Who Wins the Cook County Property Tax Appeal

Abstract

A Cook County homeowner who believes the Assessor has overvalued the house can contest that value, in writing, for no fee. This paper follows that ordinary act from the mailed notice to the certified verdict and asks a plain question. Who files these appeals, and who actually wins them. We keep two kinds of evidence strictly apart. First, we synthesize a peer-reviewed and government literature that documents, with race and income data and causal methods we do not have, that Cook County assessments were long regressive and that appeal relief tends to flow unevenly toward commercial owners and better-resourced homeowners. Second, we offer our own descriptive read of 5,000 real Cook County Assessor appeal records, of which 99.2 percent fall in tax year 2003, drawn from Cook County Open Data.^[1] Defining a win as a certified total assessed value below the mailed total, the taxpayer's real economic outcome, we find that 32.7 percent of scored 2003 appeals won a cut, 67.2 percent changed nothing, and 0.1 percent came back higher.^[1] The win rate splits hard by what kind of property is contested. Land won 60.4 percent of the time, commercial 44.7 percent, residential 37.1 percent, and condo and co-op owners just 9.6 percent.^[1] A commercial appeal was roughly 4.6 times as likely to win as a condo or co-op appeal.^[1] The gap between who files and who wins is just as sharp. Condo and co-op owners filed 23.6 percent of appeals and collected 7.0 percent of the wins, while commercial owners filed 21.6 percent and took 29.5 percent.^[1] Among winners, the median cut ran from 7.0 percent for residences to near total for a small tail of vacant land.^[1] First-stage win rates spanned 19.9 to 44.6 percent across seven townships, a 24.7-point spread.^[1] Our records carry no race, income, or sale-price field, cover one cycle, and stop at the Assessor's first stage, so every regressivity-by-race-or-income claim here rests on the cited literature and none of it on us.

What does it cost to ask for a smaller tax bill

The envelope arrives in late summer or early fall, the timing set by which township you live in, and it carries a single number that will cost you money. The Cook County Assessor has decided what your home is worth for tax purposes. That figure, the assessed value, feeds the property tax bill you pay the following year. Most people open the notice, feel a small jolt at the number, and file it in a drawer. Some do not. Some look at the figure, then at the house, then at what the two-flat down the block sold for last spring, and decide the Assessor got it wrong. For those people the county offers a remedy, and it is one of the very few things county government hands out at no charge. You can appeal.

The first-stage appeal goes back to the Assessor's own office, the same office that produced the number you are contesting. You need no lawyer. You pay no fee. You complete a form, on paper or online, and you make a case. The case almost always rests on comparison. You point to other properties that resemble yours, near in size, age, location, and construction, and you show that they carry lower assessed values, so yours is out of line. Or you argue that the office's description of your home is simply wrong, that it has the square footage too high or the lot too large. You attach what you have. You submit it. Then you wait. In time a second number comes back, the certified assessed value, which is the office's answer. It may be lower, in which case you have won a cut. It may be unchanged. In rare cases it can be higher than where you started.

That is the entire transaction, told from the kitchen table. A number arrives. A homeowner decides whether to push back. The homeowner pushes back. A verdict returns. The process is short to describe, free to use, and open to anyone with a parcel and a grievance. On paper it is the most democratic instrument in the whole property tax apparatus, a self-service correction available to the owner of a bungalow in Hanover and the owner of a lakefront house in New Trier on the same nominal terms, with the same form and the same price of zero.

The timing is not arbitrary, and it shapes who is paying attention when the window opens. Cook County reassesses on a triennial cycle, dividing the county into three regions and revaluing one of them each year, so a given parcel gets a fresh assessed value once every three years and an appeal window that opens township by township as the new values are mailed. The window for any one township stays open only a matter of weeks. Miss it and the value stands until the next notice. This is part of why the appeal, for all its formal openness, rewards a particular kind of attentiveness. You have to know the notice is coming, recognize the number as contestable, assemble a comparison, and act inside a short window that does not announce itself loudly. A management company that handles hundreds of commercial parcels has staff whose job is to watch those windows. A household that opens the mail after a long shift may register the number, feel the jolt, and let the window close before the weekend is over. Nothing in the rules favors the management company. The calendar does some of that work on its own.

It also helps to be precise about what the appeal contests and what it does not. The appeal challenges the assessed value, the office's estimate of the property, and not the tax rate or the levy. Those are set elsewhere, by the taxing bodies that decide how much revenue they need, and the assessment only divides that fixed need across parcels. So an appeal does not lower the total amount the county collects. It lowers one parcel's share of it and, mechanically, raises everyone else's by a little. That feature, easy to miss from the kitchen table, is what makes the question of who wins more than a private matter between an owner and the office. Every successful appeal is a small reallocation of a fixed bill across all the parcels that did not win one.

This paper takes that one transaction and asks what happens when you run it tens of thousands of times. The question we keep returning to is deliberately unglamorous. Who bothers to file, and who actually gets the cut. These turn out to be two questions, not one, and they have two different answers. Filing is an act of initiative. Winning is a verdict the office hands down. Between the two sits a procedure that is supposed to be neutral, a matching of your parcel against comparable ones, and the work of this paper is to walk that procedure from the mailed notice to the certified value and watch where it comes to rest.

We widen the lens as we go. One appeal is an anecdote. Several thousand appeals, read carefully, begin to describe a pattern, and the pattern is the thing we are after. The records come from Cook County Open Data, specifically the Assessor's file of appeals, and they hold what the county tracks when an owner contests a value. The class of the property. The township. The value mailed out. The value certified back. The stated reason for the decision. They do not hold everything, and we will be candid about the gaps when we reach the sections where they bite, but they are genuine records of genuine appeals, and they let us watch the remedy run at scale.

A mild irony sits inside all of this, and it deserves an early flag even though the opening will not pretend to explain it. A remedy can be free, and open to all, and still hand out its results unevenly. Equal access to a process is not equal results from it. A homeowner who files an appeal has done exactly as much, procedurally, as the corporation that files one across town, yet the two do not walk away with reductions at the same rate, and when both do win, the cuts are not the same size or anywhere close to it. We are not, here, claiming to know why. We are flagging that the gap is real and that a careful portrait of the data can show its shape even where it cannot supply its cause.

For the reader who wants one number to carry forward, here is the headline, and it is ours, taken from our own read of the records rather than from any outside source. Among the scored appeals we examined, about a third won a cut.^[1] Roughly two in three changed nothing at all. That alone resets the expectation. The appeal is not a quiet courtesy that trims your bill when you ask. It is a contest, and most filers lose it. The more revealing finding is not the overall rate but how unevenly it breaks apart once you sort the filers by what they own, and that breakdown is where this paper is going.

