessed at a rate 24% higher thanhomes in the higher quartile. That is so because ap-peals lower taxable value proportionally more at higherhome values. Assessment ratios are 3.3% less after ap-peals for the 25th percentile, and 4.1% less for the 75thpercentile. Overall, appeals make property tax assess-ments more regressive by granting larger reductions athigher market values.
The BOR introduces regressiveness into the prop-erty tax system in the way it grants appeals. The BORoften grants reductions in assessed value to propertieswhich are already under-assessed. More than half ofunder-assessed properties sold 2009-2015 filed an appeal,and the CCAO & BOR granted 22% of those appeal-ing properties some reduction in their assessed values.On average, the CCAO & BOR reduced assessed val-ues for already under-assessed properties which appealedby $5,700, or 23% of the average assessed value for ap-pealing under-assessed properties. By comparison, over-assessed properties saw an average reduction of $8,169,
28% of average assessed values. Simple regressions ofthe probability of winning an appeal, and the percent-age reduction in assessed value on first-pass assessmentsshows very little correlation between winning appealsand first-pass assessment ratios.
Taxpayers are more likely to appeal their assess-ments if they are over-assessed, and if their homes areworth more. Even after accounting for these factors,however, neighborhood demographic characteristics arestill significantly correlated with assessment ratios, ap-peals, and effective tax rates. Table ?? shows the re-sults of regressions on outcomes of interest. Areas withhigher levels of education, higher property values, andfewer minorities pay significantly lower tax rates. Figure8 shows selected correlations with demographic charac-teristics of census tracts and mean effective tax rates.Geographic patterns of tax appeals, assessment ratios,effective tax rates, and home values are shown in 9 -12. These maps convey a sense of the distribution ofproperty taxes to those familiar with the demographictopography of Chicago.
Some of these correlations may be driven by the factthat taxpayers self-select as to whether they appeal theirtaxes or not. This self-selection may drive some regres-siveness in the appeals process as well. For example,many larger condominiums buildings hire attorneys tofile appeals for all of the units in the building, taking
4
advantage of the returns to scale in this type of appeal.Effective tax rates may be consequently lower for largercondominium buildings. Figure 6 shows the strong, pos-itive relationship between condominiums’ building sizeand the probability of any individual unit filing a prop-erty tax appeal. The difference in effective tax rates be-tween condominium properties and single-family homesseen in figure 7 may be a product of the different ratesof appeal between these types of properties, althoughassessment methods also differ between these types ofproperties.
Another factor in tax appeals are attorneys: tax-payers spent an estimated $22 million in lawyers’ fees in2015 appealing their taxes. Tax lawyers in Cook Countysolicit customers during tax season, and it may be thatthose lawyers solicit more valuable property more of-ten, since those properties are more likely to yield largernominal fees. Lawyers may also target English-speakingpopulations, or areas with more density of sales, whereit is easier to find clients and win appeals. There mayalso be other ways in which lawyers select clients whichincrease regressiveness in property taxes.
Finally, there may be simple geographic informa-tional spillovers in appeals off all residential properties,making neighbors of taxpayers that appeal more likelyto appeal themselves. This may be because neighborsvoluntarily share tax information with each other, or be-cause neighbors look at each others’ property tax infor-mation online. This may result in some neighborhoodshaving a very high rate of appeal, simply because of thecompounding effect.15
4.2 Vertical equity in taxes
Regressive property tax assessments lead to regressiveproperty taxes. From 2009-2015, Chicago properties atthe 25th percentile of sale value were taxed at a rate 24%higher than properties at the 75th percentile of value.16
The 25th percentile sale values in Chicago was about$156,000, paying an average tax of about $2,300.
The 75th percentile of sale value was nearly $370,000,paying an average tax of $4,700. The average effectivetax rates at these quartiles were about 1.5% and 1.2%respectively. If homes at the 25th percentile paid thesame tax rate as those at the 75th percentile, their taxeswould be about $465/year lower on average. On theother hand, if Chicago properties in the 75th percentilefaced the average effective tax rate that properties in the25th percentile faced, they would have paid $1,142/year
higher taxes on average.
