← Back to the Chicago LVT reports

Chicago, IL — LVT Model Methodology & Limitations

Model: 2.5:1 split-rate, revenue-neutral (canonical). Also computed: 100% building abatement (pure LVT), and a classification-preserved variant. Explainer generated: 2026-06-17 · Probe depth: deep (cached parcel data present) One-line read: The headline result — Chicago housing pays substantially more and commercial pays less — is driven as much by removing Cook County's classification system as by the land shift itself, because the reform base is market value. Read every category result through that lens.

---

1. What was modeled

• Jurisdiction / body: City of Chicago — the City Corporate levy only (the City's own general + pension extension). Not the composite tax bill.
• Reform: revenue-neutral shift toward land. Canonical export is a 2.5:1 split-rate (land taxed at 2.5× the improvement rate); the ratio is capped at 2.5:1 to respect the Illinois Constitution's classification limit (Art. IX §4(b)). A 100% building abatement (pure LVT) and a classification-preserved variant are also produced for comparison.
• Revenue-neutrality basis: the modeled new revenue is solved to equal the modeled current revenue, which is pinned to the official City Corporate levy.

2. Data provenance

• Committed/reproducible source: Cook County open-data Socrata — Assessor Assessed Values (uzyt-m557) for land/building/total assessed values, joined to Parcel Universe (nj4t-kc8j) for location. City = the eight Chicago assessment townships (codes 70–77), which are coterminous with the municipal boundary.
• Local cache seeding (this run): because the public Socrata API was severely rate-limited, the cache (cities/chicago/data/parcels.gpq) was seeded from the Chicago Cityscape Postgres — assessed values from assessor_assessed_values2, and parcel-centroid lat/lon from parcel_2025_unioned (ST_Centroid, SRID 3435→4326), with a fallback to propertytaxes_combined.geom_2025. The data is identical to the public datasets; the committed notebook still references Socrata.
• Tax year: 2024 (the data) vs. the official levy figure (also TY2024, Cook County Clerk 2024 Tax Rate Report) — no tax-year drift.
• Parcel count: 882,904 total (matches Socrata 2024 for townships 70–77). 830,169 modeled after holding out 52,735 fully-exempt parcels.
• Geometry match: 100.0% of parcels assigned a centroid (67.1% on pin14, → 99.7% via pin10 parent for condo units, → 100% via the propertytaxes_combined fallback). Note the geometry vintage is the 2025 parcel file applied to 2024 assessments; ~2,549 taxable PINs (0.3%) that churned between years were dropped earlier in the pure-Socrata build but recovered here.
• Census join: 100.0% of parcels matched to a 2022 ACS block group (std_geoid non-null = 100%); median_income non-null = 90.5% (some block groups lack an income estimate).

3. Tax base & millage

• Assessed → market value. Cook County publishes assessed values, not market values, and assesses by class: residential/vacant/multi-family (major classes 1/2/3/9) at 10%, not-for-profit (4) at 20%, commercial/industrial (5) at 25%, incentive classes (6/7/8) at a reduced 10% in their active phase. Each parcel's market value = assessed ÷ its class level of assessment. The split-rate / abatement are applied to market value, so a dollar of land is taxed uniformly across classes. (This is the single most consequential modeling choice — see §9.)
• Equalized Assessed Value (EAV) = assessed × the IDOR Cook County state equalizer 3.0355 (TY2024 final). Current tax is computed on gross EAV.
• Current millage / rate. The City Corporate levy is spread across taxable EAV at a uniform citywide rate. Rather than look up the rate, the model derives it from the levy: city_rate = CITY_LEVY / Σ(taxable gross EAV) = 1.1733%. Published City Corporate rate (Cook County Clerk) = 1.4958%. The derived rate is lower because gross EAV exceeds the net taxable base by exemptions + TIF (≈ −22% / ratio 0.78) — see §6/§9.
• Modeled reform millages (2.5:1 split-rate): land 8.358 / improvement 3.343 per $1,000 of market value (ratio exactly 2.5:1). Abatement variant: land 19.28, improvement 0.