Before we get there, two clarifications about what this document is and is not, because the distinction governs every figure that follows, and we would rather draw the line in plain sight than smudge it.

What exactly are we claiming, and what are we not

This paper does two separate things, and the separation is the point. It synthesizes a published literature. It reports our own descriptive analysis of one real public dataset. Those are different kinds of evidence with different reaches, and we keep them apart so the reader always knows which one is speaking.

Begin with what we did not do, since the list is short and it clears away the most common misreadings before they start. We ran no experiment. We did not scrape anything live. The records came to us as a file from Cook County Open Data, and we analyzed what was in the file. We interviewed no homeowners, no Assessor staff, no tax attorneys. We built no statistical model and estimated no causal effect. Nothing in our own numbers establishes a cause. When we report that a commercial appeal wins more often than a condo appeal, we are describing a difference in rates that sits in the data, not naming a mechanism that produces it. The numbers describe. They do not explain, and we will not smuggle in an explanation by reaching for a causal verb where a descriptive one is the honest choice.

That restraint matters most around race and income, which is the question a reader of a civic housing paper will rightly hold in mind from the first page. The published literature, which the next two sections summarize, documents that property tax burdens in Cook County and across the country have fallen unevenly by race and by income, and that the appeals process has been part of how that came to pass. These are serious findings from serious work, and we lean on them. But they are not our findings, and we want the boundary exact. Any claim in this paper that the appeal system is regressive by race or by income is carried entirely by the cited literature.^[2][8][9] Our own file cannot reach it. The records describe who files and who wins. They contain no race field, no income field, and no sale-price field. We could not name the race of a single filer in this dataset if we tried, because the information is not in it. So when our descriptive numbers sit beside the literature's regressivity findings, the reader should read the relationship as consistency and context, never as proof. Our data are consistent with the literature's account. They do not establish it, and we will say so again, in those words, wherever the temptation to blur the two is strongest.

With the boundary drawn, the positive claim. We hold a sample of 5,000 Cook County Assessor appeal records. Of those, 4,962, or 99.2 percent, fall in tax year 2003, so for every practical purpose this is a portrait of one assessment cycle.^[1] We treat it that way throughout. Every rate and every share we report is computed on the tax-year-2003 rows alone, not on the few dozen stray rows from other years that ride along in the file. This is not a trend line across decades. It is a single cycle, seen in some detail, and we let it be exactly that.

The measure of a win runs through the entire analysis, so we state it once, carefully, and then trust it. A win is a certified total assessed value that comes in below the mailed total.^[1] Put it plainly. The office told you the property was worth some figure, you appealed, and the office sent back a lower figure. That lower figure is the taxpayer's real economic outcome, the thing that actually shrinks the eventual bill, which is why we anchor on it rather than on a softer signal. The county does keep its own flag marking whether an appeal resulted in a change, but that flag is noisier than the value arithmetic, for a reason we lay out in the methodology, and we preferred the arithmetic because it answers the only question the homeowner actually has. Did the number go down.

Four descriptive cuts of the data follow. We look first at the win rate by type of property, the sharpest contrast in the paper. Then we set who files appeals against who collects the wins, two distributions that turn out not to match. Then the size of the cut among the appeals that won, since a token reduction and a large one are both wins and counting them the same would flatten a real difference. Last, the win rate across townships within the single cycle, a geographic spread wider than one might guess. None of the four is a causal story. Each is a way of arranging the same records until a pattern becomes legible, and each comes with its limits attached.

One procedural promise, and then the backdrop. We state the limits of the data as they arise, in the section where each one bites, rather than herding them into a single confessional paragraph at the end where they are easy to skip past. When a number rests on a small count, we say so beside the number. When a rate reflects only the first stage of a multi-stage appeals system, we flag it where it matters. The dataset and the reproducible files behind every figure are ours to stand behind, and they are posted for anyone who wants to recompute the arithmetic.^[1] The discipline is simple. Better to claim less and be right than to claim more and be caught.

Now to the structure our single filer steps into, which is where the literature does its work.

Why do cheaper homes carry more than their share

The homeowner at the kitchen table cannot see the system from where they sit. They see one number on one notice for one house. But that notice arrives inside a much larger structure, and a substantial body of research has spent years describing it. What the research finds is uncomfortable. In a great many places, with Cook County prominent among them, the property tax has leaned hardest on the people who own least.

The cleanest statement of the problem comes from Christopher Berry, whose national analysis of assessment data found a tilt that is consistent and stable in direction. Inside a typical jurisdiction, the least valuable homes are assessed too high relative to what they actually sell for, and the most valuable homes are assessed too low. The gap is not faint. Berry found that the bottom tenth of homes by value were assessed at roughly twice the rate, measured against sale price, of the top tenth.^[4] Apply one identical tax rate to two such assessments and you do not get two fair bills. You get a bill that takes a proportionally larger bite out of the cheaper house. Berry then closed off the obvious objection. One might guess this is just noise, that cheap homes are harder to value and the errors cancel out. They do not cancel out. The pattern survives correction for sale-price measurement error, which means it is built into how assessment works rather than into how sales happen to be recorded.^[4] When a regressive result holds after you strip out the measurement noise, the regressivity is the signal.

David Schleicher, surveying the broader literature, frames the same result as something close to a national regularity. Expensive homes come out under-assessed and cheaper homes over-assessed almost everywhere anyone has looked, across jurisdictions with very different politics, very different staff, and very different procedures.^[7] The consistency is the part that should unsettle a reader, because it rules out the comfortable theory that regressivity is a local accident of one bad assessor or one corrupt office. Schleicher's synthesis points to several mechanisms that recur from place to place. Assessors tend to misvalue at the level of the neighborhood, missing how individual homes inside a neighborhood diverge from its average. Reassessment happens infrequently, so values drift away from the market between updates, and the drift does not land on everyone evenly. And appeals, the very remedy this paper follows, are used unevenly, so the correction they offer reaches some owners far more than others.^[7] That last mechanism is the hinge, because it is the only one of the three our own data touch, and it is the one we spend the rest of the paper on.

