How Banks Calculate the AECB Credit Score: What Your Bank Actually Weighs

Behind the Curtain: Who Actually Computes the Number

Start with the most common misconception: banks do not calculate your AECB credit score. The number, from 300 to 900, is produced by Al Etihad Credit Bureau, the UAE's federal credit bureau, from records submitted by banks, finance companies, telecom operators and utility providers. Your bank is a data supplier and a data consumer; the scoring model itself sits at the bureau.

The confusion is understandable and persistent — enough people type 'how bank calclates AECB credit score' into search boxes, typo and all, that the question clearly needs a straight answer. What banks really do is calculate around the score. They feed it, they weigh it alongside the debt burden ratio, and they overlay internal policy that can approve, price or decline a file the bureau data alone would not predict.

Separating the two layers is the key to understanding every credit decision you will face in the UAE. The bureau layer is descriptive — it summarises your recorded behaviour. The bank layer is commercial — it decides what that summary is worth to this lender, for this product, at this moment. This guide walks both layers, from the monthly data pipeline to the underwriting desk, and finishes with how to audit the inputs yourself.

The Data Pipeline: From Monthly Feeds to Your File

Everything begins with furnisher submissions. Each bank, finance company and reporting utility sends structured records to AECB on a regular cycle, typically monthly: account openings, limits, balances, instalments, and payment conduct down to days past due. The bureau consolidates these into a single file keyed to your Emirates ID, which is why the file follows you across emirates and employers rather than resetting with each move.

The scope of the feeds is wider than bank credit. Telecom operators report account conduct; utility providers such as DEWA in Dubai, ADDC in Abu Dhabi and SEWA in Sharjah contribute payment records; and certain legal information, including bounced-cheque matters, can surface under the bureau's mandate. Even rental-adjacent behaviour leaks in indirectly — a household juggling an Ejari-registered tenancy's dated cheques and a Mollak-managed service charge leaves exactly the kind of payment trail the system is built to read.

Cycle timing matters for anyone planning an application. Because submissions are periodic, the file is a snapshot of the most recent feeds, not a live window: a card paid down today reads stretched until the next cycle lands. Underwriters know this and look for trends across cycles, which is why three consecutive clean months move a file more than one perfect week.

The Ingredients: What the Scoring Model Commonly Weighs

AECB does not publish its model weights, and any article claiming exact percentages is guessing. What is well established, from the bureau's own descriptions and consistent UAE lending practice, is the family of inputs the score draws on, listed below. Read them as categories with relative influence, not as a formula.

The practical reading is that conduct dominates: paying on time and using a modest share of available credit outweighs almost everything else you can optimise. Salary size, notably, is not a direct score input — the bureau scores behaviour, and income enters the decision later at the bank layer through affordability testing. That split explains why high earners with messy habits can score below modest earners with boring ones.

Each factor is also time-sensitive in a different way. Utilisation responds within a cycle or two; enquiry clusters fade over months; derogatory records decay over years, with recency weighed heavily. That staggered sensitivity is why score repair is sequenced rather than simultaneous, and why the same file rewards patience differently at different starting points.

• Payment history: whether instalments and statements were paid on time, and how late when they were not.
• Utilisation: the share of card limits and facilities actually in use at reporting time.
• Depth and age of credit: how long facilities have been running and how many you manage.
• Recent credit-seeking: the volume and spacing of applications and lender enquiries.
• Types of facilities: the mix of cards, personal loans, auto finance and mortgages on file.
• Derogatory records: defaults, write-offs, bounced-cheque matters and collection items where furnished.
• Adjacent payment conduct: utility and telecom records such as DEWA, ADDC and SEWA data where providers report.

The Debt Burden Ratio: The Ceiling That Frames Everything

Parallel to the score runs the number banks are regulated around: the debt burden ratio, the share of documented income already committed to debt repayments. UAE Central Bank retail banking frameworks have commonly been cited capping this at around 50% of gross income for individuals, counting proposed instalments, existing loans and a proportion of card limits — verify current figures. Individual banks apply stricter internal ceilings on top of the regulatory floor.

The interaction is where applicants get confused. A strong score does not override the burden ceiling, and a modest score does not shrink it; the two tests run simultaneously. A 780 score with 45% of income already committed may approve for less than requested, while a 660 score with a 20% burden and clean history can sail through at standard terms. Banks size loans at the intersection of the two tests.

For planning, compute your own burden before applying: total monthly obligations — loans, minimum card payments, anything furnished — divided by documented income, then test the proposed instalment on top. The levers when it is tight are unglamorous and reliable: a smaller loan, a longer tenure where age limits allow, or clearing a small facility before the application. The score cannot do this arithmetic for you; only the file's structure can.

What the Bank Adds: Internal Scoring and Policy Overlays

On top of the bureau data sits the lender's own machinery. Banks run internal scorecards tuned to their book: employer categories, sector exposure, income stability, account conduct with the bank itself, and history with the product being requested. Two lenders can read the identical AECB report and reach different decisions because the models answer different questions about risk appetite.

Policy overlays do much of the quiet work. Salary transfer requirements, minimum income floors by product, expatriate versus national down-payment tiers — commonly cited at 20% for expatriate first homes and 15% for UAE nationals under Central Bank frameworks, verify current figures — and sector-specific cautions all shape outcomes before the credit officer sees anything. This is why identical files meet different ceilings at different institutions, and why a decline from one bank is not a verdict on your file.

Islamic home finance follows the same logic with different vocabulary. Diminishing musharaka and ijara structures run on the same bureau pull, the same burden ceiling and the same internal overlays; the Sharia structure changes the contract, not the credit test. Applicants sometimes assume otherwise and under-prepare the file — the safer assumption is that everything in this guide applies, and to verify product-specific rules with the provider.