To further contextualize these figures, I randomly se-lected two representative properties which sold in 2011,the year in which assessments in the County were mostregressive. I restricted my selection to owner-occupied,single-family homes in the 25th and 75th percentiles ofproperty value with successful property tax appeals.
The property in the 25th percentile of home values isin the Village of Stickney, IL, a mostly-white suburbancommunity bordering Chicago to the south west. Stick-ney has a median household income of about $43,000,and about 10% of residents are college educated. Theproperty in the 75th percentile of property values lies inthe North Park community of Chicago, where medianhousehold income is about $71,000 and about 50% ofthe residents hold a college degree.
The Stickney property sold in 2011 for $130,000,and the CCAO estimated its market value at $238,660,nearly double its actual market value. In contrast, theNorth Park property sold in 2011 for $309,000, but theCCAO estimated its market value at $281,100. Bothproperties filed successful appeals in that year: the Stick-ney won a reduction of $12,640 of taxable market value,while the North Park won a reduction of $12,300 of tax-able market value. The statutory property tax rates inStickney are nearly twice as high as they are in the Cityof Chicago, a reflection in part of the lower amount ofcommercial property value in that community. For a di-rect comparison of the impact of assessments on taxes, Iapply the statutory rate in Chicago to both properties.Under uniform tax rates, the Stickney’s property tax billwould have been $4,300, and the North Park property’sbill would been $5,100. If both properties had been as-sessed perfectly, those bills would have been $2,400 and$5,900 under uniform tax rates.
Regressiveness shifts a considerable portion of thetotal property tax burden onto lower value properties.Consider the 13,539 residential sales in the City ofChicago in 2011. Cumulatively, the total value of thosesales was about $5.1 billion. Those properties paid $59.1million in property taxes in that year. Those revenuesare not taken in equal proportion from every property.The County’s property taxes were most regressive in2011; the average effective tax rate on properties inChicago in the 25th percentile of property value was 30%higher than that on properties in the 75th percentile ofproperty values. The nature of property taxes is suchthat decreases in taxes for one group is mechanicallyoffset by increases in taxes for another group. Suppose
15In Cook County, the public may look up property tax records, including taxes paid, exemptions applied, and appeal status, online.16Throughout this analysis, I use Chicago properties to compare effective tax rates, even though I use properties Countywide to
compare assessment ratios. This is so because all residential properties in Chicago lie in the same taxing jurisdictions, so comparisonsof effective tax rates reflect differences in the assessment process.
5
Figure 2: Impact of appeals on vertical equity
−.0
05
0.0
05
.01
.015
.02
.025
.03
.035
.04
Change in P
RD
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
year
Single−family homes
Effect of appeals on PRD
−.0
05
0.0
05
.01
.015
.02
.025
.03
.035
.04
Change in P
RD
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
year
Condominiums
Effect of appeals on PRD
2003−2015
Change in PRDcaused by appeals
Margin of errorfor first−pass PRD
that the revenues from these properties was collecteduniformly according to property value, as is legally stip-
ulated: what impact would this have on tax bills?
Re-calculating each property’s tax bill by applyingthe uniform rate required to raise these funds in equalproportion from all properties, I can quantify the nomi-nal level at which properties are over- and under-taxed.In 2011, nearly 9,500 properties paid taxes higher thanthat calculated from a uniform rate, while about 4,000paid lower taxes. 17 Redistributing that tax burden inequal proportion to housing wealth, I find that the av-erage decrease in tax bills for properties paying a highereffective tax rate is about $1,500, while the average in-crease in tax bills for properties paying lower effectivetax rates is about $4,600. There are about 735,000 resi-dential properties in the City of Chicago. If this sampleof about 1.8% of all properties is representative of thegeneral pattern of vertical inequity across the City, themagnitude of redistribution of tax burdens from higherto lower value homes could be as much as 50X larger.