The argument is worth slowing down on, because it inverts the intuition most people bring to the appeal. A homeowner who wins a reduction naturally reads the outcome as fairness restored. The office had the number too high, the owner objected, the office corrected it. Schleicher's synthesis, and the Chicago appeal studies under it, say the picture looks different from above. If the regressivity is structural, baked into how cheap and expensive homes are valued relative to the market, then the appeal does not erase it. The appeal lets whoever can work the system claw back some of their own assessment, and the people best able to work the system are, on average, not the people the regressivity hit hardest. So the correction is real for the individual and perverse for the distribution. The system can be growing less fair, parcel by defensible parcel, with every reduction it grants. That is the claim our own data are positioned to illustrate in outline, though not to prove, and it is why the question of who wins the appeal is not a procedural footnote but the center of the regressivity story.

The racial dimension has to be handled with care, and we handle it strictly through the published work, none of it ours. Carlos Avenancio-Leon and Troup Howard studied assessment data nationally and found that, holding the taxing jurisdiction and the tax rate fixed, Black and Hispanic residents bore a property tax burden roughly 10 to 13 percent higher than white residents.^[2] The conditioning is the whole point. This is not the familiar and less interesting observation that different towns tax at different rates. It is a gap that opens between neighbors inside the same jurisdiction paying the same nominal rate, and it traces back to assessment. Assessments, the authors find, track the value of the neighborhood more than the actual market price of the specific home. Because race and neighborhood are so tightly bound together in American cities, the legacy of decades of segregation in where families were permitted to buy, that tracking error falls along racial lines.^[2] A Black-owned home and a white-owned home of the same market value, sitting in differently valued neighborhoods, are assessed as though they belonged to their neighborhoods rather than to the market, and the bill follows the assessment.

The mechanism Avenancio-Leon and Howard identify is worth stating carefully, because it is more specific than a vague charge of bias. The problem is not, in their account, that assessors deliberately mark Black homes higher. It is that assessment leans on neighborhood-level information, and within any given neighborhood the assessed values do not spread out as much as the actual sale prices do. A neighborhood gets an estimate, and individual homes are pulled toward it. In a neighborhood where market prices are depressed for reasons bound up with race and the history of disinvestment, that pull leaves the typical home assessed at a higher fraction of its true market value than a comparable home in a higher-valued, whiter neighborhood would be. The bias enters through the geography of valuation rather than through any single appraiser's intent, which is precisely why it persisted for so long without anyone choosing it. It is the kind of harm that an automated, neighborhood-anchored process can produce while every individual step inside it looks neutral.

In a companion study, the same authors examined what happens when a jurisdiction caps how fast an assessment can grow from year to year. They found that such caps are associated with a smaller racial assessment gap.^[3] That result is useful for a civic audience precisely because it is undramatic. It says the gap is not a fact of nature, not something baked irreversibly into land or into race. It responds to the rules. Change a piece of the assessment machinery, in this case the speed at which assessed values are allowed to move, and the distribution of the burden moves with it. The corollary, which matters for everything that follows, is that the appeals process is also machinery. It is also a rule about how values get adjusted. If a growth cap can shift the racial gap, an appeals process can too, in either direction.

Cook County did not have to wait for the national literature to learn its own system was tilted, because the county got its own reckoning, and it arrived loudly. The Chicago Tribune investigation that became known as the Tax Divide series laid out, across 2017 and 2018, how the county's residential assessments had been systematically regressive under the prior administration. The reporting did something useful beyond the headlines. It prompted an independent study, commissioned through the county rather than the newsroom, and built on the industry standards that assessment professionals use to grade their own work, the ratio measures the International Association of Assessing Officers has long published. That study reached the same destination by the formal route. It confirmed that Cook County had been running, in the plain phrase that came out of it, a very regressive system, one that effectively transferred wealth from owners of lower-value homes to owners of higher-value ones.^[9] The mechanism was assessment error sorted by value, exactly the pattern Berry and Schleicher describe in the national data, but here documented inside Cook County's own parcels, with the county's own standards, by an inquiry the county itself had set in motion.

The standards matter to how the finding should be read, so they deserve a sentence. Assessment professionals do not grade an office by eye. They use ratio studies, which compare assessed values against actual sale prices across many parcels and then test whether the assessment-to-sale ratio stays level as you move from cheap homes to expensive ones. A level ratio is a uniform, fair assessment. A ratio that falls as prices rise, with cheap homes assessed at a higher fraction of their worth than expensive ones, is the technical signature of regressivity, and it has a name in the field, namely a price-related differential outside the accepted band. The county-commissioned study found Cook County outside that band, by the field's own yardstick, not by a journalist's. That is what gives the word regressive its weight in this context. It was not rhetoric. It was a measurement, taken with the instrument the profession itself trusts, and it came back failing.

The political aftermath was not quiet. The findings became a live issue in county politics, the prior Assessor lost his office, and a successor ran in part on the promise of fixing exactly the regressivity the study had documented. We mention the sequence not to take a side in county elections but because it bears on how to read our own 2003 data. The cycle we analyze sits years before the Tax Divide reporting and the reforms that followed it. Whatever our records show about 2003, they describe a county that had not yet been forced to confront its own ratio studies in public, which is one more reason to treat the snapshot as dated rather than current.

The scale of the transfer is what lifts it above a technical complaint. Berry's analysis of the Cook County numbers, reported alongside the Tribune's work, put the shift in the billions. Regressive assessments shaved something on the order of a billion dollars off the assessed value of the most expensive residential properties, and that burden did not evaporate. It moved. It landed on lower-value homes, with the total reallocation estimated at roughly two billion dollars over the years studied.^[10] A two-billion-dollar shift is not a rounding error in a system that pays for schools, parks, libraries, and county services. It is a quiet, machinery-driven redistribution running underneath the ordinary business of mailing notices and certifying values, and it ran for years before the reporting forced it into daylight. The money was real, the homes were real, and the people who paid more than their share to cover the people who paid less were real.

This is the structure the filer at the kitchen table is sitting inside, and from the kitchen table none of it is visible. They see one number. The literature sees the pattern those numbers make in aggregate, and the pattern is that the system, left to run, asked more of the people who had less and less of the people who had more. Which sets up the question this paper is built around. The appeal exists as a correction, a way for an owner to contest a number that is too high. So does the correction undo the tilt, or does it ride along with it.

Does filing an appeal level the field or tilt it

If assessments come out regressive, the appeal is the obvious place to look for a cure. An owner who has been over-assessed can contest the figure and bring it back toward the market, and if enough over-assessed owners did exactly that, the appeals process would behave like a corrective filter, draining error out of the system one parcel at a time. That is the optimistic reading, and it is not foolish. It is just incomplete. The research on appeals specifically, as distinct from assessments generally, complicates it in a way that turns out to matter.