Why Two Identical Salaries Get Two Different Answers

Consider the classic pair: two applicants, the same AED 25,000 salary, the same requested loan. The first carries three cards near their limits, a fresh car loan and two applications last month. The second carries one old card at a tenth of its limit, a finished personal loan marked settled, and no enquiries for a year. The bureau layer sees very different behaviour; the bank layer prices and sizes accordingly.

The first file's problem is not any single item but the story the items tell together: high utilisation plus new credit-seeking plus young facilities reads as a household running ahead of its income. The second file reads boring, and boring is what underwriting rewards. Neither applicant knows their own file reads this way until the report is pulled — which is why the self-pull before applying has become standard broker advice.

The corollary is that identical scores can still diverge. A 700 built on a stable five-year card history and a 700 assembled last year from a thin file are not the same risk to an internal model, because depth and age differ even where the headline matches. Chasing the number alone misses what the number is made of — and the ingredients, unlike the output, are fully within your control.

Enquiries, Rate Shopping and the Multi-Application Trap

Every time a lender pulls your file for an application, the enquiry registers. One or two are unremarkable; a cluster inside a month tells the model you are shopping hard, and tells the bank reading it that other institutions have already had their chance. The scoring impact of enquiries is modest compared with payment history, but the human impact at underwriting is larger than the points suggest.

The fix is batching, not abstinence. When genuinely comparing mortgage offers, submit the serious applications within the same short window so they read as one shopping event, and keep exploratory pre-checks informal until you are ready to commit. Space out facility openings — a car loan in March, a card in May, a personal loan in June is a pattern, not a coincidence, as far as the model is concerned.

Be equally deliberate about who pulls. Sign-up bonuses, store cards and BNPL sign-ups all generate enquiries, and none of them helps a mortgage file. The months before a major application are a curfew period: no new credit of any kind, every existing facility paid on time, utilisation drifting down. Verify how your specific lender treats enquiry windows — implementations differ — but the conservative pattern protects you under every version of the rules.

How the Number Behaves Over Time

The score is dynamic in a very specific rhythm: it steps when furnisher data lands, typically monthly, so it moves in staircases rather than slides. A paid-down card shows at the next cycle; a dispute correction lands the cycle after the furnisher confirms; a new loan drags the file the month it reports. Applicants who understand the staircase stop checking weekly and start planning in cycles.

Negative entries weigh by recency, not just by existence. A late payment from five years ago that has been followed by clean conduct barely registers at underwriting; the same marker from last quarter colours everything. Derogatory records persist for periods commonly cited in years and vary by entry type — verify current retention rules with AECB — but their practical weight decays long before their administrative life ends.

Recovery is faster than most fear and slower than most hope. Utilisation-driven dips can lift within a couple of clean cycles; history-driven rebuilds take months of consistent reporting. The asymmetry works in your favour if you start early: the best day to begin feeding the model clean data is the day you first imagine needing a mortgage, not the week the adviser asks for the report.

From Score to Offer: How Pricing and Limits Actually Set

Approval is binary but pricing is a slope. The same approved loan can carry a different profit or interest rate, tenure and maximum amount depending on where the file sits in the lender's risk bands, and the AECB score is a first-order input to that banding. Stronger files negotiate; weaker files accept; the difference over a 25-year mortgage is measured in tens of thousands of dirhams, not rounding errors.

Amount sizing follows the burden ceiling and the property maths together. The deposit tiers, the DLD transfer fee of 4% in Dubai, mortgage registration at 0.25% of the loan plus AED 290, valuation and trustee charges — all are cash costs that apply regardless of score — while the score shapes what financing is available on top. A file repaired before application therefore improves both the borrowing terms and the timeline, which has its own cash value in off-plan and resale negotiations.

Remember that offers expire and conditions attach. Rate holds, valuation validity and pre-approval windows run in weeks, and a file that drifts during the window — a new card, a job change, a missed bill — can re-price at the finish line. Treat the period between offer and transfer as part of the underwriting: the behaviour that earned the offer is the behaviour that keeps it.

Auditing the Machine: Check Your Own Inputs

Because the whole edifice runs on furnished data, the highest-leverage consumer action is a periodic audit of the raw material. Pull the full report through aecb.gov.ae or the AECB app — fees are modest and published; verify the current schedule — and read it against your own records: every facility you hold, its limit and balance, its payment rows, every enquiry, every utility and telecom entry.

Where a line is wrong, dispute it with evidence: settlement letters, closure references, final bills marked paid. Where a line is right but unflattering, the correction is behavioural, and the sequence is known — bring arrears current, push utilisation down, stop new applications, let cycles accumulate. Where the file is thin, time and a few modest, well-managed facilities build the depth internal models like to see.

Do the audit once a year, and again three months before any major application, and the credit decision stops being a black box. You will know what the bureau layer says about you, what the burden arithmetic permits, and which bank policy overlays are likely to bind. That is precisely the information the underwriter will use to price your file — and in credit, as in property, the party who has already done the homework negotiates from the strong side of the table.

• Facility inventory: every account, limit, balance and instalment exactly as the file reports it.
• Payment rows: late markers and days past due, matched against your own bank records.
• Enquiries: every pull, its date and whether you initiated it.
• Utility and telecom entries: DEWA, ADDC, SEWA and operator accounts marked current and closed where applicable.
• Personal data: name and Emirates ID details free of mismatches and fragments.
• Stale entries: settled facilities still reading active, flagged for dispute with evidence attached.
• The score's driver notes: what the bureau says is pulling your number down, as your repair priority list.