Figure 13 graphs the results of this exercise. On theleft side of the panel, the average difference between ac-tual taxes paid and taxes paid under a uniform tax rateis plotted against property values. The x-axis is scaledby value percentiles. This graph shows that more than
half of the properties in the sample would experience areduction in their property taxes if a uniform tax ratewas applied across home values. The right panel showsthe average effective tax rates paid at each percentile ofsale value, as well as the uniform effective tax rate re-quired to raise an equivalent amount of revenues fromthis subsample of properties. To the degree that the biasin assessment ratios exhibited among properties whichsold is a good approximation of bias in assessment ratiosfor all properties, this exercise illustrates the magnitudeof the impact of regressiveness in the County.
5 Policy recommendations
Estimating the value of real estate is a difficult and com-plex task. Concerning California’s property tax system,Frederick Stocker wrote
(The property tax) resembles a structure de-signed by a mad architect, erected on a shakyfoundation by an incompetent builder, andmade worse by the well-intentioned repairwork of hordes of amateur tinkerers.18
17In calculating the uniform rate, I exclude properties in the top and bottom 1% of effective tax rate.18Stocker (1991)[27], p. 1.
6
To some degree, such can be said of any democratic in-stitution which is the product of negotiated compromise.With advances in computing and statistical sciences,however, it increasingly seems as though the technicalaspects of property tax assessments are in want of re-design. Improvements in the County’s assessments maytranslate into large improvements in the administrationof the property tax nationwide.
First-pass assessments in the County need to be ac-curate, and their errors should not be correlated withproperty values. If traditional methods of estimation donot yield such errors, the CCAO should explore non-traditional methods. For example, the CCAO couldconduct a competition with a large cash award for theperson or group who designs a significantly better as-sessment algorithm, similar in spirit to the famous Net-fliz Prize. Another example might be to explore surveymethods of predicting property value. Are self-reportedproperty values, with some threat of audit, a more reli-able way to estimate values? Do neighbors truthfully re-port their neighbors’ property values? The CCAO couldinvestigate these and other approaches to property taxassessment.
Transparency can help improve government servicessimply by changing the political incentives faced by thecounty assessor. The CCAO should make the propertytax database available, in full, to the public, along withall computer code necessary to replicate the assessors’estimation process. This would not only increase theaccountability of the system, but also may encourage anentrepreneurial data scientist to suggest improvementsto the predictive models used by the CCAO.
The BOR explicitly evaluates assessments with re-spect to uniformity, or horizontal equity. This ensuresthat properties with similar characteristics are assessedat similar levels. This does not address vertical equityat all, as I have shown. Incorporating a systemic re-view of assessments with respect to vertical equity intoBOR considerations might improve the end result of thereview process.
Improving the property tax assessment system in theCounty is not merely a technical exercise. There is areat least two entrenched political classes which benefitfrom poor-quality assessments. From 2009-2015, taxattorneys made an estimated $133 million in revenuesfrom successful tax appeals. Poor assessments createrevenues for tax attorneys, and any attempt to improveassessment quality will likely be met by resistance fromthese lawyers. In addition, most local politicians as-sist constituents with tax appeals, in some cases hiringstaff members specifically for that purpose. This gener-ates political capital elected officials will be reluctant to
trade for better quality assessments.
References
[1] Anderson, John E., Roy, Atrayee Ghosh, Shoe-maker, Paul A. “Confidence Intervals for the SuitesIndex.” National Tax Journal, n. 1, pt. 1, 2003.
[2] Anderson, Nathan and Andreas Pape. “A Model ofConstitutional Constraints on Benevolent Govern-ments and a Reassessment of the 1970s PropertyTax Revolt.” Working Pape, 2011.
[3] Anderson, Nathan, and Robert Ross. “Rethinkingthe Property Tax.” The Illinois Report: Institutefor Government and Public Affairs, 2013.
[4] Borg, Mary O., Stranahan, Harriet A., and WilliamVoorhees. “The Incidence of the Property Tax andProperty Tax Preferences.” 2014.
[5] Cabral, Marika, and Caroline Hoxby. “The HatedProperty Tax: Salience, Tax Rates, and Tax Re-volts.” NBER Working Paper 18514. 2012.
