What the Affordable Requirements Ordinance Has Produced and Where

Abstract

Chicago's Affordable Requirements Ordinance was meant to fold affordability into the city's building boom, yet the units it has yielded sit mostly where land is already dear. This paper joins a reading of the inclusionary-zoning literature to our own descriptive analysis of the City of Chicago Department of Housing's affordable-rental roster, the public file that flags which developments carry ARO units. We make no causal claim, ran no experiment, and modeled no counterfactual. We count, rank, and cross-tabulate what the file holds, which is 156 ARO-tagged developments and 1,727 affordable units across 24 of Chicago's 77 community areas ^[1]. ARO is about a quarter of the listed affordable-rental developments but under six percent of the listed units, because each set-aside is a small pocket inside a market-rate building ^[1]. Those developments concentrate in higher-cost North Side, Northwest Side, and near-downtown community areas, while historically disinvested South and West Side neighborhoods carry long affordable rosters built under other programs and almost no ARO ^[1]. Median set-aside size is five units ^[1]. Published research on inclusionary zoning anticipates much of this, tying output to program design, tenure, and program age rather than to housing need ^[2][3][5]. We are direct about what the file cannot show, which is fees collected, units forgone, rents, affordability depth, and any trend over time, and we read the roster as a point-in-time snapshot rather than a complete census of the ordinance. Where we report a number as our own, it comes from that roster and nowhere else ^[1]. Where a number is borrowed, it is attributed to the work that produced it.

A Promise Made Citywide, Units Booked in a Few Neighborhoods

The Affordable Requirements Ordinance arrived with a citywide framing. The plan was to ride the private development Chicago was already permitting and peel off a share of it for households the market was pricing out, so that wherever cranes went up, some affordability would follow. That framing is geographic by design. A mandate written for the whole city carries the implication that its product will appear, in some measure, across the whole city. The City's own roster of standing ARO rental units tells a narrower story. Those units cluster in a short list of higher-cost neighborhoods, and most of the South and West Side carries none.

We can state the contrast with our own counts and nothing borrowed. The City of Chicago Department of Housing's affordable-rental file flags 156 developments as ARO, and those developments hold 1,727 affordable units ^[1]. The 156 developments sit in just 24 of Chicago's 77 community areas ^[1]. By this file, 53 of the city's 77 community areas, more than two thirds, host no listed ARO building at all. The concentration runs deeper than the area count suggests. The top five community areas by ARO unit count hold 1,277 of the 1,727 units, which is 73.9 percent of every standing ARO rental unit the file records ^[1]. A program written for seventy-seven neighborhoods has booked nearly three quarters of its listed units in five of them.

That gap frames the paper. The ordinance promised citywide reach, and the file shows a concentrated footprint. We want to be careful about how much weight that sentence can carry, because a roster of standing units is not the same thing as a record of everything a policy has set in motion. A building enters this list because, at the moment the file was compiled, it carried units the City tracks, which is a narrower fact than the program's full history. Even so, the pattern is plain in the public data, and it is the pattern most worth examining, because it bears directly on who lives near the affordability the ordinance has produced.

The phrase "wherever cranes went up" hides the trouble, because in Chicago the cranes have never been spread evenly. The city's housing geography was set long before the ordinance, by decades of policy that decided where credit flowed and where it did not. In the 1930s the federal Home Owners' Loan Corporation graded Chicago neighborhood by neighborhood and marked the Black South and West Sides in red, instructing lenders to treat those areas as hazardous, and the appraisal practices that followed kept private mortgage money out of them for a generation. Richard Rothstein's account of how federal housing policy wrote segregation into law, rather than letting it form by accident, is the standard treatment of how that machinery worked across American cities. On the West Side in the 1950s and 1960s, as Beryl Satter reconstructed in detail, contract sellers exploited the credit vacuum directly, selling homes to Black families on terms that built no equity and could be voided by a single missed payment. By the time the ARO was written in 2003, the parts of Chicago where private capital readily financed new apartment towers and the parts where it did not had been sorted for seventy years. A mandate that rides on private construction inherits that sorting whether or not anyone intends it to. The geography in the City's file is, in part, the long shadow of the redlining map, refracted through an ordinance that attaches to the kind of building the redlined neighborhoods were starved of.

Two questions sit inside that gap, and they are the only two this file answers cleanly. One is where the units sit. A roster of registered developments, each carrying a community area and a unit count, is built to answer exactly that. We can rank neighborhoods, tally the areas with zero ARO, and put the concentration in numbers. The other is how many units there are. The same file gives a unit count for nearly every development, so we can total them, describe their per-building size, and weigh ARO's share of developments against its share of units. Where, and how many. Those two questions are the spine of everything we report.

A great deal that matters about the ordinance lives outside those two questions, and outside this file. The ARO lets a developer pay a fee instead of building units on site, and the dollars collected through that option are a central part of any honest accounting. So are the units a developer might have built had the fee option not existed, the affordability depth of the units that do get built, the income tiers they serve, and the rents tenants actually pay. None of that sits in the roster. We flag it here at the outset so the reader knows where the boundary runs. When this paper discusses fees, units forgone, rent effects, or the cumulative built-or-promised total, it is reporting what other researchers found and crediting them, not summing it from our file. Where we present a figure as ours, it comes from the roster and only the roster ^[1].

The register of this paper is even-handed, and we mean that as a working discipline rather than a pose. The siting pattern we describe is not, on its own, an indictment of the ordinance. A policy that attaches to private market-rate construction will tend to produce units where that construction happens, and in Chicago that has meant the North Side, the Northwest Side, and the blocks near downtown. Concentration of that kind is partly mechanical, an artifact of how the instrument is built rather than proof it has failed. The pattern is not a defense either. That the units landed where building was easy does not answer whether the city won a fair share of affordability for the development it approved, or whether the neighborhoods that most need affordable housing saw any of it. We hold both readings open because the data supports holding both open, and we resist the pull to collapse them into a verdict the file cannot deliver.

The reason the where question is worth this much care is that location is most of what affordable housing buys beyond shelter. An affordable apartment in a neighborhood with good schools, frequent transit, low violence, and a full-service grocery store is a different good than the same rent in a neighborhood that the city spent a century starving of all four, even when the unit and the rent are identical. The fair-housing tradition treats access to opportunity-rich neighborhoods as the thing segregation denied and the thing integration is meant to restore, and inclusionary zoning has been argued, by some of its defenders, as a tool that places affordable units precisely in the appreciating neighborhoods low-income families would otherwise be priced out of. So a map of where ARO units land is not bookkeeping. It is a map of which families the program lets into which neighborhoods, and that is the register in which the University of Chicago equity study reads the ordinance when it asks what the program has meant for Black Chicagoans ^[9]. We keep our own claims descriptive, but we want the reader to understand why the geography carries the weight it does.