Rachel Weber and Daniel McMillen studied appeals in Chicago directly, asking who files them and who succeeds, and the answer was uneven in a way that does real damage to the optimistic reading. Appeals are not filed at the same rate everywhere, and they are not granted at the same rate everywhere. Both the filing and the winning cluster, sorting unevenly across neighborhoods, and the consequence is the genuinely counterintuitive part of their work. When appeals are distributed unevenly, the appeals process can make assessment uniformity worse rather than better.^[5] The logic is plain once it is stated. If the owners best positioned to appeal, by information, by resources, by access to the right comparable sales, are not the same owners who were over-assessed in the first place, then granting their appeals pulls the system away from fairness even though each individual reduction looks, on its own, like a correction. The filter can run backward. Every cut it grants is defensible parcel by parcel, and the sum of those defensible cuts can still be a less uniform, less fair distribution than the one the office started with.

McMillen and Weber sharpened the picture with a study of what they called thin markets, meaning areas where few comparable properties change hands, so there is little recent sale data to anchor a value. In those thin-market areas, assessment-to-sale ratios swing to extremes in both directions, running high on some parcels and low on others, because there is less information to discipline the estimate. And those same areas show more appeals and more successful appeals.^[6] The link the authors draw is between information and outcome. Where the data are thin, values are noisier, and an owner who can marshal an argument, or pay a professional to marshal one, is better placed to convert that noise into a reduction. Appeal behavior tracks access to information, and access to information is not evenly spread. It is the same uneven distribution Weber and McMillen found across neighborhoods, observed from a second angle and confirmed by a second method.^[6] Two studies, two approaches, one finding. The appeal does not reach everyone the same way, and where it reaches furthest is not necessarily where the over-assessment was worst.

There is a quiet but important point buried in the thin-markets result, and it pushes back on a comforting story one might otherwise tell. The comforting story says that appeals correct errors, so wherever errors are largest, appeals should cluster and fix them. The thin-markets finding says the clustering is real but the correction is not symmetric. A noisy value can be wrong in either direction, too high or too low. An owner who has been over-assessed has every reason to appeal and bring the value down. An owner who has been under-assessed has every reason to stay quiet and keep the windfall. So in a thin market full of two-sided error, the appeals flow in one direction only, downward, pulling the over-assessed parcels back toward the line while leaving the under-assessed ones sitting comfortably below it. The result is not a market nudged toward accuracy. It is a market whose high errors get shaved and whose low errors are protected by the silence of the people who benefit from them. That is a built-in ratchet, and it works in favor of whoever was already paying too little, which in the regressivity literature tends to be the owner of the more valuable property. The appeal, in this light, is not a neutral correction device at all. It is a device that systematically removes the errors that hurt the well-positioned and preserves the errors that help them.

To see what these mechanics add up to in the present, rather than in the 2003 records we analyze further on, the Cook County Treasurer's office offers a recent and pointed accounting. In a 2025 report on the appeals system, the office reported that business appeals had secured 25.5 billion dollars in valuation reductions, which translated into something on the order of 3.26 billion dollars in tax relief for those businesses.^[8] Relief in a property tax system is rarely free, though, because the levy still has to be met. The taxing bodies set how much money they need, and the assessment process only divides that fixed need across parcels. So a dollar a business successfully appeals off its own bill is, in large part, a dollar that reappears on someone else's. The Treasurer's report estimated that the business reductions shifted roughly 1.9 to 2.0 billion dollars onto homeowners.^[8]

This is the feature of the property tax that makes the appeal a zero-sum contest rather than a private negotiation, and it is worth being explicit about the arithmetic because it is the engine under the Treasurer's numbers. A taxing district decides it needs, say, a fixed sum to run its schools. That sum does not fall when a shopping center wins a valuation cut. The bill is simply spread across a smaller total of assessed value, so the rate ticks up, and every parcel that did not win a reduction pays slightly more to cover the parcel that did. When the winners are concentrated, large commercial owners with the staff and the representation to appeal year after year, and the non-winners are diffuse, the households who did not appeal or appealed and lost, the shift has a direction. It runs from business to household, and within the household side it runs hardest toward the owners least able to claw any of it back. The Treasurer's distributional figures show exactly that direction. Across the years the report examined, households in the lowest income band, under 50,000 dollars, saw their bills rise by about 10 percent, while households over 150,000 dollars saw increases closer to 5 percent, and majority-Black and majority-Latino areas saw larger increases than the county as a whole over the 2021 to 2023 window.^[8] That is the uneven-relief pattern of the academic work, restated in the county's own current dollars, with the appeal sitting squarely at the center of it. The households at the bottom of the income distribution absorbed twice the percentage increase of the households at the top, and a meaningful part of what pushed their bills up was relief granted, through appeals, to commercial owners across the county.

Honesty requires the other side of the ledger, because Cook County is not a fixed story and reform has happened. A University of Chicago evaluation of the assessment reforms undertaken during the Kaegi administration found that residential regressivity fell substantially over the period studied. The bottom 70 percent of homes by value paid roughly 1.9 billion dollars less than they would have under the prior administration's approach, a reform-side counterweight to the harm the earlier reporting documented.^[11] We include this deliberately, and not as a throat-clearing nod to balance. A paper that cited only the damage and skipped the correction would be telling half the story and would deserve the skepticism such a paper earns. The point is that the appeals system the Treasurer criticizes and the assessment reforms the University of Chicago credits are operating in the same county at overlapping times. Both can be true at once. Assessments can grow fairer at the front end, where the office sets the initial value, while the appeals process at the back end still redistributes relief unevenly. Untangling how much of the recent change owes to fairer assessment and how much to shifting appeal behavior is exactly the kind of causal question our own data are not built to answer, and we will not pretend the question away by treating the reform and the appeal pattern as the same phenomenon.

So here is what our data can and cannot do, stated before we put a single figure of our own on the page. The literature has established, with race and income data we do not possess and with causal methods we did not use, that Cook County assessments were regressive and that appeal relief has flowed unevenly. Our file cannot test any of that. What our file can do is describe, for one cycle, the shape of the appeal outcomes themselves. It can show whether the win rate differs by the type of property, whether the people who file are the people who win, how large the cuts are when they come, and whether outcomes vary by place. Those descriptions cannot tell you why, and they carry no race or income field, so they cannot confirm the regressivity story on their own. They can only show whether the visible pattern of wins is consistent with a system that, as the literature argues, hands out its relief unevenly. That is a modest offering set beside the literature. It is also what we honestly have, so we turn to it now.