[6] Cornia, Gary C., and Barrett A. Slade. “HorizontalInequity in the Property Taxation of Apartment,Industrial, Office, and Retail Properties.” NationalTax Journal, no. 1, 2006.
[7] Doerner, William and Keith Ihlanfeldt. “An Empir-ical Analysis of the Property Tax Appeals Process”Journal of Property Tax Assessment and Admin-istation, v.10, n. 4, 2012.
[8] Ely, Richard T. Taxation in the American Statesand Cities. New York: T. Y. Crowell and Co, 1888.
[9] Fisher, G. The Worst Tax: A History of the Prop-erty Tax in America. Lawrence, KA: UniversityPress of Kansas, 1996.
[10] Fisher, Ronald. State & Local Public Finance. 3rdEd. Thomson South-Western, 2007.
[11] Gallup, 2005. “Which is the Unfairest Tax of themall?”
[12] Hodge, Timothy, McMillen, Daniel, Sands, Gary,and Mark Skidmore. “Assessment Inequity in a De-clining Housing Market: The Case of Detroit.” RealEstate Economics, 2016.
[13] International Association of Assessing Officers stan-dard on ratio studies. Kansas City, MO: IAAO.2013.
[14] Krupa, Olha. “An Analysis of Indiana Property TaxReform.” State Tax Notes, V. 65 N. 10, 2012.
7
[15] McMillen, Daniel. “Assessment of Regressivity: ATale of Two Illinois Counties.” Land Lines, LincolnInstitute of Land Policy, January, 2011.
[16] McMillen, Daniel.“The effect of appeals on assess-ment ratio distributions: Some nonparametric ap-proaches.” Real Estate Economics, n.1, 2013.
[17] McMillen, Daniel P. and Rachel N. Weber.“ThinMarkets and Property Tax Inequities: A Multino-mial Logit Approach.” National Tax Journal, n. 4,2008.
[18] Metcalf, Gilbert. “The Lifetime Incidence of Stateand Local Taxes: measuring Chnes During the1980s.” In Tax Progressivity and Income Inequality,edited by Joel Slemrod. Cambridge, UK: CambridgeUP, 1994.
[19] Musgrave, Richard A. “Commentary on LocalProperty Taxation in Theory and Practice.” InProperty Taxation and Local Government Finance,edited by Wallace E. Oates, 339. Cambridge, MA:Lincoln Institute of Land Policy, 2001.
[20] Phares, Donald. Who Pays State and Local Taxes.Cambridge: Oelgeschlager, Gunn and Hain, 1980.
[21] Plummer, Elizabeth. “Evidence on the Incidenceof Residential Property Taxes Across Households.”National Tax Journal, no. 1, 2003.
[22] Rosen, Harvey. Public Finance. 6th Ed. McGraw-Hill Irwin, 2002.
[23] Ross, Myron H. “The Property Tax Assessment Re-view Process: A Cause for Regressive Property Tax-ation?” National Tax Journal, no. 1, 1971.
[24] Sands, Gary and Skidmore, Mark. “Detroit and theProperty Tax: Strategies to Improve Equity andEnhance Revenue.” Lincoln Institute of Land Pol-icy, 2015.
[25] Suits, Daniel B. “Measurement of Tax Progressiv-ity.” American Economic Review, n. 4, 1977.
[26] Stanahan, Harriet A., Coorhees, William, andBorg, Mary. “The Incidence fo the Property Taxand Property Tax Preferences.” Researgate, 2014.
[27] Stocker, Frederick C. Proposition 13: A Ten YearRetrospective. Cambridge, MA: The Lincoln Insti-tute of Land Policy. 1991.
[28] Wassmer, Robert W. “Property Taxation, PropertyBase, and Property Value: An Empirical Test of the‘New View’.” National Tax Journal, No. 2, 1993.
[29] Youngman, Joan. A Good Tax: Legal and PolicyIssues for the Property Tax in the United States.Cambridge, MA: Lincoln Institute of Land Policy,2016.
[30] Appendix G of Marshalls Principles of Economics(1948), Tiebout (1956), Oats (1969), Epple, Fil-imon, and Romer (1984), Hoxby (1999), Gal-lagherm Kurban, and Persky (2013).
