A condo questionnaire answer needs its document.

Before a lender finances a unit in a condominium, it has to decide whether the project as a whole meets the investor’s standards: how the association is funded, how many owners are behind on their dues, whether anyone owns too much of the building, whether the building needs repairs nobody has paid for. Almost none of that is visible from the unit. It lives with the homeowners’ association, and the usual way to ask for it is a condo questionnaire.

Fannie Mae’s version is the Condominium Project Questionnaire, Form 1076. The Selling Guide describes it plainly: it helps lenders collect data to determine condo project eligibility, its use is optional, and a substantially similar form may be used in its place. The same topic also says something that matters more for anyone automating the work: lenders are responsible for the accuracy of any information they obtain from sources such as HOAs and management companies.

That sentence changes how a questionnaire should be read. A questionnaire is not evidence. It is a set of answers written by a manager or a board officer, and each answer points to a document the association already holds — the budget, the delinquency report, the reserve study, the insurance certificates, the governing documents. This piece describes how to extract a condo questionnaire so that every answer arrives with the document that should support it, and how to surface the places where the two disagree.

A questionnaire is a set of claims

Read a completed questionnaire and most answers are short: a yes, a no, a count, a percentage, a dollar amount. “Number of units 60 or more days delinquent: 3.” “Is the HOA named in any pending litigation? No.” “Percentage of the budget allocated to reserves: 10%.” Each of those is the end of a calculation or a search that someone did on the association’s side, usually from memory or from a report they ran that day.

Nothing in the form shows the working. The delinquency count does not say which aging report it came from or on what date. The reserve percentage does not say which budget lines went into the numerator and the denominator. A “No” to the litigation question does not say whether anyone asked the association’s attorney. The person filling in the form is often careful and usually right, but the lender cannot tell the careful answers from the rushed ones by reading the form alone.

So the unit of extraction is not the answer. It is the pair: the answer as the association gave it, and the value the supporting document shows, each with its page. Where the two agree, the file is stronger than either alone. Where they do not, the disagreement is the most useful thing the extraction can produce.

Where each answer can be checked

Most questionnaire answers map to one of a small number of association documents. The set a lender typically has, or can request, looks like this:

• The annual budget — assessment income, operating lines, and the replacement reserve allocation. Most of the financial answers are arithmetic on this one document.
• The delinquency or aging report — units grouped by how long their assessments have been past due, often in 30, 60 and 90-day buckets, sometimes with special assessments listed separately.
• The reserve study — components, remaining useful lives, the funded balance and the recommended annual contribution, prepared by an outside specialist on a stated date.
• Insurance certificates and declarations — the master policy, liability and fidelity coverage, with their limits and policy periods.
• Governing documents and minutes — declaration, bylaws and board minutes, where commercial space, rental restrictions, special assessments, inspections and lawsuits tend to appear long before anyone summarises them on a form.

The work is to read each of these as structured data, not as attachments to a questionnaire. Once the budget is a set of lines with amounts, the reserve percentage can be computed rather than copied. Once the aging report is a list of units with buckets, the delinquency count can be counted rather than trusted.

The numbers the Selling Guide asks about

The questionnaire exists because the investor has specific tests. Fannie Mae publishes them in Chapter B4-2 of the Selling Guide, and several are numeric. In the versions of the topics dated 08/05/2026, the full review process and the list of ineligible projects include, among others:

• Reserves — the budget must fund replacement reserves for capital expenditures and deferred maintenance of at least 10% of the budget, measured as the annual reserve allocation divided by the association’s annual budgeted assessment income.
• Delinquent assessments — no more than 15% of the total units may be 60 days or more past due on common expense assessments, and the same limit applies to each special assessment.
• Single-entity ownership — in projects of 21 or more units, a single entity may not own more than 20% of the units; smaller projects have lower unit counts.
• Commercial space — the total space used for nonresidential or commercial purposes may not exceed 35%.
• Critical repairs — projects needing critical repairs are ineligible, and the topic’s examples include unfunded repairs costing more than $10,000 per unit that should be undertaken within the next 12 months.

The same topics also cover pending litigation, projects operated as hotels or motels, and other characteristics that are not numbers at all. Two cautions apply to all of it. These are the investor’s tests, and they are revised; the topic dates are part of the citation for a reason. And the tests belong to the lender’s underwriting, not to the extraction. A pipeline that reads documents should deliver the values the tests need, with their sources. It should not decide that a project passes.

The arithmetic the form hides

The reserve percentage is the clearest case of an answer that is really a calculation. The Selling Guide does not just set the 10% floor; it says how to compute it. The lender divides the annual budgeted replacement reserve allocation by the annual budgeted assessment income, which includes regular common expense fees. Certain income may be left out of that denominator: incidental income the project does not rely on, income collected for utilities owners would normally pay themselves, such as cable or internet, income already allocated to reserve accounts, and special assessment income.

Two associations with the same budget can answer the question differently depending on which lines they include. One manager divides by total revenue, interest and laundry income included, and reports 9.4%. Another divides by assessment income only and reports 10.2%. Both numbers are honest. Only one follows the method the lender has to apply. If the extraction carries the budget lines themselves — each income line with its amount and its category, the reserve contribution as its own line — the percentage can be recomputed the same way every time, and the questionnaire’s figure becomes a check rather than an input.

