How the estimate is built

The story deck covers the main steps. This deck holds the whole method: UIC's findings in detail, every allocation step with its check, the evidence inside immigrant neighborhoods, the open-data tests and the places where our method is weakest. The last slides list every setting and every source.

UIC marks a number from the research; added marks a choice we made to apply it to your ward. Data powered by Ward Wise.

What comes from UIC and what we add

UIC's Great Cities Institute estimated what 2025 immigration enforcement cost Cook County's stores and restaurants: $1.26B a year. We don't produce a new causal estimate. We split UIC's number across block groups, wards and community areas by stated rules, and we scale it to the Midway Blitz weeks. You can change every rule on the ward map.

The table pairs each piece of UIC's method, which we rebuilt from public data, with the allocation choice we layered on top. Every slide and every control on the map carries the same two badges.

UIC's findings in detail

UICTrips across community lines

Before January 2025, residents of high Latin America-born neighborhoods did most of their shopping and dining in similar neighborhoods. Still, nearly four in ten of their store and restaurant visits went to businesses in low-share parts of the county. UIC calls these cross-community trips, and they are the trips that fell.

UIC measured them with weekly foot-traffic data from Advan (via Dewey), which records where each visitor lives and where each business is, from January 2024 through December 2025.

UICHow UIC measured the drop

UIC compared each high block group's weekly trips to low-share businesses before and after January 20, 2025, with a difference-in-differences model whose treatment is the block group's Latin America-born share. A block group at a higher share gets a larger dose, so UIC reads the effect at any share from one coefficient.

At the average high share of 19 percent, trips fell 9 percent for retail and 10 percent for restaurants countywide. Chicago residents cut retail trips by 6 percent and restaurant trips by 10; suburban residents cut them by 13 and 8.

The model uses 14 weeks before and 19 weeks after January 20. V counts weekly visits from block group i to businesses in low-share block groups; a and l are block-group and week effects.

UICThe trips did not move elsewhere

The same residents kept visiting businesses in other high Latin America-born neighborhoods at their usual rate, and their trips outside the county did not rise. Businesses in low-share areas did not pick up customers from anyone else either. UIC therefore counts the missing trips as lost sales rather than sales that moved.

This finding is why we send losses outside immigrant neighborhoods only to businesses in low-share areas.

UICFrom visits to dollars, and the later weeks

UIC priced each lost visit at Cook County's 2022 Economic Census receipts, in 2025 dollars, divided by the visits the same panel recorded in 2024: $58 for retail and $33 for restaurants. About 18 million retail visits and 7 million restaurant visits a year went missing, which comes to $1.04B plus $0.22B (95 percent confidence interval $0.83B to $1.67B). At an 8.5 percent blended sales-tax rate, governments lost $107M.

The drop deepened over the year. Trips ran 8.9 percent lower for retail and 9.9 percent lower for restaurants over the first 19 weeks, and 12 and 16 percent lower over all 46 weeks UIC observed. The later weeks include Midway Blitz, which began in week 33. UIC declines to call the September deepening a separate effect because fall shopping has its own seasonal pattern.

Every allocation step

UICStep zero: which block groups count as high

We take each tract's Latin America-born share from the 2019–2023 American Community Survey (B05006 over B01003) and give it to every block group in the tract, as UIC did, because the Census Bureau doesn't publish birthplace below the tract. The top 40 percent of Cook County's 4,002 block groups count as high. A block group's centroid decides whether it is in Chicago.

The share counts heads, and some heads live in group quarters. The Cook County Jail tract at 26th and California has 8,470 residents, 8,048 of them in group quarters and 190 households, so its share is 5.9 percent and the jail block group counts as low. We keep UIC's rule and handle the jail on the destination side. A map setting recomputes shares on household population instead, which moves the jail tract to 100 percent and departs from UIC's counts.

addedStep one: spread the baseline, apply the dose

UIC published only the average number of weekly trips per high block group in each region. We split each region's total across its high block groups by population, then applied UIC's dose-response at each block group's own share, so a 40 percent block group loses a larger slice of its trips than a 10 percent one does. Multiplying by UIC's dollars per visit gives each block group's withdrawn spending.