What follows moves from framing to evidence and back. We say plainly what this study is and is not, read the ordinance's mechanics against the published research on inclusionary mandates, and then turn to four descriptive findings drawn from the roster, each a different angle on the same gap between a citywide promise and a concentrated set of booked units. We close by setting our picture beside what the literature would predict and naming the questions a fuller public record could answer.

What This Paper Is, and What It Is Not

This is a synthesis of published inclusionary-zoning research joined to our own descriptive analysis of one real public dataset. The dataset is the City of Chicago Department of Housing's affordable-rental roster, pulled from the Chicago Data Portal, the city's Socrata-hosted open-data platform ^[1]. That file is the spine of our original numbers. Everything we present as a finding of ours is a count, a share, a ranking, or a cross-tabulation of rows in that file. Everything we present as a finding of someone else's is attributed in the text to its source.

It helps to be concrete about the work we did and did not do. We read the roster into a standard data tool, counted its rows, grouped them by community area, summed and described their unit counts, and cross-tabulated ARO developments against the rest of the affordable-rental list. That is the whole of it. We did not interview developers, aldermen, or housing officials. We audited no fee account and traced no dollar of in-lieu revenue. We built no statistical model, fit no regression, and estimated no counterfactual. At no point did we try to recover what Chicago's housing market would look like without the ordinance, because nothing in a roster of standing buildings supports that kind of inference. The reader should treat our contribution as careful description of a public file, and should read any causal-sounding claim in these pages as belonging to the literature we cite rather than to us.

The provenance of the file is worth stating exactly, because a descriptive paper is only as good as the source it describes. The roster is the City of Chicago Department of Housing's Affordable Rental Housing Developments resource, published on the Chicago Data Portal, the city's open-data platform hosted on Socrata ^[1]. Each row is one development, with fields for the community area and community-area number, the property type, the property name, the address and ZIP code, the management company, the unit count, and a set of geographic coordinates ^[1]. The property-type field is the one that makes this analysis possible, because one of its values is the literal string ARO, which is how the City flags a development as carrying inclusionary units. We treated that flag as our definition of an ARO development and counted it case-insensitively so that a stray capitalization could not drop a real row ^[1]. Everything downstream, the share calculations, the rankings, the size distribution, and the cross-tabs, rests on that single field, so we checked it against the live version of the dataset rather than trusting the local copy alone. The field schema in the repository file matches the live Socrata endpoint, which returns the same fourteen columns, and the record count agrees across both ^[1]. We mention the parallel live feed not because we queried it for new numbers but because its agreement with the static file is part of why we trust the static file.

Before we report a single finding, one correction comes first, because it governs how the rest of the numbers should be read. A naive look at the file suggests it holds 1,776 developments, and that figure even surfaced in our own early planning notes. It is wrong. The true count is 598 developments ^[1]. The inflation comes from the file's geometry field, a location column that stores each development's mapped point in a format that embeds line breaks. A tool that counts physical lines, such as the Unix wc -l utility, reads those embedded breaks as new rows and overcounts badly. Parsed properly, with Python's csv module treating each record as a record rather than each line as a row, the count resolves to 598, and the live Socrata endpoint's own record count agrees at 598 ^[1]. We lead with this because every share we report leans on it. ARO's quarter-of-developments figure and its sliver-of-units figure are both computed against the real denominators, 598 developments and 29,550 units, not against the inflated line count ^[1]. A reader who carried the 1,776 figure forward would misread every ratio in the paper.

A second clarification matters as much, and it concerns what kind of file this is. The roster is a point-in-time snapshot of buildings that have registered affordable units with the City. It is the Department of Housing's working list of rental developments, not a cumulative ledger of everything the ordinance has ever produced or pledged. A development appears because, at the moment the file was compiled, it carried units the City tracks. That has a direct consequence for our headline count. Our 1,727 ARO units is the number standing on the roster, and it sits below the built-or-promised totals other researchers report. A critical 2025 report from the Illinois Policy Institute puts ARO at roughly 2,798 units built or promised over twenty-one years ^[8]. The two numbers do not conflict, because they count different things. A snapshot of currently listed units will fall below a cumulative tally that folds in pledges, units in the pipeline, and developments that have since dropped off the roster. We name the difference here so that no reader takes our 1,727 as a contradiction of the larger figure. It is a narrower measure, drawn from a narrower file, and the 2,798 stays attributed to its source throughout.

There is a smaller piece of housekeeping worth stating in the same breath, because it shapes one of the four findings. Of the 156 ARO developments, 155 carry a reported unit count and one does not ^[1]. When we describe the size of ARO set-asides, the median, the mean, the range, and the distribution, we compute those statistics on the 155 developments that report a count, and we say so each time. We did not impute a value for the missing row, and we did not drop it from the development count. It is one row, reported plainly, and it moves none of the conclusions.

The reason for laying all this out before the evidence is trust. A reader who knows exactly which numbers are ours, which are borrowed, what the file is, and where it miscounts can weigh each finding for what it is worth. We would rather spend the words here, in front, than have a reader stumble on the snapshot caveat or the line-count correction halfway through and wonder what else went unsaid. The discipline of this paper is to claim less than the data might tempt us to claim, and the place that discipline starts is here.

How the Ordinance Works and What Researchers Have Found About Mandates Like It

To read the roster well, it helps to know what the ordinance asks of developers and what scholars have learned about instruments like it. The mechanism first, kept brief and drawn from the civic record.

Chicago adopted the Affordable Requirements Ordinance in 2003 and amended it through 2015, with the Metropolitan Planning Council's case study serving as the standard civic summary of that history ^[7]. The core idea is a set-aside. When a residential project crosses certain thresholds, typically because it receives city financial assistance, a zoning change, or a land purchase from the City, the ordinance requires that a share of its units be affordable ^[7]. That trigger is worth dwelling on, because it defines the program's reach. The ARO does not attach to every new apartment building in Chicago. It attaches to projects that ask the City for something, a rezoning, a subsidy, a parcel of public land, and in exchange the City asks for affordability back. A project that needs none of those things and builds entirely as-of-right on privately assembled land can sidestep the requirement altogether. The set-aside share is the heart of the policy, and it is also the reason the units arrive in small pockets, because a percentage of one building's units is, for most buildings, a handful of homes.