4. Revenue

	Amount	Source
Modeled current revenue	$1,642,587,611	notebook §4 output; recomputed Σ current_tax from export = $1,642,587,611
Revenue-neutral new revenue	$1,642,587,611	notebook §5 output; Σ new_tax from export = $1,642,587,611
Official figure	$1,642,587,611	City of Chicago Corporate extension, Cook County Clerk 2024 Tax Rate Report
Gap	0.000%	exact by construction — the levy is the revenue-neutral target

Because current revenue is set to the official levy, Gate-1 "revenue match" is satisfied by construction; the meaningful check is the derived rate (1.17%) vs published (1.50%) — the 0.78 ratio quantifies the gross-EAV gap (§9, limitation 3).

5. Levies modeled vs NOT modeled

• Modeled: City of Chicago Corporate levy only.
• Excluded (all also appear on a Chicago parcel's actual bill): Chicago Board of Education / CPS (~$3.99B, the largest single levy), City of Chicago Library and School Building & Improvement funds, Cook County, Forest Preserve, Metropolitan Water Reclamation District, City Colleges (Community College Dist. 508), Chicago Park District, and any Special Service Areas / TIF. A resident's composite TY2024 rate was ~6.62%; the modeled City Corporate piece is ~1.50% of that. The model therefore describes the incidence of the City's own levy only, not a household's total bill. (No legality brief present to cross-check the vehicle.)

6. Exemptions

• Full exemptions: a parcel is flagged fully exempt if its class is EX or RR (railroad, state-assessed) or it has zero/blank assessed value. 52,735 parcels (6.0%) were flagged and held out of the revenue-neutral solver and excluded from the export and charts (the export's is_fully_exempt sums to 0 because they were removed upstream).
• Partial / dollar relief: NONE modeled. The public Assessed Values dataset carries no per-parcel homeowner / senior / senior-freeze exemptions, so the model runs on gross EAV. This is a deliberate, documented choice (the user opted for gross-EAV over an owner-occupancy proxy). Consequence: the derived city rate (1.17%) sits below the published 1.50%, and absolute burdens are not exemption-exact.
• Caps / circuit breakers: none exist for this levy / none modeled.
• Credits: none modeled.
• How they stack: with no partial relief, no caps, and no credits, the per-parcel computation reduces to: full-exempt zeroing → millage. calculate_current_tax is called with exemption_flag_col='full_exmp' (no exemption_col), so exempt parcels are zeroed; everyone else is taxable_value × millage / 1000. The split-rate solver receives only the non-exempt set.

7. Property categories

• Source field: Cook County 3-character class code (class), mapped via a function (not a flat table) that splits residential subcodes, condos, townhomes, multifamily, mixed-use, and the commercial/industrial/incentive classes per the CCAO Classifications of Real Property (rev. 12/16/2024).
• Transportation - Parking category: classes ending in -22 (5-22/7-22/8-22 automotive) + 7-52, ending in -23 (gasoline stations), and ending in -90 (minor improvements incl. 1-90 and surface lots) — and these are held out of the $0-improvement→Vacant override so surface lots aren't reabsorbed into Vacant Land.
• $0-improvement → Vacant Land override: applied (except Parking). Any parcel with zero building value is recategorized Vacant Land regardless of class.
• 2-41 (vacant land under common ownership w/ adjacent residence) → Vacant Land, taxed at the 10% residential level.
• Distribution (modeled, non-exempt = 830,169):

Category	Parcels		Category	Parcels
Condominium	287,166		Industrial	11,404
Single Family Residential	278,520		Large Multi-Family (5+)	11,258
Small Multi-Family (2-4)	120,202		Other Commercial	4,359
Vacant Land	39,925		Other Residential	4,309
Townhome / Rowhouse	23,020		Office / Commercial Condo	4,245
Mixed Use	15,962		Hotel	436
Transportation - Parking	15,957		Other	172
Retail / General Commercial	13,232		Agricultural	2

• Residual buckets are small (each <0.6%): Other Residential (4,309) is 94% Class 2-01 residential garages (mostly land — explains its near-vacant behavior), Other Commercial (4,359), Other (172). None exceed the ~10% threshold.