Each region's baseline volume equals UIC's panel volume. The map can spread it by households or evenly instead of by population.

addedWe scale to UIC's total

Summed over the county without adjustment, our version comes to $1.16B for retail and $0.24B for restaurants, $1.40B in all, on 20.0 million and 7.4 million lost visits. UIC reports $1.04B and $0.22B on 18 million and 7 million. The gap comes from rounding in UIC's published inputs and from UIC annualizing its 46-week series.

We scale each sector so the county matches UIC's headline, a factor of 0.90. Our rules then decide only whose dollars they are. The map's presets swap in either end of UIC's confidence interval, keep our unscaled $1.40B or take any total you type.

addedStep two: where the trips would have gone

UIC's panel recorded where each withdrawn trip would have ended, but the published results say only that the destinations sit in low-share block groups, and that suburban residents mostly shopped on nearby suburban corridors. We filled that gap with a gravity model. Each neighborhood's withdrawn dollars go to low Latin America-born block groups in proportion to how many stores and restaurants they have, discounted with distance (exponential decay over 3 miles for Chicago residents and 5 miles for suburban residents).

We measure distance in a straight line from where the origin's 2020 residents live to where the destination's businesses cluster. The average modeled trip runs 6.5 miles for retail and 6.9 for restaurants, longer than λ because suburban trips and big commercial clusters pull it out. Dollars balance: destination losses add up to origin losses in each sector.

This is our softest step. Changing the distance setting moves losses between downtown, the North Side corridors and the suburbs, and the county total stays the same. A ward that is almost entirely high Latin America-born, such as the 14th or 22nd, shows large withdrawn spending and no lost sales here, because its residents' withdrawn trips were trips elsewhere.

addedNo lost sales at the jail

A block group whose residents mostly live in a jail, a dormitory or a nursing home has a head count that doesn't shop, and that head count dilutes its tract's share. The Cook County Jail block group had 6,030 people in the 2020 Census, 4,948 of them in group quarters and 4,678 in custody. Its 5.9 percent share falls under the 6.8 percent cut, so by the letter of UIC's rule it is the only low block group in South Lawndale.

Our first run sent it every dollar the gravity model assigned to South Lawndale: $2.2M in the Blitz weeks, $10.2M a year, which then showed up in Wards 25 and 12 as lost sales at a jail. Dormitory block groups at the University of Chicago and IIT had the same problem on a smaller scale.

The 2020 Census (PL 94-171, table P5) puts twelve Cook County block groups above half group quarters, ten of them low. We give those ten no destination weight. Their businesses still count toward their ward's business total, and the dollars go to the next nearest low-share businesses instead: South Lawndale's to the Lower West Side, the Near West Side and the western suburbs.

The map can switch this off, or reclassify on household population, which makes the jail block group high.

addedCounting businesses

Our business counts come from OpenStreetMap through OSM US Layercake, the only free source that covers the whole county: shops (minus service shops) plus fuel stations and pharmacies for retail, and restaurants, fast food, cafes, bars, pubs and ice cream for restaurants, 19,254 places in all. Compared with Ward Wise's Google Maps sweep of Chicago (2026Q3, 11,304 open eating and drinking places), OpenStreetMap lists a median 44% of them: about 70 percent in Lincoln Park and Logan Square, and 10 to 20 percent in South Lawndale, Austin and Belmont Cragin.

Raw counts would push lost sales toward the North Side. In Chicago we use Google's restaurant counts directly and divide retail counts by each community area's measured coverage, floored at 10 percent. No sweep covers the suburbs, so there we divide both by the city median.

We keep the Google rows private and publish only counts per block group and ratios per community area. The map can switch to raw OpenStreetMap counts or ignore business density.

addedStep three: the Blitz weeks

Operation Midway Blitz ran 9.4 weeks, from September 8, 2025 until most Border Patrol agents left on November 13. UIC didn't estimate the Blitz on its own. Its averages imply that weeks 20 to 46 ran about 14.2 percent lower for retail and 20.3 percent lower for restaurants, deeper than the year as a whole. We price the Blitz weeks at that later rate, 1.18 times the year average for retail and 1.27 for restaurants.

The map offers two other readings. Prorating the year (1.00) gives the Blitz its share of a year under the enforcement climate. The incremental reading (0.44 and 0.65) counts only the deepening after September 8, which answers what the Blitz added on top of the climate, with UIC's caveat that part of that deepening may be seasonal.