The ordinance pairs the set-aside with an alternative, a tiered in-lieu fee that lets a developer pay into a city fund instead of building the affordable units on site ^[7]. The fee schedule varies by location, with higher fees in higher-cost areas, an attempt to reflect what on-site affordability would have cost in that part of the city ^[7]. The logic of the tier is straightforward. A foregone affordable unit in a neighborhood where market rents are high represents a larger public loss than the same unit foregone where rents are low, so the fee tracks the local cost of buying affordability out of the deal. The MPC summary lays out this history along with the projected unit and revenue targets the City attached to the program at adoption ^[7]. The fee option is central to how the ordinance behaves, because every dollar paid is a unit not built on site, and it is exactly the part of the story our roster cannot see. A building that satisfied its obligation entirely by writing a check leaves no affordable units to register, so it never appears on the list we count, which is one of the cleanest illustrations of why a roster of standing units is a partial record of what the ordinance has done.

The history behind the 2015 revision is worth a sentence, because it explains why the ordinance grew teeth. The original 2003 version leaned heavily on the fee option, and for years developers chose to pay rather than build, which meant the ordinance's footprint in actual on-site units stayed thin while a fund accumulated. The 2015 rewrite raised the in-lieu fees, sorted the city into pricing zones that demanded more on-site units in higher-cost areas, and was followed by a 2017 pilot that pushed further in the same direction ^[10]. The 2021 round tightened the requirements again. That trajectory matters for reading the geography, because the file we analyze is a present-tense snapshot that mixes buildings produced under the looser early rules with buildings produced under the stricter later ones, and it carries no dates that would let us separate the two.

With the mechanism in view, the published evidence on inclusionary zoning gives a frame for what to expect, and the frame holds across studies even though the studies look at different places.

The anchor is a body of work by Jenny Schuetz, Rachel Meltzer, and Vicki Been. Their study with the pointed title asking whether inclusionary zoning is a silver bullet or a Trojan horse reaches a measured answer. It is neither, and its effects on housing prices and on production are market-dependent and tied to how long a program has been in place ^[2]. The framing of the question is itself instructive. The silver-bullet hope is that an inclusionary mandate quietly solves affordability at no cost to the public budget, since the units come out of private projects. The Trojan-horse fear is that the mandate functions as a hidden tax on construction, raising prices and choking supply. Schuetz, Meltzer, and Been find the truth sits between those poles and shifts with the local market, which is the most useful thing one can say to a city deciding whether to adopt or tighten such a policy ^[2]. The phrase worth carrying forward is no silver bullet. Inclusionary zoning, in their reading, is a modest instrument whose output depends heavily on local conditions and on the maturity of the program, not a lever that reliably delivers large numbers of affordable units ^[2].

Their companion cross-city work, comparing the many varieties of inclusionary zoning across San Francisco, Washington, and suburban Boston, documents how widely these programs differ from one another ^[3]. The title's count of thirty-one flavors is not a joke. It is the point. Two jurisdictions can both say they have inclusionary zoning and mean almost entirely different things by it, because the design parameters that govern output vary across nearly every dimension. Whether a program is mandatory or voluntary, how large its set-aside share is, how it sets income targets, and how its fees are structured all drive how much affordable housing gets built ^[3]. Design is not a detail in their account. It is the variable that explains the spread in results. That finding is the reason we keep returning to design when we read Chicago's numbers, and it is the reason a single descriptive snapshot of one city's output cannot be read as a verdict on inclusionary zoning in general. Chicago's ARO is one flavor among many, and its results reflect its own parameters and its own market.

The empirical literature reinforces the modest-output expectation. Vinit Mukhija and colleagues examined inclusionary zoning across Los Angeles and Orange Counties and found it produced only modest unit counts, with no measurable effect on overall housing supply ^[4]. Their work is a useful caution against expecting an inclusionary mandate to move the supply needle at the metropolitan scale. The units it generates are real, but in their California cases they were few, and they did not register as a detectable change in how much housing got built overall ^[4]. That second finding matters as much as the first. One of the standing arguments against inclusionary requirements is that they suppress production by making projects pencil out worse, and one of the standing arguments for them is that they meaningfully add to the affordable stock. Mukhija and colleagues found neither effect at a detectable scale in their two counties, which suggests the policy is, in practice, a smaller intervention than either its boosters or its critics often claim ^[4]. A reader who expects ARO to have reshaped Chicago's housing supply, in either direction, is expecting something the comparative evidence does not lead us to expect from this kind of policy.

If the literature stopped there, it would read as uniformly skeptical, but it does not. Ruoniu Wang and Xianghong Fu took up the design question directly and asked which program features track with more affordable units. Their answer is specific. Mandatory programs produce more than voluntary ones. Older programs, with more years to accumulate output, produce more than newer ones. Jurisdiction-wide programs produce more than narrowly scoped ones. Programs with stricter income targeting produce more than loose ones ^[5]. Stringency, in their finding, tracks with production. This is why the trajectory of Chicago's revisions matters. An ordinance that grows more demanding over time is, by Wang and Fu's logic, an ordinance that should track with more on-site units than its earlier, looser form ^[5]. We cannot test that claim with our file, because the roster carries no dates, but the literature explains why the revisions are not a footnote.

Layered on top of the national literature is a set of Chicago-specific findings that address this ordinance directly, and they cut in more than one direction. The City's own watchdog weighed in first. In 2017 the Chicago Office of Inspector General audited the administration of the ordinance and found that the City did not properly account for ARO and density-bonus fees, nor did it follow administrative best practices, in ways the audit tied to weaker outcomes ^[6]. That finding sits at the center of the fee question, and it deserves to be read for what it is. The OIG was not arguing that the ordinance is a bad idea. It was finding that the administration of the in-lieu fees fell short of the standard the City should hold itself to, which is a narrower and in some ways more damning conclusion. A policy can be sound in design and still leak value through weak bookkeeping. If the agency administering the in-lieu option was not tracking the dollars well, then the public record of what those fees bought is itself shaky, which is one more reason we treat fees as outside what we can responsibly count. The audit also looked at geographic outcomes and at how the in-lieu option was used, which makes it one of the few official documents that engages the same siting question we approach from the roster, and it reached its conclusions with access to the fee records we do not have ^[6]. A University of Chicago study took up the equity question and reached a mixed verdict, finding that the ordinance delivers some equity for Black Chicagoans but does so only at a small scale, with integration outcomes the study treats as contested rather than settled ^[9]. Small scale, contested outcomes. The ordinance is doing something for equity in that account, but not at a scale that resolves the segregation it engages, and whether the units it places actually integrate neighborhoods remains, in that study, an open question.