8. Land/improvement split + deep-probe diagnostics

Each parcel's land and improvement market values come directly from the assessor's separately-set land and building assessed values (÷ class LOA) — not an allocation of a combined total. The deep probes confirm this produces real variation:

Land-ratio uniformity (land ÷ (land+building), parcels with building>0). A category with >50% of parcels at one modal ratio would be a flat-allocation artifact (the Maricopa case). None are:

Category	n	median land ratio	std	% at modal (2dp)
Single Family Residential	278,520	0.252	0.123	3.5%
Condominium	287,166	0.109	0.133	7.3%
Small Multi-Family (2-4)	120,202	0.216	0.127	4.0%
Industrial	11,404	0.723	0.302	9.0%
Office / Commercial Condo	4,245	0.148	0.164	13.6%
Retail / General Commercial	13,232	0.380	0.243	2.2%
Transportation - Parking	15,899	0.944	0.243	12.0%
Other Residential (garages)	4,309	0.919	0.195	8.9%

Land ratios are economically sensible (condos & offices building-heavy; industrial, parking, garages land-heavy) and vary widely — the split-rate genuinely differentiates parcels.

Collapsed % change (share of a category's parcels within 1 percentage-point of the category median, split-rate):

• Vacant Land: 99.8% — collapsed, but inherently, not as an artifact: a zero-building parcel's % change is land_millage / (0.10 × equalizer × city_rate) − 1, in which land value cancels, so every 10%-assessed vacant parcel moves an identical +134.7% citywide. This is correct land economics, not a bookkeeping flat-rate.
• Every other category: 1.7%–16.8% within 1pp of median — i.e., genuine spread. No collapsed signal in the built categories.

9. Limitations (ranked)

1. Market-value base bundles two reforms. Because the split/abatement run on market value, they simultaneously (a) shift tax to land and (b) undo Cook County's classification system (residential assessed at 10% vs commercial 25%). The large residential increases (SFR +29%) and commercial cuts (offices −54%, hotels −55%) are driven mostly by (b). The classification-preserved variant isolates (a): there, housing is roughly flat-to-down and land-heavy commercial rises. Undermines: any claim that "the land tax raises homeowner bills" — most of that is the assessment-uniformity change, not the land shift.
2. City Corporate levy only. Excludes CPS, County, MWRD, parks, library, community college, TIF, SSAs. Undermines: any statement about a household's total tax bill — this is ~1.5 percentage points of a ~6.6% composite.
3. Gross EAV — no exemptions applied. Homeowner/senior exemptions and TIF increment are not subtracted (public data lacks them), so the derived rate (1.17%) is ~22% below the published 1.50% and per-parcel levels are not exemption-exact. Undermines: precise dollar burdens, especially for owner-occupied homes that currently hold exemptions. (Revenue-neutral distribution is robust.)
4. TY2024 is the City reassessment year. Board of Review values are still rolling out (~90% Board-certified); the model uses Board → certified → mailed coalesce, so a minority of parcels reflect a pre-final stage. Undermines: precision for not-yet-finalized parcels; will settle as certification completes.
5. Incentive classes (6/7/8) assumed at 10%. A minority in years 11–12 (15%/20%) or expired would have market value slightly overstated. Small share. Undermines: exact treatment of incentive parcels only.
6. Geometry vintage mismatch. 2025 parcel centroids on 2024 assessments; resolved to 100% via fallback, but a handful of churned PINs use a parent-parcel centroid. Undermines: nothing material for block-group-level equity.
7. Median_income coverage 90.5%. Equity-by-income charts exclude ~9.5% of parcels in block groups without an ACS income estimate. Undermines: completeness (not direction) of the income-quintile analysis.

10. What the model CAN and CANNOT support

• CAN: the distributional direction and relative magnitude of shifting the City's own levy toward land — which property types and which neighborhoods pay more vs. less, revenue-neutral; the finding that vacant land and (in high-value areas) land-heavy commercial bear large increases while building-heavy commercial cores (Loop, Near North/West) receive large cuts; the geographic North-vs-South contrast; and — via the classification-preserved variant — separation of the land-shift effect from the classification-removal effect.
• CANNOT: a household's actual total tax bill (one levy of many); exemption-exact dollar burdens (gross EAV); the effect on a specific parcel without its real exemptions and final Board value; or any claim that the headline housing increase is "the land tax" without crediting the classification-removal effect baked into the market-value base.