Whose appeals actually win a cut

Return to the kitchen table, and now multiply it. Our filer's yes-or-no outcome, the certified value coming back lower or not, is one of 4,962 such outcomes in the tax-year-2003 records, and gathered together those outcomes give the first real picture of how the remedy performs.^[1]

The topline first. Among the scored 2003 appeals, 1,621 of them, or 32.7 percent, won a cut in total assessed value. Some 3,334, or 67.2 percent, saw no change at all. And a bare seven, 0.1 percent, came back with the value raised.^[1] Put those three counts side by side and the dominant outcome of filing an appeal, by a wide margin, is that nothing happens to your number. Two filers in three got back the same value they had contested. The raise is vanishingly rare, seven cases in nearly five thousand, which is itself useful information for a homeowner weighing whether to bother, because it means the downside risk of filing was almost nil in this cycle. You were not, in any practical sense, gambling that the office would punish your nerve by raising the value. Filing was close to a free option, costless to exercise and almost never backfiring. But the upside was far from assured. The appeal is a contest most filers lose, and the headline rate of about a third winning sets the floor for everything that follows. Whatever sorting we are about to find happens on top of a baseline where losing is the norm and a no-change letter is what most people get back.

It is worth holding that baseline in mind against the way the appeal is often described, as a routine step a savvy owner simply takes to shave the bill. In this cycle it was not routine in its results. The savvy owner who filed faced two-to-one odds of getting nothing for the effort. That alone should make anyone cautious about treating appeal volume as if it were appeal success, a conflation the next section takes apart in detail. Filing and winning are different acts with different rates, and the distance between them is where the interesting structure lives.

The average hides the interesting part, and the interesting part surfaces the moment you sort the filers by what kind of property they brought to the office.

Land appeals won a cut 60.4 percent of the time, on a base of 96 appeals. Commercial appeals won 44.7 percent of the time, on 1,071 appeals. Residential appeals, the largest single group at 2,620, won 37.1 percent of the time. And condo and co-op appeals, 1,173 of them, won just 9.6 percent of the time.^[1] Hold the two ends of that range against each other and the distance is hard to ignore. A commercial appeal in this cycle was roughly 4.6 times as likely to win a cut as a condo or co-op appeal. Land, the small category at the top, won more than six times as often as condo and co-op, though we flag at once that 96 appeals is a thin base, and that land behaves strangely in ways the section on the size of the cut will make plain, so it belongs in the read as a small-sample descriptive figure rather than a typical case. The condo and co-op number is the one that stops a reader. Fewer than one in ten of those appeals succeeded, inside the same office, in the same year, under the same form, where more than four in ten commercial appeals did.

Notice that residential and condo or co-op are both, in the ordinary sense, housing. A condo is where a person lives. Yet the two housing categories sit at opposite ends of this range, with residential at 37.1 percent and condo or co-op at 9.6 percent, nearly thirty points apart.^[1] We do not, from this file, know why a condo appeal fared so much worse than a single-family residential appeal in 2003. The dataset does not record how condo assessments were set, whether by building or by unit, or how the comparisons were drawn, and we will not invent a mechanism to fill the silence. We only note that the gap is internal to housing, which means it cannot be waved away as a simple story of homes versus businesses. The sorting is finer than that, and our file shows the sorting without explaining it.

We should be careful about what this contrast is and is not. It is an association, a difference in success rates sitting in the records, sorted by a category the records do carry, namely the type of property. It is not a mechanism. We are not saying that being a commercial property causes an appeal to win, or that the office favors one class over another, or that condo owners simply file worse appeals. Any of those could be true. The literature gestures at reasons that uneven appeal success arises, chiefly the uneven access to information and representation that Weber and McMillen document.^[5][6] But our data cannot adjudicate among the possibilities, and we will not pretend they can. What we can say, flatly, is what wins. Commercial and land appeals win cuts often. Condo and co-op appeals mostly do not. Residential sits in between, nearer to commercial than to condo. Those are the rates, computed on the cycle, and they are what they are.

These rates sit beside the literature without overstepping the line we drew at the start. The published work argues that appeals relief flows unevenly and can redistribute rather than equalize. Our cut of the data shows relief flowing very unevenly across property types in one cycle, which is consistent with that picture. Consistent is the operative word, and we choose it on purpose. We have shown an unevenness the literature would predict, but we have shown it in property-type terms, not in race or income terms, because property type is what our file records and race and income are not. The reader who wants the racial and income reading of uneven relief should take it from the cited work, where the data and the methods exist to support it, and not from our four descriptive rates.^[5][8]

There is a dry observation to make here, and we will make it once and lightly. The appeal is the cheapest remedy in county government, free at the point of use and open to every parcel owner on the same nominal terms. And it still sorts its winners. The form costs nothing to file, yet the outcomes are distributed as though something were being charged that not everyone can pay. What that something is, our data do not say. That it shows up in the win rates, our data say plainly.

The win rate answers whether you win. It leaves two further questions that sharpen the picture considerably. One is whether the population that files appeals is the same population that collects the wins, because a group can file heavily and still leave with very little. The other is how large the cuts are when they land, because a win of 7 percent and a win that zeroes out a parcel are both counted as wins, and treating them as equal would erase a real difference. We take those in turn.

Are the people who file the same as the people who win

There is a second way to read the same records, and it cuts sharper than the win rate alone. Ask not just how often each group won, but how the total pool of wins was divided across the groups that filed. Who shows up at the office, and who walks out with a cut, turn out to be different crowds.

Condo and co-op owners filed 23.6 percent of all scored 2003 appeals, very nearly a quarter of everything that came through the door, and collected 7.0 percent of all the wins.^[1] Commercial owners filed 21.6 percent of appeals, a slightly smaller share of filings than the condo and co-op group, and collected 29.5 percent of the wins.^[1] Residential owners sat in between and a little above their weight, filing 52.8 percent and taking 59.9 percent of the wins.^[1] Land, tiny on both sides of the ledger, filed 1.9 percent and took 3.6 percent.^[1] These shares are computed on the 4,960 scored rows that carry an appeal type. Two rows in the cohort are missing it, which is why this denominator is two short of the topline's 4,962, a difference small enough to change nothing and worth stating anyway.

Line the condo and co-op figures up against each other and the asymmetry is impossible to miss. Filing close to a quarter of all appeals and leaving with one win in fourteen is not a rounding error and not a quirk of how the categories were drawn. It is the whole story of this section in two numbers. Commercial owners are the mirror image. They file about a fifth of the appeals and leave with nearly a third of the cuts. The plain takeaway, and we will not dress it up, is that the population that files and the population that wins are not the same population. The appeal does not simply ratify the choices of whoever bothers to file. It selects among them, and it selects hard.