A piece of independent data work rounds out the Chicago picture and speaks directly to the geography we are about to examine. Rashada Graham-Bailey digitized the Department of Planning and Development's quarterly ARO reports for 2016 through 2018 and analyzed them at the ward level, where the concentration is severe. The 27th Ward alone triggered 2,690 ARO units in that analysis, a single ward accounting for a volume of units that dwarfs most of the city ^[10]. The 27th covers the West Loop, Fulton Market, and the river corridors just west and northwest of downtown, which were among the most active apartment-construction zones in Chicago over those years, so the ward total is the development boom and the ordinance arriving in the same place at the same time. That same work found that the stricter 2015 rules, together with the 2017 pilot, materially increased on-site unit production, which is the Wang and Fu stringency finding turning up in Chicago's own numbers ^[10]. We lean on this study twice below, once for the concentration it documents and once for the production lift the stricter rules produced, because it is the closest thing in the literature to the descriptive geography we run ourselves.

The counterweight comes from the Illinois Policy Institute's 2025 report, which is openly critical of the ordinance and which we present as the critical perspective it is. That report puts ARO at roughly 2,798 units built or promised over twenty-one years, about 140 units a year and on the order of 2 percent of new housing stock over the period, alongside more than 200 million dollars in in-lieu fees ^[8]. The report reads those figures as evidence that the ordinance has delivered limited, concentrated output while pushing costs into the development it touches ^[8]. We do not adopt the report's interpretation, and we do not treat its built-or-promised total as interchangeable with our roster count, but the figures themselves are the most complete public tally of cumulative output and fees, and they are the right backdrop for our narrower snapshot. The 2,798 units, the roughly 140-per-year pace, the 2 percent share, and the 200 million dollars in fees all belong to that report and stay attributed to it throughout these pages ^[8].

Pulled together, the literature points one way. Inclusionary mandates produce modest unit counts whose size tracks how the program is designed, how mature it is, and how it handles tenure and income targeting ^[2][3][4][5]. In Chicago specifically, the units are concentrated, the equity gains are real but small and contested, the fee accounting has been weak, and the stricter rules lifted on-site production ^[6][9][10]. That is the expectation we carry into the data. The roster should show modest, design-sensitive, concentrated output, and the value of running our own numbers is to see, in the City's own file, exactly where that concentration falls and how small the pockets are. The findings that follow are descriptive, and they line up with what the literature would predict, which is the most honest thing we can say about them in advance.

A Quarter of the Developments, a Sliver of the Units

The first thing the roster reveals is a split between how often ARO appears and how much housing it accounts for. ARO shows up frequently as a development and rarely as a unit, and the gap between those two facts is the key to reading everything that follows.

The numbers are these. Of the 598 affordable-rental developments the file lists, 156 are tagged ARO, which is 26.1 percent of the developments ^[1]. Just over a quarter of the City's listed affordable-rental developments carry the ARO label. Turn from developments to units and the picture changes sharply. The file's 598 developments hold 29,550 units in total, and the 156 ARO developments hold 1,727 of them, which is 5.8 percent of the units ^[1]. ARO is a quarter of the developments and under a sixteenth of the units. A reader who knew only the development share would badly overestimate how much housing the ordinance accounts for. A reader who knew only the unit share would underestimate how often it appears across the City's affordable portfolio. Both numbers are true, and they describe the same set of buildings from two angles.

The arithmetic behind the split is not mysterious, and it follows directly from how the ordinance works. An ARO set-aside is a small pocket of affordability carved inside a larger market-rate building. The ordinance takes a percentage of a private project's units, and a percentage of one building is, in most cases, a modest number of homes. So each ARO development contributes a small unit count to the roster while still counting as one development. Multiply that across 156 buildings and the result is exactly what the file shows, many developments contributing few units each.

The non-ARO developments behave differently, and understanding that group is worth a moment because it is the backdrop against which ARO's smallness reads. That group takes in Chicago Housing Authority properties, dedicated multifamily affordable buildings, senior housing, supportive housing, and single-room-occupancy buildings, and many of those are large, purpose-built affordable developments where most or all of the units are affordable ^[1]. The history of how those buildings came to exist is the history of twentieth-century affordable housing in Chicago. The CHA properties on the list are what remains and what replaced the high-rise system the Plan for Transformation began demolishing in 1999, when the agency set out to tear down tens of thousands of units of distressed public housing and rebuild at lower density. The dedicated affordable and supportive developments came up through federal tax-credit financing, project-based vouchers, and nonprofit development, programs built to produce whole buildings of affordable units rather than slices of market-rate ones. A single CHA mid-rise or a large supportive-housing building can carry more units than a dozen ARO set-asides combined. The non-ARO group is 442 developments holding 27,823 units, and its developments are, on average, far larger in affordable-unit terms than the ARO pockets ^[1]. That size difference is the whole of the explanation. ARO is common but small, and the rest of the roster, built by programs designed to produce affordability in bulk, is built at a scale that dominates the units.

We want to be careful about what this comparison is and is not, because it would be easy to overread it. The 5.8 percent unit share is a description of this roster, not a verdict on the ordinance's worth. The non-ARO group is heterogeneous. It mixes federal public housing with privately operated affordable buildings, senior developments with supportive housing, each built under its own program with its own funding and its own rules. Comparing ARO's unit share against that mixed group tells us how ARO sits within the City's listed affordable-rental portfolio, and nothing more. It does not tell us whether ARO is an efficient use of the development it touches, whether its units run deeper or shallower in affordability than the alternatives, or whether the city won fair value for the zoning and assistance it granted. Those are real questions, and they are precisely the questions the literature takes up and our file cannot. A small unit share is consistent with a modest but worthwhile program. It is equally consistent with a program critics judge too thin. The roster alone does not adjudicate between them.

One technical point about the comparison group deserves a plain statement, because it bears on how much weight the cross-tabulations later in the paper can carry. The property_type field that distinguishes ARO from everything else is free text, and the non-ARO values contain spelling variants, including misspellings of multifamily that we observed in the raw file ^[1]. We did not merge those variants into clean categories. For the ARO finding this does not matter, because the ARO tag itself is consistent and we counted it case-insensitively to be safe ^[1]. It matters only if a reader wants to slice the non-ARO group into its sub-types, which we do not do, precisely because the labels are not clean enough to support it. When we say non-ARO, we mean the entire remainder of the roster taken as one block, 442 developments and 27,823 units, treated as a single comparison category rather than carved into a precision the free-text field does not allow ^[1].

The two numbers also explain why people talk past each other about the ordinance. A supporter who points to 156 developments is counting honestly, and so is a critic who points to 1,727 units, or under six percent of the listed stock. The development figure captures how widely the policy reaches into the city's pipeline of new construction, touching a quarter of the developments on the affordable roster. The unit figure captures how little of the actual affordable housing supply the policy is responsible for. Neither number is spin. They measure different things, and a fair account of ARO has to hold both, which is the same discipline the rest of this paper tries to keep. The share split is not a gotcha against either side. It is the structural fact that any honest verdict on the ordinance, favorable or not, has to begin from.