8
Appendix A: Graphical results and tables
9
Figure
3:
Tre
nds
inpro
per
tyta
xap
pea
ls
10%30%50%
Percent of allproperties appealing
20
03
20
05
20
07
20
09
20
11
20
13
Ye
ar
Pro
babili
ty o
f a
pp
ea
ling
40%60%80%
Percent of all propertieswinning an appeal
20
03
20
05
20
07
20
09
20
11
20
13
Ye
ar
Pro
babili
ty o
f w
innin
g a
n a
ppe
al
10%30%
Percent reduction intaxable value
0.5
11
.52
F
irst−
pa
ss a
sse
ssm
en
t ra
tio
2009−
2015
85%95%100%
Percent of propertieswinning an appeal
0.5
11
.52
F
irst−
pa
ss a
sse
ssm
en
t ra
tio
2009−
2015
Co
ok C
ou
nty
20
03
−2
01
5
Sin
gle
Fa
mily
Co
nd
om
iniu
ms
Cook
Cou
nty
isdiv
ided
into
thre
eas
sess
men
tar
eas,
each
ofw
hic
his
reas
sess
edev
ery
thre
eye
ars.
The
cycl
ical
pat
tern
inth
era
teof
app
eal
obse
rved
inth
efirs
tgr
aph
ofth
ispan
elre
flec
tsth
efa
ctth
atth
ere
are
man
ym
ore
pro
per
ties
inth
eC
ity
ofC
hic
ago,
the
firs
tas
sess
men
ttr
aid,
than
ther
ear
ein
the
other
two
tria
ds.
10
Figure 4: Impact of the Housing Bubble on volume and prices in Cook County’s housing market.
$150k
$2
00k
$250k
$300
k$350k
Sa
le v
alu
e o
f hom
es
(net
pe
rsonal p
rop
ert
y)
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
Tax year
by triennial reassessment area
Median sale value of homes
01
0000
20000
30
000
4000
0
Re
sid
ential sale
s(t
housand
s o
f sale
s)
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
Tax year
by triennial reassessment area
Volume of residential sales
Chicago NW Suburbs SW Suburbs
11
Figure 5: Price-Related Differential
11
.11
.21
.31
.41
.5
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
year
Single−Family Condominiums
Price−related differential
12
Figure 6: Correlation between condominium building size and probabilty of appeal
0%
20
%4
0%
Pro
ba
bili
ty o
f a
pp
ea
ling
0 200 400 600 800Number of units in
condominium building
By size of building
Probability of appealing
13
Figure
7:
Reg
ress
ivnes
sin
pro
per
tyta
xes
and
asse
ssm
ents
.005.01.015.02.025
Effective tax rate on housing wealth
$5
00
k$
1m
ln$
1.5
mln
$2
mln
Sa
le v
alu
e o
f p
rop
ert
yF
or
113750 p
ropert
ies s
old
in C
hic
ago 2
009−
2015.
Chic
ago 2
009−
2015
Effective
tax r
ate
s
.511.5
Assessment ratios
$5
00
k$
1m
ln$
1.5
mln
$2
mln
S
ale
va
lue
of
pro
pe
rty
For
230177 p
ropert
ies s
old
in C
ook C
ounty
2009−
2015.