The published work anticipates this shape exactly. Weber and McMillen's central finding was that an appeals process can redistribute relief rather than equalize it, so that the act of appealing reshuffles who carries the burden instead of returning the system to neutral.^[5] Our records are consistent with that shape, and the filing-versus-winning split is the cleanest version of it our file can produce. But we stop well short of their conclusion, because we have to. Their claim is about uniformity and, threaded through the later Treasurer work, about race and income.^[8] Our file holds no race field and no income field. It cannot tell us whether the condo and co-op owners who filed and lost were poorer, or whiter, or older, or anything else, than the commercial filers who won. That inference lives in the cited literature, and it stays there. What lives in our file is the bare arithmetic, that a group filing close to a quarter of the county's appeals walked away with a sliver of its cuts. The redistribution is visible in our numbers. Its demography is not, at least not here, and we will not borrow the literature's demography and pin it onto our spreadsheet to make the finding sound larger than it is.

It is worth pausing on why the two readings, the win rate and the share split, are not the same finding told twice. The win rate is a within-group statistic. Of the condo owners who filed, how many won. The share split is a between-group statistic. Of all the wins the office handed out, how many went to condo owners. A group could in principle have a low win rate and still collect a large share of wins, if it filed in overwhelming numbers. Condo and co-op owners did file in large numbers, nearly a quarter of the total, and still their slice of the wins collapsed to 7.0 percent, because their win rate was so low that volume could not rescue it. Commercial owners filed somewhat fewer appeals than condo and co-op owners did, yet their high win rate carried them to nearly a third of all cuts. The two statistics agree, and their agreement is the point. The group that wins rarely also wins little of the total, and the group that wins often also wins much of the total, even when it files less. The sorting compounds rather than cancels.

There is a reason to care about the property-type sorting on its own, even though it carries no demographic label, and it has to do with how the sorting connects to the literature's larger claim. The regressivity findings are ultimately about the distribution of a fixed bill. Someone pays less, someone pays more, and the question is who lands in each group. Property type is not race and it is not income, but it is not unrelated to them either, and more to the point it is the axis along which the relief in our data visibly concentrates. Commercial owners, the group that wins most and wins biggest, are the same broad category whose appeals the Treasurer's recent report ties to the multibillion-dollar shift onto homeowners.^[8] We are not collapsing the two findings into one. The Treasurer had income and geography and we do not. But our descriptive split shows the relief concentrating in the commercial category in 2003, and the Treasurer shows the cost of that concentration falling on households two decades later, and a reader is entitled to notice that the two pictures point the same way even though only one of them can prove its direction. Ours shows the funnel. Theirs shows where the water lands.

When an appeal wins, how big is the cut

Winning is yes or no. The cut is how much. Following our filer past the certified value means asking what a win was actually worth, because two appeals can both be wins and not be the same victory. Among appeals that won a reduction, the median size of that reduction in total assessed value varied by type about as sharply as the win rate did. Land winners saw a median cut of 99.6 percent. Commercial winners, 29.6 percent. Condo and co-op winners, 18.3 percent. Residential winners, just 7.0 percent, with the middle half of residential cuts running from roughly 4.6 percent to 12.3 percent.^[1]

Land first, because it needs quarantining before it distorts anything else. That 99.6 percent median is computed on 58 winners, and it reflects what those parcels are, mostly low-value or vacant land effectively zeroed out, brought down to a certified total near a single dollar.^[1] It is a real number and we report it as one, but it is a small-sample descriptive curiosity driven by extreme cases, not a model of what relief looks like for anyone with a building on their lot. Read it as the behavior of a strange tail of the data, not as a typical outcome. That is the last time land needs the warning label. It has now carried three.

The number that matters for the household story is the residential one, and it is modest. A homeowner who beat the odds, filed, and actually won a reduction took home a median trim of 7.0 percent off the assessed value.^[1] That is not nothing. On a real bill it is money, and for a household watching its costs it is worth the hour the form takes. But set it against the commercial winner's median of 29.6 percent and the shape of the thing comes into focus.^[1] The two groups are not playing the same game. A residential win is a haircut. A commercial win is closer to a re-valuation, a reduction four times deeper at the median, on a property that is very often worth far more to begin with, so the dollars behind the percentage are larger again. The interquartile range on the residential side, roughly 4.6 to 12.3 percent, tells the same story from inside the residential group.^[1] Even a residential winner near the top of the usual range, at about 12 percent, took a smaller proportional cut than the typical commercial winner did. The homeowner's best ordinary case still trailed the commercial median.

Even the condo and co-op winners, rare as they were, took a larger median cut at 18.3 percent than residential winners did.^[1] That is a genuine oddity and we flag it rather than smooth it over. It means the condo and co-op story splits in two. On the one hand, a condo or co-op owner was the least likely of anyone to win at all, at a 9.6 percent win rate. On the other hand, on the uncommon occasion that such an owner did win, the reduction was meatier than a house owner's, 18.3 percent against 7.0 percent.^[1] We do not have an explanation for the combination, and we will not manufacture one. It may have to do with how condo assessments were set or contested in 2003, or with which condo owners found it worth pursuing an appeal at all, but the file does not record any of that, so the honest report is the pairing itself. Hardest to win, larger when won.

A word on medians versus means, since the choice is not cosmetic and the gap between the two carries information. We report medians because the distribution of cuts is skewed, with a long tail of large reductions pulling the average up above the typical case. The means make the skew visible. Among residential winners the median cut is 7.0 percent but the mean is 13.4 percent, nearly double, which says a minority of residential winners took unusually deep cuts and dragged the average well above what the middle winner actually received.^[1] Commercial winners show a median of 29.6 percent against a mean of 32.4 percent, a gentler skew.^[1] Condo and co-op winners show the widest gap of the housing types, a median of 18.3 percent against a mean of 32.4 percent, which means the rare condo or co-op win was sometimes very large indeed and the average overstates the typical one badly.^[1] Land, predictably, runs the other way, a median of 99.6 percent against a mean of 69.6 percent, the median sitting above the mean because so many parcels were zeroed to near a dollar while a few took smaller cuts.^[1] We lead with the median in every case because it answers the homeowner's real question, what the person in the middle of the winners got, and we report the means here so the reader can see the shape of the distribution behind that middle. Quoting a mean as the headline would have inflated the residential and condo figures in particular, and inflation is the opposite of what this paper is for.