The reason this finding leads is that it sets up the three that follow. The development-versus-unit split is why the geography looks the way it does, since a program that produces many small developments will scatter its presence across neighborhoods differently than one that builds a few large projects. It is why the per-development size distribution, which we examine later, skews so heavily toward small set-asides. And it is the quantitative backbone of the gap the paper opened on, because a program that accounts for a quarter of developments but a sliver of units is, almost by construction, a program whose footprint is wide in label and narrow in housing. Where that narrow footprint lands is the next question, and it is the one the roster is best equipped to answer.

Where the Units Landed

If the share split tells us what ARO is, the community-area ranking tells us where it sits, and the answer is a short list of neighborhoods rather than a map of the whole city. Ranked by development count, the roster's ARO leaders are West Town with 24, Near West Side with 20, Lake View with 19, Uptown with 18, Logan Square with 13, and Near North Side with 12, with Lincoln Square at 7 and Portage Park at 5 trailing the pack ^[1]. These eight areas alone account for the bulk of the file's ARO presence, and every one of them sits on the North Side, the Northwest Side, or close to downtown. The geography is not subtle. A reader who knew nothing about the ordinance and saw only this list would correctly guess that the program rides on the parts of Chicago where private apartment construction has been busiest.

Ranking by units rather than by development count reshuffles the order without changing the story. Counted by affordable units, Near West Side moves to the top with 584, followed by Near North Side with 260, then West Town with 179, Uptown with 147, and Logan Square with 107 ^[1]. A few areas that barely register on the development list climb when units are the measure. Lincoln Park, with just 3 ARO developments, carries 80 units, and Hyde Park, with 4, carries 75 ^[1]. The reordering happens because a handful of larger set-asides sit in these areas, and a single building with dozens of affordable units outweighs many buildings with one or two. West Town is the clearest case of the divergence. It leads the city in ARO developments at 24 yet holds only 179 units, which works out to fewer than eight per development, while Near West Side reaches 584 units from 20 developments ^[1]. The same program produces a dense scatter of tiny set-asides in one neighborhood and a smaller number of larger ones in the next.

The concentration is worth stating in plain figures rather than leaving it as an impression. Of Chicago's 77 community areas, only 24 host any ARO building at all ^[1]. That leaves 53 areas, more than two thirds of the city, with no listed ARO development whatsoever. Narrow the lens further and the picture tightens again. The top five community areas by unit count hold 1,277 of the 1,727 ARO units in the file, or 73.9 percent ^[1]. Close to three quarters of the standing ARO units the City lists sit in five of seventy-seven neighborhoods. The remaining quarter is spread across the other nineteen areas that carry any ARO at all, and most of the map carries none.

The common thread among the leaders is not hard to name. West Town, Near West Side, Lake View, Uptown, Logan Square, Near North Side, Lincoln Square, and Portage Park are, by and large, higher-cost areas on the North Side, the Northwest Side, and the near-downtown ring where market-rate apartment construction has been active for years ^[1]. Near West Side and Near North Side wrap the Loop, taking in the West Loop and the Gold Coast and River North, where the towers have gone up fastest. West Town and Logan Square sit on the Milwaukee Avenue corridor that gentrified hard through the 2010s. Lake View and Uptown run up the lakefront on the North Side. This is the pattern one would expect from a mandate that attaches to private building. The ordinance sets aside a fraction of units in qualifying projects, so it can only produce affordable units where qualifying projects go up, and qualifying projects go up where the development math works, which in Chicago has meant the appreciating North and Northwest Sides and the blocks around downtown. The published literature frames the same dependence in general terms, since inclusionary mandates produce output that tracks where private development is occurring and how the program is designed rather than where need is greatest ^[2][3]. Graham-Bailey's ward-level finding is the local version of the same point, with the 27th Ward and its downtown-adjacent construction corridor triggering thousands of units while most of the city triggered few ^[10].

It is worth pausing on how our community-area picture and Graham-Bailey's ward picture fit together, because they are measured on different maps and from different sources, yet they point the same way. Chicago's 50 wards and its 77 community areas are different geographies, drawn for different purposes, and a single ward can straddle several community areas. Graham-Bailey worked from the City's quarterly ARO reports and counted units triggered by ward, and found the 27th Ward alone responsible for 2,690 units ^[10]. We work from a present-tense roster of standing buildings and count by community area, and find the top five areas holding 1,277 of 1,727 listed units ^[1]. The two counts are not the same measurement, and the 2,690 figure is theirs, not ours, drawn from a fuller cumulative source than our snapshot. What is striking is that two independent slices of the data, on two map systems, both land on the West Loop and the river corridors just outside downtown as the program's center of gravity. The 27th Ward covers much of that ground, and so do our Near West Side and Near North Side leaders. When a description holds across the ward map and the community-area map, across a cumulative source and a snapshot, the concentration it describes is unlikely to be an artifact of how the geography was sliced.

A word of caution belongs here before the next section leans on this geography. What the roster shows is where ARO units have been booked and listed, not why they landed there, and the distinction matters for everything that follows. We are reading a snapshot of registered buildings and counting them by neighborhood. We are not estimating what would have happened under a different ordinance, a different fee schedule, or a different building cycle, and we are not claiming that the program steered units toward these areas by design or away from others by intent. The concentration is real in the file. Its causes belong to the development market, the fee-in-lieu option, and the program's structure, all of which sit outside what this dataset can adjudicate. With that boundary stated, the more pointed question is not where ARO is, but where it is not, and the roster answers that one too.

The Map Where ARO Is Absent

The siting story has a second half the leader list cannot show, because a ranking of where ARO concentrates says nothing about whether those same neighborhoods needed it most. To see that, the roster has to be read the other way around, starting from the areas that carry the longest affordable rosters and asking how much of that housing is ARO. Across much of the South and West Sides, the answer is almost none of it.

Consider the neighborhoods the file shows carrying large affordable inventories built under other programs. North Lawndale lists 35 non-ARO affordable developments and zero ARO ^[1]. Grand Boulevard lists 30 non-ARO developments and zero ARO ^[1]. Washington Park and Oakland each carry 18 non-ARO developments and no ARO at all, and Woodlawn carries 17 with none ^[1]. Humboldt Park holds the largest non-ARO roster in the city at 44 developments, against which it lists a single ARO building ^[1]. These are areas with deep affordable-housing footprints on the City's own list, assembled over decades through public housing, project-based subsidy, supportive housing, and senior developments, and the inclusionary program is effectively absent from every one of them.