Cook C
ounty
2009−
2015
Asse
ssm
ent
Ratios
Sin
gle
−F
am
ilyC
on
do
min
ium
s
14
Figure
8:
Dem
ogra
phic
corr
elat
esw
ith
effec
tive
tax
rate
s
.012.013.014.015.016
Effective Tax Rateon housing wealth
0%
20
%4
0%
60
%8
0%
10
0%
% A
fric
an A
merican
.01.012.014.016.018
Effective Tax Rateon housing wealth
0%
20
%4
0%
60
%8
0%
% H
ispanic
1.3%1.5%1.7%
Effective Tax Rateon housing wealth
$3
1K
$5
K$
10
0K
$1
20
K
Household
incom
e
.01.012.014.016.018
Effective Tax Rateon housing wealth
20
%4
0%
60
%8
0%
% C
olle
ge E
ducate
d
Fo
r 9
79
18
pro
pe
rtie
s s
old
be
twe
en
20
09
−2
01
5
Ch
ica
go
20
09
−2
01
5
15
Figure 9: Map of probability of filing a tax appeal by 2010 Cenusus tract
Quintiles of probabilityof appealing5th quintile4th quintile3rd quintile2nd quintile1st quintile
Cook County 2010 Census Tracts 2009−2015
Appeal rates
16
Figure 10: Map of post-appeal assessment ratios
Quintiles of post−appealassessment ratio5th quintile4th quintile3rd quintile2nd quintile1st quintile
Cook County 2010 Census Tracts 2009−2015
Final Assessment Ratios
17
Figure 11: Map of mean effective tax rate by 2010 Cenusus tract
Quintiles of effectivetax rates5th quintile4th quintile3rd quintile2nd quintile1st quintile
2010 Chicago Census Tracts 2009−2015
Effective tax rates
18
Figure 12: Map of mean residential sale value by 2010 Cenusus tract
Quintiles of meanproperty value5th quintile4th quintile3rd quintile2nd quintile1st quintile
2010 Cook County Census Tracts 2009−2015
Mean home value
19
Figure
13:
Eff
ect
ofap
ply
ing
aunif
orm
tax
rate
topro
per
ties
sold
in20
11
−202468
Avg. difference in uniform rate andactual tax bills
(Thousands of $)
30
13
02
10
36
02
94
0
Sa
le p
rice
of
pro
pe
rty
(Th
ou
sa
nd
s o
f $
)
1%
Uniform
rate2%3%
Effective tax rateon housing wealth
30
13
02
10
36
02
94
0
Sa
le p
rice
of
pro
pe
rty
(Th
ou
sa
nd
s o
f $
)
Fo
r 1
35
39
pro
pe
rtie
s s
old
in
Ch
ica
go
in
20
11
20
11
20
Table
1:
Eff
ect
ofap
pea
lson
PR
Dby
year
Yea
rN
um
ber
ofsa
les
Fir
st-p
ass
asse
ssm
ents
Mar
gin
ofer
ror
for
firs
t-p
ass
PR
D
Fin
al
ass
essm
ents
Eff
ect
of
ap
pea
lson
PR
D
Condominiums
2003
2364
7.9
8049
41.0
0096
67.9
864979
.0060037
2004
2468
7.9
9921
08.0
0106
991.0
029
.0036888
2005
2636
11.
0033
14.0
0100
11.0
05927
.0026129
2006
2262
3.9
9297
95.0
0108
.9989655
.005986
2007
2037
31.
0204
22.0
0139
681.0
20143
-.0002795
2008
1275
71.
0791
48.0
0297
11.0
86757
.0076089
2009
1349
71.
1111
39.0
0487
211.1
26737
.0155981
2010
1215
51.
2142
19.0
0895
861.2
16193
.0019739
2011
1093
21.
4535
66.0
1123
591.4
87601
.0340348
2012
1198
91.
1400
18.0
0639
241.1
61866
.0218487
2013
1570
11.
0995
89.0
0293
51.0
9847
-.001119
2014
1505
91.
0633
91.0
0224
681.0
69806
.0064142
2015
1505
71.
0025
97.0
0178
781.0
15389
.0127921
Single-familyhomes
2003
4784
81.
0332
48.0
0100
41.0
34837
.0015888
2004
5159
91.
0164
08.0
0102
091.0
18406
.0019974
2005
5278
51.
0354
01.0
0115
081.0
37576
.0021751
2006
4096
21.
0256
03.0
0106
41.0
28046
.0024426
2007
2862
41.
0253
14.0
0154
141.0
3179
.0064753
2008
1735
21.
0829
53.0
0346
861.0
91926
.0089734
2009
1678
21.
0981
44.0
0349
111.1
12224
.0140803
2010
1777
61.
1368
79.0
0375
951.1
52979
.0161002
2011
1622
11.
2089
11.0
0525
921.2
18689
.0097774
2012
1842
61.
