Brooklyn research brief ·

What Brooklyn’s condo tax records can tell us

NYC’s condo tax abatement is generally for owners who live in their units. We looked for units where the owner might qualify, using the address where the City sends the tax bill as a clue.

2,400 candidate tax lots in the random sample
Exact match, absent from Aug. 2026 credit extract
687
Exact match, listed in that extract
99
Different street address or apartment
1,542
Incomplete or conflicting address
72

An exact match means the street address and apartment on the billing record agree with the assessment roll after standardizing their format.

What the model found

The study screened 54,166 Brooklyn residential Class 2 condo tax lots. Of those, 15,512 passed the bulk filters and had a positive modeled benefit. The address sample came from this smaller group.

$8.19 million

Modeled annual gross tax reduction in Brooklyn, assuming every exact match absent from the August 2026 credit extract qualifies and receives the benefit. The model uses FY2026 tax figures, before fees and any compliance costs.

The adjusted sampling interval is $7.27–$9.11 million. It measures variation from drawing a sample. Errors in the address rule, eligibility, assessment inputs and later changes can move the result beyond that interval.

About 74% of the modeled dollars came from developments flagged by our screen for a possible prevailing-wage affidavit requirement. The screen used development size and average assessed value. It did not verify wages, affidavits or the City’s determination for any development. Removing every flagged development would bring the modeled total to $2.10 million. That scenario illustrates how much of the result depends on that group.

About $5.99 million (73%) came from developments without a positive condo abatement row in the FY2026 extract. The other $2.20 million (27%) came from developments with a positive record. Both groups can include developments flagged for the prevailing-wage requirement.

We observed a billing and credit-record pattern. The records establish neither who lives in these units nor what the City owes. The study did not establish a refund, an approved application or a credit posted as a result of Atlas’s work.

How we counted

The study covers Brooklyn Class 2 residential condo lots in building classes R1, R2, R3, R4 and R6. Co-ops, condops, rental-condo blocks, commercial condos and Class 1 condos are outside its scope.

  1. Screen the FY2026 roll

    We counted each tax lot once and grouped lots by condo development. A development can span multiple buildings. We screened out lots with a condo abatement record, apparent entity or ambiguous ownership, more than three units under the same recorded owner name within a development, and potentially incompatible or unclear tax benefits. Some of these exclusions can omit legally eligible cases.

  2. Check the billing address

    We drew 2,400 lots across 1,045 developments using a fixed random ordering of the 15,512 candidates, without replacement. We obtained each selected page on November 4, 2026. Exact address and apartment matches counted toward the estimate. Other addresses and unresolved cases counted as zero. A matching address can be used by someone who lives elsewhere, and an owner who lives in the unit may have a bill sent to an agent or another address.

  3. Estimate the benefit and check a newer extract

    We applied the FY2026 Class 2 rate of 12.439% to billed assessed value less billed exemptions, subtracted modeled compatible credits and applied the development’s abatement tier of 17.5% to 28.1%. Where the City’s development value or tier was unavailable, we used the mean assessed value of its residential lots to model the tier.

    Using only the FY2026 credit file, the model gives $9.64 million (adjusted sampling interval $8.66–$10.62 million). We then removed 99 sample matches that appear in the FY2027 Q2 extract dated August 15, 2026. The resulting $8.19 million retains the FY2026 assessments, tax rate and screens. An absence in that quarter’s file leaves open credits recorded in other quarters or later adjustments.

Sampling formula and address scenarios

Let N be the candidate count, n the sample count and s the sample standard deviation. We multiplied the mean modeled dollars per sampled lot (including zeroes) by N. The standard error is N × s × √((1 − n/N)/n). The approximate t interval applies a Bonferroni correction across 24 checks, one per 100 lots, for a nominal 95% level across those checks. The Q2-adjusted interval has an 11.2% relative half-width at 2,400 lots, above the study’s 10% precision target.

Illustrations using the August 2026 exclusion. The assumed error rates have no measured basis.
ScenarioAnnual model
Remove 10% of exact-match dollars, with no added matches$7.37 million
Remove 25% of exact-match dollars, with no added matches$6.14 million
Keep all exact-match dollars and assume every unresolved case absent from Q2 qualifies$9.26 million

The last illustration leaves all differently addressed lots at zero. None of these scenarios measures actual occupancy or brackets what the City might award.

Data and sources

The aggregate file includes the counts and sample totals needed to recalculate both estimates and the adjusted sampling intervals.

Download aggregate data (CSV)

The same FY2026 roll classifications cover 227,157 residential condo tax lots across the five boroughs. We estimated abatement amounts for Brooklyn only.

About this project

Joel Witten led this independent study and is developing Abatement Atlas as a potential paid service to help Brooklyn condo boards with routine abatement renewals. The research also covers developments that may need to apply for the first time. The proposed pilot serves condos that already receive the benefit. It is gathering feedback, with no paid engagements open. The pilot’s sample board packet uses fictional records and outcomes. Proposed renewal pilot

Send a research question or correction