These are not arbitrary names. They are the heart of Black Chicago and the heart of the city's twentieth-century disinvestment. Grand Boulevard, Washington Park, and Oakland make up much of the historic Black Belt, the narrow band on the South Side into which restrictive covenants and the color line packed Black Chicagoans through the Great Migration. North Lawndale on the West Side was the ground Beryl Satter reconstructed in Family Properties, where contract sellers stripped equity from Black families building by building in the 1950s and 1960s, and where the 1968 unrest after Martin Luther King's assassination accelerated a disinvestment that never fully reversed. Grand Boulevard held the Robert Taylor Homes, the four-mile wall of CHA high-rises along State Street that the Plan for Transformation began tearing down in 1999. The reason these community areas carry long non-ARO rosters is that the public and nonprofit affordable-housing system spent decades concentrating subsidized units exactly here, in the places the private market had abandoned. That history is the backdrop against which the ARO zeros read. The ordinance is missing from the very neighborhoods the rest of the affordable-housing system was built to serve.

There is a hard irony in this distribution that the earlier discussion of stakes sets up. If the case for inclusionary zoning is partly that it places affordable units in opportunity-rich neighborhoods low-income families are otherwise priced out of, then the program works, where it works at all, by adding affordability to the North Side and the near-downtown ring. The mirror image is that it adds almost nothing to the South and West Side neighborhoods that already carry the city's deepest concentrations of subsidized housing and its deepest history of disinvestment. Both halves of that sentence can be virtues or failures depending on which goal one holds. Read as an integration tool, scattering units into appreciating neighborhoods is the point, and the South and West Side absence is beside it, since those areas were never the target of an opportunity-access strategy. Read as a tool for directing investment back to neighborhoods stripped of it, the same pattern is a near-total miss. The roster cannot tell us which goal the ordinance should be measured against. It can only show that on the first measure the units land where the theory wants them, and on the second they are absent from the places that most need reinvestment, which is why reasonable people read the same map and reach opposite conclusions ^[9].

The structural pattern behind the examples is stark when the whole file is tallied. Forty-four of Chicago's community areas have other affordable developments on the roster but zero ARO ^[1]. Twenty-two areas carry both ARO and non-ARO listings, and only two areas carry ARO with no non-ARO presence at all ^[1]. Read together, those three counts describe a near-total separation between where the inclusionary program operates and where the rest of the affordable inventory sits. ARO appears alongside other affordable housing in a minority of neighborhoods, is absent from a large block of areas that carry substantial non-ARO rosters, and is the only affordable listing in just two community areas. The cross-tab against the leader list sharpens the contrast. Uptown carries 18 ARO developments beside 21 non-ARO, and Logan Square 13 beside 19, so in a few North and Northwest Side areas the two coexist ^[1]. On the South and West Sides the columns do not coexist. The non-ARO column is long and the ARO column is empty.

The published research helps explain why this distribution is the one to expect rather than a surprise. The Chicago-specific study of the ordinance finds it delivers some measurable equity benefit for Black residents but works only at small scale, with integration outcomes the authors treat as contested rather than settled ^[9]. Graham-Bailey's digitized analysis of the City's quarterly ARO reporting documents severe concentration at the ward level, with the 27th Ward alone triggering 2,690 ARO units, a figure that signals how unevenly the program's activity has clustered around particular corridors of development ^[10]. Neither finding is ours, and we attribute both to their sources, but both point the same direction the roster does. An inclusionary mandate that produces affordable units as a byproduct of market-rate construction will be thick where that construction is thick and thin where it is thin, and in Chicago the thin areas have historically been the disinvested South and West Sides.

The candor this paper owes is to stop the inference there. The file establishes the distribution. It does not establish the mechanism, and it certainly does not establish intent. A neighborhood with zero ARO and 35 other affordable developments is not, on this evidence, a neighborhood the ordinance avoided. It is a neighborhood where little of the private market-rate construction that triggers ARO has occurred, which is a different statement with different causes. Disinvestment, lending patterns, land values, and the long arc of redlining all shape where apartment towers get built, and all of them sit upstream of the ordinance and outside this dataset. The maps the Home Owners' Loan Corporation drew in the 1930s, which marked the Black Belt and the West Side in red and told lenders not to write mortgages there, set a pattern of withheld credit that no inclusionary ordinance was designed to undo and that still tracks, loosely, with where private towers rise and where they do not. We can say with confidence that ARO is missing from the places the City has historically concentrated its other affordable housing. Why those places see little triggering construction is a question the roster raises and cannot answer, and we leave it raised rather than pretend to close it.

Small Pockets, Not Large Projects

The development-versus-unit gap from the earlier section has a simple explanation, and the per-development size distribution is where it becomes visible. If ARO accounts for a quarter of the listed developments but a sliver of the units, the individual set-asides must be small, and they are. Across the 155 ARO developments that report a unit count, the median is 5 affordable units, the mean is 11.1, and the range runs from a single unit to 75 ^[1]. The gap between the median and the mean is itself informative, since a median of 5 against a mean of 11.1 says the distribution is pulled rightward by a small number of larger projects while most developments sit well below the average. A typical ARO set-aside is not a building. It is a pocket inside one.

The buckets make the skew concrete. Nineteen ARO developments carry exactly one affordable unit, and 61 carry between two and five ^[1]. Thirty-three fall in the six-to-ten range, 22 in the eleven-to-twenty-five range, and 20 sit at twenty-six units or more ^[1]. Translated into shares, 51.6 percent of ARO developments contain five or fewer affordable units, and 12.3 percent contain exactly one ^[1]. More than half of the program's developments are set-asides of five units or fewer, and roughly one in eight is a lone affordable apartment in an otherwise market-rate building. The twenty developments in the top bucket carry the weight that lifts the mean and that drives the unit-count leaders from the earlier section, but they are the exception in a file dominated by small numbers.

It is worth picturing what a one-unit or five-unit set-aside means on the ground. A developer puts up a fifty-unit apartment building on a hot block in West Town, and the ordinance requires that a small fraction of those units rent at an affordable rate, which comes to a few apartments scattered through the building. Those few apartments are a real benefit to the few households that win them, and they are also, by the design of a percentage requirement, a rounding error against the neighborhood's need. The roster shows this happening again and again. The median ARO development puts five affordable apartments into the city's stock, and the most common single outcome, nineteen times over, is one ^[1]. The program produces affordability in ones and fives, building by building, which is a different kind of output than a forty-unit supportive-housing development or a CHA mid-rise contributes in a single stroke.