The reserve study adds a second path. The Selling Guide allows a lender to rely on an acceptable reserve study instead of the 10% calculation, under conditions it lists, including that the budget carries the highest recommended reserve allocation in the study. That is another comparison between two documents: the contribution the study recommends and the contribution the budget actually funds. Extracting both, with their pages and the study’s date, is what makes that comparison possible.

Delinquency works the same way. The question asks about units 60 or more days past due, but aging reports group balances in buckets that do not always line up with the question. A report with 31–60, 61–90 and 90+ columns has to be summed carefully, and a unit with a small balance in an older bucket is still a delinquent unit. Special assessments are tested separately, so a report that merges them into one balance cannot answer the question as asked. The extraction should count units, per bucket and per assessment type, and say when the report’s structure does not allow the count the question needs.

Question, document, divergence, action

The table below shows the shape of the output for a fictional project, Harbor Point Condominium, 120 units. The questionnaire was completed by the management company; the documents came from the association’s resale package.

Three things in that table do work that a copied questionnaire would not. The reserve row shows the arithmetic, so a reviewer sees in one line why 10% and 9.5% are both on the page. The litigation and assessment rows come from minutes, a document the questionnaire never mentions, which is where boards record decisions months before they reach a form. And the single-entity row does not resolve the question; it raises it, because whether two LLCs at one address are one owner is a judgment, not a match.

Dates decide which answer is current

Every document in the package has its own date, and they rarely agree. The questionnaire is signed on one day. The budget covers a fiscal year. The aging report is run as of a date. The reserve study was prepared, perhaps, three years ago. The insurance certificate covers a policy period that may end before the loan closes. Minutes are dated by meeting.

An extraction that drops those dates loses the ability to explain most disagreements. A delinquency count of 3 on the questionnaire and 7 in the aging report may both be correct if one is from June and the other from September. A budget from the prior fiscal year can make the reserve answer look wrong when the current budget fixed it. The record should carry the as-of date for every document and every answer, and it should flag answers whose supporting document is older than the questionnaire or from a different fiscal year. That flag does not say who is right. It tells the reviewer which document to ask for.

Not the estoppel certificate

Condo questionnaires and estoppel certificates come from the same association and are often requested in the same week, so they are easy to confuse. They answer different questions. An estoppel certificate is about one unit: what that owner owes the association, as of a date, so the closing can pay it. We covered it in The estoppel certificate closes the association’s account. A condo questionnaire is about the project: how the whole association is funded, governed and maintained, so the lender can decide whether to lend in the building at all.

The documents overlap — both lean on the budget and the delinquency report — but a value extracted for one should not be reused for the other without checking which question it answers. The unit’s own balance belongs on the certificate. The count of delinquent units belongs on the questionnaire.

What the extraction does not decide

It is worth being clear about the edges. Extraction does not determine whether a project is eligible. Fannie Mae’s process runs through its Condo Project Manager, where the lender completes its own certification and documents the decision in the loan file; insurance compliance is validated against separate chapters of the Selling Guide; and some review paths require a qualified attorney to confirm the project’s legal documents in writing. None of that is replaced by reading documents faster.

Extraction also does not interpret. Whether a demand letter is pre-litigation activity, whether two owners are a single entity, whether a deferred repair counts as critical — these are judgments for the people who sign the file. What extraction can do is make sure those people see the minutes, the roster and the engineering report next to the answer that summarised them, instead of only the answer.

What the extraction has to deliver

For each questionnaire, a useful record carries:

• The answers — every questionnaire response as its own field, with the question text and the page.
• The documents as data — budget lines, aging buckets, reserve study figures, insurance limits and periods, owner counts, each with its source.
• The computations — reserve share, delinquent-unit counts, commercial-space share and largest-owner share, recomputed from the documents with the method written down.
• The dates — the as-of date of every answer and every document, and which fiscal year each financial figure belongs to.
• The flags — each place where an answer and its document disagree, or where a document cannot answer the question as asked, with a plain reason and the request that would resolve it.

This is the scope the real estate solutions page describes for HOA document indexing: CC&Rs, bylaws, budgets and reserve studies extracted for estoppels and condo questionnaires, with the source of every value attached, for a person to act on.

Keep the answer and the document together

A condo questionnaire compresses a building’s finances, ownership and condition into a few pages of short answers. That compression is useful, and it is also where errors hide: a denominator chosen differently, a report run on another day, a board decision that never reached the form. Reading the questionnaire alone leaves the lender trusting a summary it is responsible for.

Extract the answer and the document side by side, recompute what can be recomputed, keep every date, and show every disagreement with both sides and their pages. Then the questionnaire becomes what it should be: an index into the association’s documents, checked against them, rather than a substitute for them.

Sources

• B4-2.1-01, General Information on Project Standards (Fannie Mae Selling Guide, 08/05/2026) — the Condominium Project Questionnaire (Form 1076) as optional, a substantially similar form allowed, and lenders responsible for the accuracy of information obtained from HOAs and other sources.
• B4-2.2-01, Full Review Process (Fannie Mae Selling Guide, 08/05/2026) — the 10% replacement reserve calculation and its exclusions, the reserve study alternative, the 15% limit on units 60 or more days delinquent, and the insurance chapters B7-3 and B7-4.
• B4-2.1-03, Ineligible Projects (Fannie Mae Selling Guide, 08/05/2026) — single-entity ownership, the 35% commercial space limit, pending litigation, critical repairs including unfunded repairs over $10,000 per unit within 12 months, and projects operated as hotels.