The discipline holds here as everywhere. The sizes differ by type. That is the finding, stated and finished. We make no claim about why commercial winners obtain deeper cuts, nothing about effort or legal representation or the quality of the comparable sales they were able to assemble, and nothing about whether any of it is fair beyond what the cited literature already supplies.^[1] If a dry observation is permitted, it is only this. The distance between a 7 percent trim for the homeowner who wins and a near-total write-down for the scattering of vacant land parcels is wide enough to see from the street. The file lets us measure that distance precisely. It does not let us account for it, and we have tried not to dress measurement up as explanation.

Does it matter which township you file in

One cut remains in the records, and it is geographic. Cook County assesses by township, the unit that also sets the calendar for when those late-summer notices go out, and our 2003 rows are thick enough in seven townships, the ones with at least 100 scored appeals, to compare first-stage win rates across place. The seven do not agree with one another.

The rate ran from 19.9 percent in Hanover to 44.6 percent in Barrington, a spread of 24.7 percentage points.^[1] In between sat New Trier at 39.4 percent, Northfield at 36.7, Palatine at 36.1, Wheeling at 28.7, and Schaumburg at 26.0.^[1] So a filer in Barrington won a first-stage cut more than twice as often as a filer in Hanover, at least in this one year and at least at this one stage. The spread is not subtle, and it is not the kind of thing that vanishes if you round more coarsely. From the bottom township to the top, the chance of walking away with a cut roughly doubled.

Now the hard caveat, and it is genuinely hard, the kind that constrains the finding rather than decorating it. This is a single year, a single stage, and seven townships, and the differences among them almost certainly reflect what kind of property sits in each township in 2003 far more than any difference in how the townships treat the people who file. We already know, from the section on property type, that commercial and land appeals win at high rates and condo and co-op appeals win at low ones. A township heavy in the categories that win will therefore post a higher first-stage rate for reasons that have nothing to do with hospitality toward appellants and everything to do with its mix of parcels. The North Shore and northwest suburban townships in this list each carry their own blend of housing, commercial strips, and vacant land, and that blend, not some local generosity, is the likeliest driver of the spread. We are not claiming Barrington is lenient and Hanover is stingy. We are reporting that in this sample, outcomes varied by place, and the variation was large.^[1] Those are two different sentences, and we mean only the second one. The first one, the tempting causal one about generous and stingy townships, is exactly the claim our data cannot support, and naming the temptation is the best way we have of refusing it.

The literature lends the geography some texture without lending us permission to draw a causal line. McMillen and Weber's thin-markets result implies that areas differ in how legible their property values are, and therefore in how appeals resolve, since a township with little recent sale data offers noisier values and more room for a well-argued appeal to find a cut.^[6] Weber and McMillen's broader finding is that appeal outcomes are not spread smoothly across a city in the first place.^[5] Our township spread is consistent with a world in which place matters to outcomes, both because the property mix differs by place and because the information environment differs by place. It is not evidence of why, and we have no intention of manufacturing any. Outcomes varied by township in this sample. We stop there, deliberately, with the door to causation left shut.

How much weight can one cycle of records carry

Following one appeal teaches you the mechanics. It does not teach you the trend, and being honest about what these records cannot do is the price of trusting what they can. This section gathers the limits in one place, not because we have hidden them until now, we have flagged each where it bit, but because a reader deciding how much weight to put on the findings deserves them assembled and unflinching.

Every rate and share in this paper rests on tax-year-2003 rows, which make up 99.2 percent of the shipped sample.^[1] That is one assessment cycle. It is not a time series, and nothing here should be read as a direction of travel. The patterns held in 2003. Whether they held in 1993 or in 2013, this file cannot say, and the recent Treasurer and University of Chicago findings about a changing county are a caution against assuming that a 2003 portrait still describes the present.^[8][11] The honest stance is that our numbers are a snapshot, dated, and that the county has demonstrably moved in at least some respects since the shutter clicked.

A win, as we have defined it throughout, is the Assessor's first-stage decision and only that, a certified total coming in below the mailed total.^[1] Cook County's appeals process does not end at the Assessor. An owner denied at the first stage can carry the fight to the Cook County Board of Review, and beyond that to the Illinois Property Tax Appeal Board, and many who lose at the first door win a cut behind one of the later ones. Our figures therefore understate cumulative success across the whole system, very possibly by a wide margin, because we are looking only at the first decision and not at the two that can follow.^[1] When we report that 37.1 percent of residential appeals won, it should be read as 37.1 percent won at the first stage. The lifetime figure, after the Board of Review and the state board have had their say, is higher, and we do not have it. This also means the uneven sorting we found is itself only a first-stage portrait. It is possible that the later stages widen the gaps between property types, or narrow them, and our file is silent on which.

A note on method, since it bears on every figure above. We scored wins with the value math, comparing certified total to mailed total, rather than leaning on the Assessor's own change flag, and we did so for cause rather than convenience. The flag is noisy. Cross-tabulating the flag against the actual movement of the grand total shows the problem in full. Of the rows the office flagged as a change, 1,601 did come in below the mailed total, a genuine win, but 111 came in equal to or above it, no win at all.^[1] Of the rows flagged no change, 3,151 indeed held steady, but 14 actually moved down.^[1] And a small set of rows, 85 in all, carried a blank flag while their values split 6 down and 79 unchanged.^[1] The 111 mislabeled rows are the instructive ones. In those cases the office moved value between the building line and the land line, recorded that internal shuffle as a change, and left the taxpayer's grand total exactly where it started. A flag that flips on internal bookkeeping while the bill in the mailbox does not move is the wrong instrument for measuring whether the taxpayer won. The value math measures the thing that lands in the mailbox, the grand total that drives the bill, and that is the thing we wanted to count. Had we trusted the flag instead, we would have credited those 111 rows as victories and overstated the win count accordingly, which is exactly the kind of small, defensible-looking error that compounds into a misleading headline. We print the cross-tab rather than ask the reader to take the choice on faith.

Land, one last time, is small and extreme, 96 filings and 58 winners with cuts driven by near-zero parcels, and belongs in the read as a descriptive tail rather than a rate anyone could plan around.^[1] We have kept it in the figures because removing inconvenient categories is its own kind of dishonesty, but we have labeled it at every appearance so that no reader mistakes the behavior of 58 vacant lots for the behavior of the system.