ICE's arrest records back up the timing. The Deportation Data Project's files show about 3,470 ICE arrests in Illinois in those weeks, 6.6 times the weekly pace earlier in 2025, and ICE recorded 42% of them in Chicago, 7% in Broadview and 3% in Cicero. Border Patrol led the Blitz on the street and its arrests are missing from that release, so the real increase was larger.

DHS said operations continued after November 13, and the Border Patrol commander returned in mid-December. The map lets you move both dates.

UICaddedSales tax

For all governments together we use UIC's 8.5 percent effective rate: the 9 to 10.25 percent general rate, with grocery and drug sales at 1 percent, weighted by the 2022 Economic Census mix of sales.

For the City of Chicago's own share on sales inside the city, we add its 1.25 percent home-rule tax to the 1 percent municipal share of the state rate, 2.25 percent on retail, plus the 0.5 percent Chicago restaurant tax, 2.75 percent on restaurants. Groceries and drugs return only 1 percent to the city, and we don't reduce the retail rate for them, so the city figure runs slightly high.

addedStep four: wards and neighborhoods

Block groups cross ward lines. We split each block group's withdrawn spending by where its 2020 census-block residents live, and its lost sales by where its businesses sit. A ward's number is the sum, and the estimator shows that arithmetic for any address.

share_pop is the fraction of block group i's residents who live in area W; share_biz the fraction of d's businesses in W; k the 0.90 anchor; w the gravity weights. A block group with business weight but no mapped point of a sector gets one point at its business center.

addedPer-business figures

We divide an area's lost sales by its eligible businesses: those in low-share block groups that aren't group quarters, the only businesses our rule lets lose sales outside immigrant neighborhoods. Dividing by every business misleads where the two counts differ. In Ward 12, 99 percent of businesses sit in high block groups, and an all-business average produced a meaningless "$381 per business" from the jail sliver.

Where an area has businesses but none eligible (Wards 12, 13, 14, 22 and 31; South Lawndale, Belmont Cragin, Gage Park and fourteen other community areas), the map shows a dash with the note that residents' withdrawn trips went elsewhere. UIC found those businesses kept their own neighbors' trips. The inside-neighborhood line gives them their own figure. Per-resident figures divide by every resident.

Which settings move which numbers

Every area figure comes from the same few formulas, so you can read off what a setting can change. The control total and the window scale everything. Under UIC's headline, UIC's dollars per visit cancel out, because UIC's own $58 and $33 already sit inside its $1.04B and $0.22B. The distance decay and the business counts never touch withdrawn spending.

The same ward, four ways

The county total stays fixed. Tighter distance decay keeps losses near immigrant neighborhoods, and looser decay sends them downtown. Raw OpenStreetMap counts, which miss many South and West Side businesses, push them north.

addedThe files we publish

Anyone can rebuild our ward and community-area numbers, or aggregate them to another geography, from two point layers in the byop/v1 format that Ward Wise reads. Each layer is flagged as a synthetic estimate.

Inside immigrant neighborhoods

UICaddedThe businesses on 26th Street

UIC measured businesses inside high Latin America-born neighborhoods and found them holding roughly steady: trips from their own neighbors changed +0.1 percent for retail and +0.5 percent for restaurants, neither significant, and foot traffic there fell about 1% from January to May (p ≈ 0.08). UIC's headline therefore holds nothing for these businesses. Four owners on 26th Street told reporters their sales fell 40 to 70% during the Blitz. We found no chamber or survey figure.

We report this as its own line and never add it to UIC's total. Two measurements bound it: Illinois sales-tax receipts put Cicero (39% Latin America-born) 9% below its usual fall, and the four owners report a median loss of 62.5%. By default we use the midpoint of the two, scaled to each block group's share and applied to every business in a high block group. The map and the estimator let you switch to either end.

Our own mobility checks split. CTA entries at Cicero fell 19% and Pink Line stations 6 to 10% in the Blitz weeks; Divvy trips between high-share areas fell as much as trips out of them; weekend bus rides on corridors through high-share areas ran about 7% below their earlier 2025 level. Rideshare trips showed no decline into or within these neighborhoods.