1252
53.0
0345
231.1
35212
.0099592
2013
2436
51.
1051
95.0
0247
731.1
09264
.0040691
2014
2129
01.
0922
.002
1832
1.0
94257
.0020574
2015
2092
71.
0560
25.0
0195
251.0
63649
.0076243
21
Table
2:
Cor
rela
tes
wit
hap
pea
ling,
win
nin
gap
pea
l
Fir
st-p
ass
Pro
bab
ilit
yof
Pro
bab
ilit
yof
%re
du
ctio
nin
Eff
ecti
veta
xas
sess
men
tsap
pea
lin
gw
inn
ing
an
ap
pea
lta
xab
leva
lue
rate
Pro
pertyChara
cteristics
Sal
ep
rice
4.87
e-13
3.57
e-13
∗∗∗
8.69
e-13∗∗
∗4.6
9e-
13∗∗
∗-1
.34e-
13∗
($10
0,00
0)(1
.89)
(3.6
6)(4
.35)
(3.6
4)
(-2.3
9)
Rat
ioof
firs
t-p
ass
esti
mat
ed0.
0315
∗∗∗
0.23
6∗∗∗
0.1
74∗∗
∗1.0
99∗∗
∗
mar
ket
valu
eto
real
ized
mar
ket
valu
e(6
.61)
(14.6
7)
(5.4
5)
(33.6
8)
Con
dom
iniu
m0.
0063
10.
0842
∗∗∗
0.13
9∗∗∗
0.0
201
-0.0
0741
(0.3
7)(1
3.51
)(9
.34)
(1.1
8)
(-1.0
1)
Nu
mb
erof
unit
sin
-0.0
0442
∗∗0.
0062
0∗∗
∗0.
000148
0.0
0225
-0.0
0457∗
∗∗
ab
uil
din
g(-
3.18
)(1
0.53
)(0
.16)
(1.4
8)
(-8.7
0)
Squ
ared
nu
mb
erof
un
its
0.00
0017
4-0
.000
0246
∗∗∗
-0.0
00000105
-0.0
0000381
0.0
000239∗∗
∗
ina
buil
din
g(1
.49)
(-4.
01)
(-0.0
1)
(-0.4
6)
(3.3
1)
Pro
pertyChara
cteristics
%n
on-w
hit
e0.
0100
∗∗∗
-0.0
0026
7∗∗∗
0.00
343∗∗
∗-0
.000769
0.0
0279∗
∗∗
and
/or
his
pan
ic(5
0.22
)(-
4.15
)(1
2.3
8)
(-0.7
6)
(8.6
9)
%H
old
ing
aB
Aor
hig
her
1.45
2∗∗∗
0.18
3∗∗∗
0.45
8∗∗∗
-0.2
31∗∗
∗0.4
71∗∗
∗
(29.
32)
(9.5
3)(1
0.90)
(-4.6
5)
(10.2
2)
N58
9259
5892
5990
016
90016
279428
r20.
782
0.26
70.
661
0.0
0196
0.9
13
tst
ati
stic
sin
pare
nth
eses
Err
ors
clust
ered
on
2010
Cen
sus
Tra
cts
∗p<
0.0
5,∗∗
p<
0.0
1,∗∗
∗p<
0.0
01
22
Appendix B: Alternate measures of vertical equity
In addition to PRD, I also calculate Price RelatedBias (PRB), and a version of the Suites Index. PriceRelated Bias is calculated by regressing the adjusted as-sessment ratio against the adjusted mean sales price fora given period, and capturing the resulting coefficient β1.I trim the first and last percentile of assessment ratiosfrom my annual regression samples, which is consistentwith standard assessment practices. My regression is:
Yi = β0 + β1Xi + εi (2)
where
Yi =Vi −Median(ECCAO[V ])
Median(ECCAO[V ])(3)
and
Xi =ln( ECCAO[Vi]
(Median(ECCAO[V ]) + Vi)
2ln(2)(4)
Where Med(ECCAO[V ]) is the median CCAO estimatedmarket value.