And the limit that matters most, the one we have flagged at every turn and will flag once more here, is that this file carries no race variable, no income variable, and no sale-price variable.^[2][7][8][9] It can show who filed by property type and who won. It is structurally silent on whether those filers were richer or poorer, whiter or not, on whether the homes were over- or under-assessed relative to what they would fetch on the market, which is the very quantity the regressivity literature is built on. Every claim in this paper that the appeals system or the underlying assessment is regressive by race or by income comes from the cited literature and from nowhere in our spreadsheet.^[2][7][8][9] We can show the sorting. We cannot, from this file, name what it sorts on. The sale-price gap is worth dwelling on for a moment, because it is the cleanest illustration of the boundary. Berry, Schleicher, and Avenancio-Leon and Howard all build their case on the ratio of assessed value to sale price, the assessment-to-sale ratio, which is the standard tool for detecting regressivity.^[4][7][2] Our file has no sale prices. We literally cannot compute that ratio. So we could not test for regressivity in our own data even if we set the race question aside, and any reader who wants the regressivity result must take it from the work that has the prices, not from us.

It is fair to ask, then, what a file would need to carry for a paper like this one to make the claims we have refused to make. The list is concrete, and naming it is more useful than gesturing at limitations in the abstract. To test regressivity directly, the records would need a sale price or an independent market value for each parcel, so the assessment-to-sale ratio could be computed and checked for the downward tilt the literature describes. To carry the racial reading, the records would need a way to link each parcel to the race composition of its owners or its block, which in practice means joining the parcels to census geography or to deed records, with all the privacy care that implies. To measure true success rather than first-stage success, the records would need each appeal followed through the Board of Review and the Illinois Property Tax Appeal Board, so a denial at the first door could be traced to a win or a loss at the later ones. And to speak about trends rather than a single snapshot, the file would need many cycles, not one, so the 2003 picture could be set against the years before and after the Tax Divide reforms. None of those joins is impossible. Each has been done in some form by the researchers we cite, who had the resources and the data agreements to do it. We did not, and rather than approximate any of it with guesses, we have left those questions to the work that answered them properly and confined ourselves to what 5,000 honest rows can actually support.

So the data carry exactly this much. A descriptive portrait of one cycle, first stage only, by property type and township, with the magnitudes of the cuts attached. They carry no trend, no final cross-stage outcome, and no demography. Claiming more than that would be inventing, and not inventing was the entire discipline of the exercise.

So who wins the Cook County property tax appeal

We opened with a household at a kitchen table, a mailed notice with a number that felt too high, and a free form that could be filed to contest it. The title asked who wins when that act is repeated across a county, and at the level our evidence allows, here is the answer our own records support.

Across one 2003 cycle, the filers who won cuts most often, and won the largest cuts when they did, were commercial and land owners, not the condo, co-op, and residential owners who together make up the household side of the ledger.^[1] The population that files and the population that wins are not the same, with condo and co-op owners filing nearly a quarter of appeals and collecting a single win in fourteen.^[1] And the odds of a first-stage cut depended on which township the parcel sat in, by a 24.7-point spread.^[1] All of that is description. None of it is a claim about cause. And none of it, from our file, is a claim about race or income, however much the property-type sorting may rhyme with the demographic sorting the literature documents.

The larger interpretation belongs to the literature, which we have kept at arm's length from our own numbers on purpose, and which is clear where we are silent. Cook County's assessments were regressive. The burden tracked the value of the property and, through the long entanglement of value with neighborhood and neighborhood with race, it tracked race as well.^[2][9] The appeals process has tended to redistribute relief toward those best equipped to claim it rather than equalize it, with billions of dollars in present-day stakes, roughly 1.9 to 2.0 billion shifted onto homeowners in the most recent accounting.^[7][8] The reform record is part of the honest picture too, since the assessment side has been pulled measurably back toward fairness in recent years, by something near 1.9 billion dollars for the bottom 70 percent of homes.^[11] The mechanism our filer used, though, the appeal itself, is the part the most recent work finds still sorting its winners.^[8]

Which leaves the civic point, and it is a plain one, the same one we flagged in the opening and have now earned the right to state. A remedy can be free, and universal, and genuinely available to the homeowner at the kitchen table, and still not be equal in what it returns. Our records show a free remedy producing unequal results across property types and across places within a single year. They cannot tell us who, by race or income, those results finally fell on, and that is not a small gap to leave open. The questions that would close it need data this file does not hold, records that link each filer to race, to income, to the sale price that would reveal whether the home was over-assessed in the first place, and to the final outcome after the Board of Review and the state board have ruled. Until someone assembles that file and runs it, the most we can say honestly is what we have said. About a third of the people who asked got a cut. The cuts were not evenly distributed, not across property types and not across townships. And a form that costs nothing to file still produced winners and losers it never announced on its face. The reproducible files behind every number above are posted for anyone who wants to check the arithmetic and, better still, to push past where our data stop.^[1]

^[1]: Author analysis of the dataset. Reproducible files at rooted-forward.org/research/data/cook-county-property-tax-appeal-disparity.

^[2]: Avenancio-Leon, C. F., & Howard, T. (2022). The Assessment Gap. Racial Inequalities in Property Taxation. The Quarterly Journal of Economics, 137(3), 1383-1434.

^[3]: Avenancio-Leon, C. F., & Howard, T. (2022). Assessment Caps and the Racial Assessment Gap. National Tax Journal, 75(1), 169-200.

^[4]: Berry, C. R. (2021). Reassessing the Property Tax. SSRN Working Paper (Univ. of Chicago Harris / Mansueto Institute).

^[5]: Weber, R. N., & McMillen, D. P. (2010). Ask and Ye Shall Receive? Predicting the Successful Appeal of Property Tax Assessments. Public Finance Review, 38(1), 74-101. (Lincoln Institute working paper, 2006.)

^[6]: McMillen, D. P., & Weber, R. N. (2008). Thin Markets and Property Tax Inequities. A Multinomial Logit Approach. National Tax Journal, 61(4 Part 1), 653-672.

^[7]: Schleicher, D. (2025). Your House Is Worth More Than They Think. The Strange Case of Property Tax Regressivity. Harvard Journal on Legislation, 62(1).

^[8]: Cook County Treasurer's Office (Maria Pappas) (2025). A Broken Property Tax Appeals System. Cook County Treasurer.

^[9]: Grotto, J., & others (2018). Cook County's Residential Property Tax Assessments Deeply Unfair, Independent Study Confirms (Civic Consulting Alliance / IAAO-standard study commissioned after the Chicago Tribune Tax Divide series). ProPublica Illinois.

^[10]: ProPublica Illinois / Chicago Tribune (2018). Flawed Assessments Caused $2 Billion Shift in Property Taxes, Study Finds.

^[11]: Cook County Assessor's Office (2024). Cook County Homeowners Saved $1.9 Billion Under Fritz Kaegi's Reforms, University of Chicago Study Finds.