Why the panel and the owners can both hold

1. The panel misses people. Advan's panel comes from opt-in location tracking in smartphone apps. Li, Ning, Jing and Lessani (2024, PLOS ONE) audited SafeGraph, the panel Advan's grew out of, against the Census for 2018 to 2022 and found Hispanic residents under-represented in most states and most years after 2020. Coston and colleagues (FAccT 2021) found non-white and older voters less captured. Neither audit measures block groups, so nobody knows how thin the panel runs on 26th Street. People afraid of enforcement also had reasons to turn location off or stop carrying a phone after January, and UIC's tests checked business coverage, not who carried the phones.
2. A visit differs from a sale. One phone in a store counts once whether its owner buys for one household or three, and a volunteer, reporter or rapid responder counts like a customer. Dense strips of small storefronts also confuse GPS.
3. UIC's 1% averages about 1,600 high block groups, most of them 7 to 20% Latin America-born. South Lawndale's block groups run 40 to 50%, and even UIC's dose-response at 47% reads only about 2.5%.
4. UIC measured the 1% from January to May. Its September diagnostics show visits confounded by the season, so its data don't speak to the Blitz weeks.
5. Owners who talk to reporters have losses to report, and they compare against their own memory of a normal fall, without a denominator.

UIC's test of whether residents cut all travel, local trips included, fails its pre-trend check in 14 of 15 versions, so the data leave that question open.

addedThe four owner reports

All four owners run businesses on or just off 26th Street in Little Village, in block groups 27 to 56% Latin America-born. Summaries that repeat "20 to 70 percent" are quoting the spread of these individual reports.

• David Galecio, Atardecer Acapulqueño, 2900 block of W Cermak Rd: revenue down 60 to 70 percent. The story adds that some unnamed eateries report up to 70 percent. WGN, Oct 17, 2025
• Jose Luis Lopez, co-owner of Los Candiles, 26th St and Central Park Ave: dine-in sales down about 60 percent when the fall raids began. WBEZ, Nov 21, 2025
• Mike Moreno Jr., Moreno's Liquors, 3724 W 26th St: sales down 20 percent at first, as much as 60 percent at the worst, then 40 percent for two months. WBEZ, Feb 20, 2026
• Michelle Macias, Carniceria y Taqueria Aguascalientes, 3132 W 26th St: business down 60 to 70 percent over the year since the Blitz. ABC7 Chicago, Sept 7, 2026

Two related reports carry no sales figure: the Illinois Restaurant Association said upwards of 20 percent of staff stopped coming to work (WGN, above), and Casa Madera Furniture in Brighton Park planned to import half its usual merchandise in 2026 (Block Club, Feb 13, 2026). The estimator lists every owner report within 1.5 miles of an address, always as an individual report.

addedPricing a block group's sales from its jobs

To turn a loss rate into dollars we need each block group's annual sales. We count the retail and restaurant jobs located there in the Census Bureau's LEHD LODES 2023 workplace files and price each job at Cook County's Economic Census sales per job, which UIC also used. Both sectors add up to UIC's county receipts.

Until October 27, 2026 we priced every storefront at the county's average sales per business, about $4.65M for retail and $0.98M for restaurants. That counted a corridor of small shops at an average that includes big-box stores. On jobs, Chicago's high block groups hold $18.4B a year of sales, against $29.6B on business counts.

Suburban high block groups gain on jobs, because they hold malls, auto rows and hotels in Rosemont, Melrose Park and Northlake with many jobs and few mapped storefronts. The restaurant series includes hotel jobs, and our 0.8 share counts them as restaurant jobs there. Business counts still divide the result into per-business figures for the estimator.

addedWhat actual sales show

Illinois publishes each municipality's sales-tax receipts, a direct record of taxable sales. During the Blitz months, sales in the high-share suburbs (about 19% Latin America-born) ran 3.4% below their earlier 2025 level (95% interval −6.1 to −0.5), and Cicero (39%) ran 8 to 10% below its own seasonal pattern. The solid curve extends the suburban figure to other shares: 6.9% at 39% and 8.2% at a 26th Street block group's 47%. We chose the curve's shape, and two points can't pin it down.