PRD is positively correlated with regressiveness,while PRB is negatively correlated with regressiveness.The standard “acceptable” bounds for PRB are 0± .03.The PRB may be interpreted as the effect of doublinga property’s value on that property’s assessment ratio.In 2015, the post-appeal PRB figure for the County was-.042. The interpretation of this coefficient is that dou-bling a property’s value would be associated with andecrease in its assessment ratio by 4.2%. As with PRD,I calculate PRB from both first and final assessment ra-tios to measure the impact of appeals on progressiveness,and find qualitatively identical results to PRD.
The Suites Index is given:
S = 1 − (1/5000)
∫ 100
0T (Y )dy (5)
S ≈N∑i=0
[T (Yi) + T (Yi−1)][Yi − Yi−1] (6)
where Y and T (Y ) are the cumulative percentage of thetotal income and the corresponding cumulative percent-age of tax burden, respectively. I use property value inlieu of income, since I cannot directly observe individualtaxpayers’ incomes. I therefore calculate this Suites in-dex using the subsample of properties which sold in eachyear.
Figure 14 shows the “Suites Curve” plotted with a45 degree line, representing a proportional distributionof taxes across property values. In the figure, the SuitesCurve lies above the proportional line, indicating thatlower value properties bear a disproportional larger por-tion of the property tax burden.
Since the Suites Index can only be calculated usingactual taxes paid, I cannot calculate a pre- and post-appeals Index. In order to recalculate taxes for individ-ual taxpayers supposing they had not appealed, I wouldhave to make an assumption about the effect if a largertax base on the statutory tax rate. Tax levies are setex ante the determination of the tax base, and tax ratesare determined simply by the ratio of the tax levy tothe base. Changing the base by excluding appeals willincrease the statutory rate in a somewhat unpredictableway. Thus, I present indices only for post-appeal taxes.
23
Figure 14: Price-Related Differential
0.2
.4.6
.81
Cu
mu
lative
pro
po
rtio
n o
f to
tal ta
x
0 .2 .4 .6 .8 1
Cumulative proportion of total property values
Suites Curve Proportional Distribution
Cook County, 2003−2015
Suites Curve
24
Table
3:
Pri
ce-r
elat
edB
ias
and
Suit
esIn
dic
es
Yea
rN
um
ber
ofsa
les
Fir
st-p
ass
asse
ssm
ents
Fin
alas
sess
men
tsM
argi
nof
erro
rfo
rfi
rst-
pas
sP
RB
Eff
ect
of
app
eals
on
PR
B
Su
ites
Ind
exM
arg
inof
errr
or
2003
7149
5-.
0252
729
-.02
802
.000
8249
-.00
27471
.0037802
.0015675
2004
7628
6-.
0169
512
-.01
8527
9.0
0080
47-.
0015
767
-.0048736
.0014858
2005
7914
6-.
0378
17-.
0383
912
.000
8311
-.00
05742
-.003165
.0015807
2006
6358
5-.
0164
822
-.01
9637
5.0
0080
08-.
0031
553
.0060896
.001734
2007
4899
7-.
0205
835
-.02
1942
3.0
0100
23-.
0013
588
-.0183435
.0019752
2008
3010
9-.
0720
933
-.07
8076
8.0
0144
91-.
0059
834
-.0779939
.00343
2009
3027
9-.
1180
95-.
1312
703
.002
1423
-.01
31753
-.1106438
.0044872
2010
2993
1-.
1460
46-.
1514
736
.002
0353
-.00
54276
-.1760897
.0064089
2011
2715
3-.
1864
577
-.19
3930
8.0
0205
69-.
0074
731
-.3616682
.0078987
2012
3041
5-.
1304
321
-.13
5322
9.0
0147
16-.
0048
909
-.1064669
.0042314
2013
4006
6-.
1122
607
-.10
6440
9.0
0111
06.0
058197
-.0896962
.0029003
2014
3634
9-.
0866
059
-.08
5319
6.0
0109
66.0
012862
-.079108
.003676
2015
3598
4-.
0353
427
-.04
2676
2.0
0106
1-.
0073
335
-.0325671
.0022702
25