The smallness is not purely a weakness, and an even-handed reading should say so. Scattering a few affordable apartments through a market-rate building in an appreciating neighborhood is, in fair-housing terms, close to the opposite of the mid-century pattern that stacked thousands of subsidized units into a single high-rise in a single segregated census tract. A household in a one-unit ARO set-aside on the North Side lives in an unmarked apartment in a market-rate building, with the same address and the same lobby as its neighbors, in a neighborhood with the schools and transit and grocery access that the disinvested South and West Sides were denied. That is exactly the access-to-opportunity outcome the integration literature values, and it is the thread the University of Chicago equity study pulls when it credits the ordinance with real benefit for Black Chicagoans even at small scale ^[9]. So the size distribution carries a genuine tension. The same feature that makes ARO a small contributor to the citywide unit count, its production of affordability in ones and fives rather than in large blocks, is the feature that makes the units it does produce deconcentrated and woven into higher-opportunity neighborhoods. Whether that trade is worth making is a value judgment the roster cannot resolve, but the roster does make the shape of the trade visible.

This shape is the one the inclusionary-zoning literature would predict, and naming that keeps the finding descriptive rather than accusatory. A percentage set-aside applied building by building produces affordable counts that scale with the size of each market-rate project, so most outputs will be small because most buildings are not enormous. Published work ties affordable production under these programs to set-aside share and to tenure, with the size of what gets built following from how the requirement is structured rather than from any single project's ambition ^[3][5]. The same research finds that stricter, mandatory, longer-standing designs produce more units, which is the lever that separates a program of mostly single-unit set-asides from one that yields larger pockets ^[5]. Graham-Bailey's Chicago finding that the stricter 2015 rules and the 2017 pilot raised on-site production is that lever moving in the City's own data, and it is the reason the small counts in this snapshot should not be read as the program's permanent ceiling ^[10]. The small per-project counts in the file are the expected arithmetic of a percentage requirement on individual buildings, not a defect we are diagnosing.

One piece of honest housekeeping belongs with these numbers. The size statistics rest on the 155 ARO rows that report a unit count, because one of the 156 ARO developments in the file is missing its units value ^[1]. We report that gap rather than fill it, since imputing a number for a single missing row would dress up the distribution with a value the City did not record. The effect on the summary figures is negligible at this scale, but the principle matters more than the magnitude. Every count in this section is a count of what the roster actually holds, with the one absence noted in the open.

What This File Can and Cannot Tell Us

A descriptive paper earns its conclusions by being precise about the dataset's edges, and this roster has several worth naming plainly before the close. The file is a point-in-time snapshot of buildings that registered affordable units with the City, not a cumulative ledger of everything the ordinance has ever produced or promised ^[1]. That single fact shapes what every number in the paper can and cannot bear. A building enters the file because it currently appears on the Department of Housing's affordable-rental list, which means the roster almost certainly undercounts the program's full history, including in-lieu fees collected, units forgone when developers paid rather than built, and any ARO units that were produced but never listed on this particular resource ^[1].

The missing fields matter as much as the snapshot framing. The CSV carries no construction year, no completion date, no AMI level, no rent figure, and no unit-mix detail ^[1]. Their absence rules out whole categories of question. We cannot trace how ARO output changed over time, because there are no dates to plot against, which means the 2015 and 2021 revisions that the literature tells us should matter are invisible in our data even though they are central to the policy story ^[5][10]. We cannot measure affordability depth, because there are no income-targeting tiers in the file, so a unit renting at 60 percent of area median income and a unit renting at 30 percent look identical here. We cannot say anything about rent burden or who can actually afford these apartments, because rents are not recorded. What the roster supports is narrow, and we have tried to stay inside it, namely where listed ARO units sit and how many there are, counted and cross-tabulated, and nothing past that line.

The gap between our count and the larger totals in circulation deserves a direct reconciliation rather than a footnote. The Illinois Policy Institute's critical report puts ARO at roughly 2,798 units built or promised over twenty-one years, about 140 a year and on the order of 2 percent of new housing stock, alongside more than 200 million dollars in in-lieu fees ^[8]. Our file holds 1,727 listed units ^[1]. The two figures are not in conflict, because they count different things. The 2,798 figure is a built-or-promised cumulative tally drawn from the City's quarterly reporting, reflecting pledges and completions across the program's life, while our 1,727 is a present-tense snapshot of what one DOH roster currently lists ^[1][8]. We attribute the larger number and the fee total to that report and treat both as outside our dataset, since neither the fee accounting nor the built-or-promised history lives in the CSV we analyzed. The Inspector General's audit reached a related conclusion from inside the City, finding the program did not properly account for its ARO and density-bonus fees, which is precisely the kind of figure our file cannot recover ^[6].

It is worth being concrete about what a file capable of answering the harder questions would have to contain, because the gap between this roster and that file is the gap between description and evaluation. To trace output over time, the data would need a completion or permit date on every development, so that units produced under the 2003 rules could be separated from units produced after the 2015 and 2021 tightenings. To assess affordability depth, it would need the area-median-income tier on each unit, since an affordable unit at 30 percent of AMI serves a far poorer household than one at 80 percent. To weigh the fee option honestly, it would need a parallel record of every project that chose to pay rather than build, the dollars each paid, and the units that choice forgave, none of which a roster of standing buildings can hold. That the most useful Chicago analysis of ARO geography to date came from Rashada Graham-Bailey hand-digitizing the Department of Planning and Development's quarterly PDF reports is itself a comment on the state of the public record ^[10]. The richer fields exist somewhere in the City's reporting, but they live in PDF appendices that have to be transcribed by hand rather than in a machine-readable file like the one we analyzed. We name this not to excuse the limits of our analysis but to locate them precisely. The constraint is the published data, and a city serious about letting the public evaluate the ordinance could lift most of it by releasing the quarterly figures as structured data.

Two methodological choices need restating so the cross-tabs are read correctly. The non-ARO comparison group is everything else on the roster, which means public housing, multifamily, senior, supportive, and single-room-occupancy developments folded into one heterogeneous category built under many different programs ^[1]. That group's property-type field also contains free-text spelling variants, including misspellings, that we did not merge, so the non-ARO count is a description of how the roster is filled in rather than a clean program census ^[1]. Comparing ARO against that group is a fair way to show relative siting, but it is not a verdict on ARO's worth against any single alternative program, and we have framed every cross-tab accordingly.

The one analysis a reader might expect and not find here is a map. The dataset's latitude and longitude columns are complete, with zero missing values across all 156 ARO rows, so the coordinates are present and ready ^[1]. We did not overlay them on a HOLC redlining map in this paper because the charting component supports only bar, line, and pie, and a point map is none of those. The coordinates are noted as available for separate mapping work rather than forced into a chart type that cannot carry them ^[1]. That is a limitation of the presentation here, not of the data, and we flag it as such. A map of those 156 points against the 1930s redlining grades would make the siting argument visually rather than in a table, and it is the obvious next step for anyone extending this work.