The tax figures cover every storefront in a municipality. The orange dots are four storefronts on the hardest-hit corridor whose owners chose to talk to reporters. UIC's 1% comes from a phone panel that probably misses the most exposed households. The truth on 26th Street likely falls between the curve and the dots, and no public record says where. The shaded band runs from Cicero's sales up to the owners' median, both scaled by share, and our default sits at its middle.

addedWhere the two anchors come from

Low: Cicero's September-to-November taxable sales ran 4.3 percent above its February-to-August level in 2023 and 7.2 percent above in 2024. In 2025 they ran 3.4 percent below. Against the average of the two earlier years, that is a 9.15 percent shortfall at Cicero's 39 percent share.

High: the four owners' midpoints are 65, 60, 50 and 65 percent, with a median of 62.5 percent. Their block groups run 28, 56, 27 and 47 percent Latin America-born, with a median of 37.6 percent.

Each anchor becomes a curve through its own point, and the default averages the two rates. At a 47 percent share the low, middle and high readings take 11, 41 and 71 percent of the Blitz weeks' sales; at 15 percent they take 3, 18 and 32.

These readings were measured during the Blitz months, so the window applies only their length (weeks/52) and skips the depth factor. UIC's 1% comes from January to May and keeps the depth. Before October 27, 2026 we applied the depth to every reading, which overstated the sales-tax readings by 18 to 27 percent.

Every reading, in dollars

The table prices each reading on the same jobs base for the Blitz weeks. Curves that rise with the share move dollars toward Chicago, whose high block groups run higher shares than the suburban ones. City sales tax and the all-business figures include the line; the UIC-anchored figures leave it out.

The share-scaled reading rises from zero at the 6.8% cut to 30% at a 40% share, a press-informed upper bound. An enforcement-based reading waits on a sightings rate that the public maps can't supply.

Ward 12, worked through

Ward 12 covers Brighton Park and McKinley Park and is almost entirely high Latin America-born, so it carries no UIC-anchored losses at all. Its businesses sit in 38 high block groups with 1,621 retail jobs and 704 restaurant jobs. At county sales per job that comes to $817M a year and $148M in the 9.4 Blitz weeks, at a sales-weighted share of 23.7 percent.

Applied block group by block group, Cicero's sales give $8.3M, the midpoint $37.3M and the owner reports $66.2M, an effective 45 percent of the ward's Blitz-week sales. The estimator's "Show the math" repeats this for any address.

addedWhat the free evidence says

Illinois sales-tax receipts are the only free record of actual sales, and they put high-share suburbs a few percent short, nowhere near the owner reports. Chicago business licenses show fewer openings and no rush of closures. The other two sources can't carry a rate.

Your choice of source moves the estimate 35-fold

Outside immigrant neighborhoods we anchor to UIC, and the figure barely changes. Inside them, the figure depends on whose evidence you accept. For Chicago businesses in the Blitz weeks, UIC's phone panel puts the loss at … and the owner reports put it at …, with Cicero's tax receipts (…) in between. We show the midpoint, …, to start the conversation.

Settling it takes the business-by-business sales-tax filings that the Illinois Department of Revenue holds. A City inquiry could request them.

The chart adds the suburban sales-tax average for comparison.

Chicago's own sales tax caps the loss near $650M

Each total includes the … that we carry from UIC's estimate outside immigrant neighborhoods. UIC found the withdrawn trips did not reappear elsewhere, so every lost dollar should show in the City's receipts. October 2024 carries a one-off statewide catch-up in receipts, so the 2024 pattern gives a generous top, and we set the whole fall's shortfall against 9.4 weeks. From January 2025 Illinois counts Chicago residents' online orders as Chicago sales, and those orders, which peak in November, can mask part of a drop at storefronts.

The Blitz window

What the Blitz weeks cost

addedChecking the Blitz depth against city mobility data

We read four free Chicago mobility series by the Latin America-born share of each place. CTA 'L' entries, Divvy bike trips and bus rides (holding fixed the buses actually run) all ran lower in high-share places during the Blitz. Weekday bus rides there were lower all year, and weekend rides dropped during those weeks. Rideshare trips into those neighborhoods ran a little higher, mostly short weekday rides inside Brighton Park.

Against their own level earlier in 2025, the Blitz weeks sit at most 2 to 7% lower. That supports pricing the Blitz at the deeper late-year rate rather than the year average.

None of these series measures shopping. The two business series point the same way: Illinois sales-tax receipts put high-share suburbs a few percent below their 2025 level during the Blitz, and Chicago business licenses in high block groups lapsed at the prior year's rate while 13% fewer new ones opened.

Limits

Where our method is weakest

The inside-neighborhood line is the largest number here, and it has bounds rather than a measurement. Its low end is one municipality's taxable sales and its high end is four storefronts that spoke to reporters; a curve we chose carries both to Chicago block groups. We price its dollars from the retail and restaurant jobs in each block group at the county's sales per job, so a job at a corner store counts as much as a job at a big-box store. Chicago's own taxable sales put a ceiling of about $650M on the city's Blitz-week total, and the midpoint reading runs above it.

Outside immigrant neighborhoods, the destination side is a model. Change the distance decay and lost sales move between downtown and the suburbs while the total holds. The business counts rest on OpenStreetMap rescaled to Google coverage in Chicago, and the suburbs get the city median because no sweep covers them. The Blitz depth reads UIC's published averages and doesn't estimate the operation's own effect; UIC itself declines to treat September as a separate shock.

UIC's figure is an order-of-magnitude estimate from one foot-traffic panel (95 percent confidence interval $0.83B to $1.67B a year), and every number here inherits that uncertainty.

What we don't estimate

We estimate nothing from Chicago's own foot traffic, and everything causal here is UIC's. A local replication would need Advan or Placer data; our code is ready for it, because observed trips per block group can replace step one's baseline and coefficient.

We separate Midway Blitz from the rest of the enforcement year only through the depth reading. The treatment is a neighborhood measure of exposure to enforcement. It says nothing about any person's status, and the model never touches individual data.

What the estimate leaves out

Everything here allocates UIC's published estimate, and none of it is legal advice. UIC's numbers produce the county total; every choice about which storefront gets what is ours, and each carries an "added" badge. The estimate covers formal, in-person retail and restaurant sales only, the same scope as UIC's.

It leaves out cash and informal commerce, suppliers and closures, and the stores that never opened: Chicago's high Latin America-born block groups got 13% fewer new retail and restaurant licenses during the Blitz than a year earlier, while lapses held steady. It also leaves out lost wages, unpaid rent, missed medical visits and children kept home from school, and it says nothing about who is liable. The ballot question asks the City to find out.

Reference

Every setting and its default

The ward map exposes these settings, and assumptions.json in the repository holds the same names. Change one and the map recomputes every area in your browser.

Sources

• UIC Great Cities Institute, Hunkering Down: The Hidden Economic Cost of Federal Immigration Enforcement in Cook County, IL (Wilson, Lester and Acosta-Córdova, 2026): brief, working paper, technical appendix
• American Community Survey 2019–2023 five-year (api.census.gov): B05006, B01003, B11001, B19013, B26001
• 2020 Census PL 94-171, tables P1 and P5: block populations and group quarters by block group
• TIGERweb: ACS-2023 block groups and 2020 blocks, through the Ward Wise pipeline's cache
• LEHD LODES 8 workplace area characteristics, Illinois, 2023
• © OpenStreetMap contributors (ODbL) via OSM US Layercake pois.parquet; snapshot recorded in the layer spec
• Ward Wise Google Maps sweep, 2026Q3, for coverage ratios (aggregates only)
• Illinois Department of Revenue, local sales-tax distributions by municipality (MT receipts)
• City of Chicago data portal: business licenses, ward (2023–) and community-area boundaries, rideshare trips; CTA ridership; Divvy trip data
• Deportation Data Project ICE arrests; newsroom sightings maps from WBEZ / Chicago Sun-Times, the Chicago Maroon and the Loyola Phoenix; Opportunity Insights Economic Tracker
• Panel audits: Li, Ning, Jing and Lessani, PLOS ONE 19(1): e0294430 (2024); Coston et al., FAccT 2021
• Owner reports: WGN (Oct 17, 2025), WBEZ (Nov 21, 2025 and Feb 20, 2026), ABC7 Chicago (Sept 7, 2026); see the owner-report slide
• Blitz dates: DHS launch September 8–9, 2025; Border Patrol departure November 13, 2025 (CNN, CBS, WTTW)