Set against the published evidence, our descriptive picture lines up without needing to be inflated. The literature holds that inclusionary zoning is no silver bullet and that its effects on price and production depend on the market and on how long the program has run ^[2], and that empirical studies of comparable mandates have found modest unit counts with no measurable lift to overall supply ^[4]. A roster showing a quarter of developments, a sliver of units, heavy small set-asides, and tight geographic concentration is what that body of work would lead one to expect. We present the agreement as agreement, not as confirmation we are positioned to provide, since our file describes one city's standing inventory and the studies test broader questions our data cannot reach.

Reading the Ordinance by What It Has Booked

The gap the paper opened on holds up under the count. Chicago's Affordable Requirements Ordinance has produced real but modest affordable housing, and the standing units the City lists sit disproportionately in higher-cost North Side, Northwest Side, and near-downtown community areas rather than in the disinvested South and West Side neighborhoods that carry the city's longest non-ARO affordable rosters ^[1]. That is a statement about where the units are and how many there are, drawn from 1,727 listed units across 24 of 77 community areas with 73.9 percent of those units in five areas, and it is offered as description, not as a finding about cause ^[1].

The honest reading holds two things at once. A mandate that attaches to private market-rate construction will follow that construction, so part of the concentration is mechanical rather than chosen, an artifact of where buildings go up rather than evidence of where the program tried to put them. At the same time, the literature that calls inclusionary zoning no silver bullet is the same literature that finds stricter, older, mandatory, jurisdiction-wide designs with tighter income targeting produce meaningfully more affordable units ^[2][5]. Both readings sit inside what the evidence supports. The geography is partly structural, the scale of what a program yields is partly a matter of how it is written, and neither observation requires us to assign blame or credit from a roster that was never built to settle the question.

The two most pointed Chicago verdicts on the ordinance sit on opposite sides, and an even-handed reading has to take both seriously rather than splitting the difference. The Illinois Policy Institute reads the cumulative output, roughly 2,798 units over twenty-one years and more than 200 million dollars in fees, as proof that the mandate has delivered little while raising the cost of the construction it touches, and it titles its report accordingly ^[8]. The University of Chicago equity study reads a different ledger, finding that the ordinance produces real benefit for Black Chicagoans, though only at small scale and with integration outcomes still in dispute ^[9]. These are not simply two opinions about the same facts. They weigh different things. The critical report foregrounds aggregate supply and cost, the metrics on which a market-skeptical analyst judges a housing policy. The equity study foregrounds who gains access to opportunity-rich neighborhoods, the metric on which a fair-housing analyst judges the same policy. Our roster cannot referee between them, because it measures neither cost nor opportunity, only where units sit and how many. What it can say is that both readings are compatible with the geography we find. A program can be small in aggregate, as the critical report stresses, and still place a meaningful number of Black households into appreciating North Side neighborhoods they were historically shut out of, as the equity study stresses, because those two claims describe different cross-sections of the same modest output. Holding them together, rather than choosing the one that flatters a prior, is the honest posture the evidence allows.

There is a deeper point under the geography that the file gestures at without proving. An inclusionary mandate is a tool for capturing a slice of a hot market for public benefit. Where the market is hot, it works as designed and the slices add up, slowly, in West Town and the West Loop and along the lakefront. Where the market is cold, on the blocks that redlining and contract selling and decades of withheld credit left without the private construction that triggers the requirement, the tool has almost nothing to grip. So the ordinance does its quiet redistribution inside the neighborhoods that least resemble the ones the rest of the city's affordable-housing system was built to serve. That is not a flaw the City could fix by tightening the percentage. It is a property of the kind of instrument the ARO is, and recognizing it is part of reading the ordinance honestly. The fix for cold-market neighborhoods was never going to be a set-aside on construction that is not happening.

This is also why the design lessons in the literature are not abstract for Chicago. Wang and Fu found that mandatory, older, jurisdiction-wide programs with stricter income targeting produce more affordable units, and Graham-Bailey found that Chicago's own stricter 2015 rules and 2017 pilot lifted on-site production ^[5][10]. Put beside the silver-bullet caution that output is market-dependent, the combined message is that a city cannot make a cold-market neighborhood produce inclusionary units by writing a tougher ordinance, but it can raise how much the hot-market neighborhoods yield by tightening the requirement that applies where building is already happening. The roster we analyze is a present-tense mix of units produced under looser and stricter versions of the rule, with no dates to separate them, so it cannot show that lift directly. But the literature tells us the lever exists and that Chicago has already pulled it once, which means the small per-project counts in our snapshot describe the program as it has been, not a fixed ceiling on what a differently designed version could produce ^[5][10].

What the file cannot close points naturally to the record that could. A fuller public accounting would let Chicago answer the questions this dataset leaves open, namely how much has been collected in in-lieu fees, how many units were forgone when developers paid instead of built, how deep the affordability runs against area median income, and how the program's output has shifted across its successive revisions. Those questions live in the City's fee accounts, in the quarterly reporting, and in the audit record, not in a snapshot roster of registered buildings ^[6][8]. We point to them without prescribing an answer and without inventing a number to fill the space, because doing either would trade the discipline of the paper for a conclusion the data does not carry.

It is worth saying plainly what this kind of descriptive work is good for, since its modesty is easy to mistake for thinness. A careful count of a public file does not settle whether a policy succeeded, but it fixes the factual floor that any argument about the policy has to stand on. Before anyone debates whether ARO is too small, too concentrated, or well-targeted, there has to be agreement on where its units are and how many the City lists, and that agreement is exactly what a contested policy usually lacks. The Illinois Policy Institute and the University of Chicago researchers disagree about what the ordinance means, but a shared, checkable description of the standing inventory narrows the ground their disagreement can occupy ^[8][9]. We have tried to supply a piece of that floor, drawn from the City's own data and reported with its limits in the open, so that the next argument about the ordinance can be about values and counterfactuals rather than about what the file says. That is the proper ambition of a descriptive paper, and we have tried not to exceed it.

That discipline is the note to end on. We can say with confidence where the standing ARO units are and how many the City lists, because we counted them in a real public file and reported exactly what it holds. What those units are worth against what might have been built under a different design, a different fee, or a different building cycle is a question of counterfactuals and dollars this roster does not contain. The ordinance promised affordability woven into the building boom citywide. The file shows affordability woven into the building boom where the boom happened, which is not the whole city. Reading the ordinance by what it has booked, that is the finding, stated as plainly as the data allows and no further.