IFRS 9 ECL software in Bangladesh now has a fixed deadline. Scheduled banks have until January 2028 to bring credit facilities under Bangladesh Bank's guidance, and ECL must be estimated every quarter. Rust builds the PD, LGD and EAD curves, scenario runs and audit trail that routine needs, while your credit risk team keeps every judgment call.
IFRS 9 ECL software is a system that calculates expected credit loss from a bank’s own loan data and keeps every assumption and result in one traceable record. Rust IFRS 9 software is Prima’s version, built by its credit risk modellers, and it runs in the cloud or on a bank’s own servers.
Bangladesh Bank issued BRPD-1 Circular No. 06 of 8 March 2026 as the minimum requirements for IFRS 9 in scheduled banks, after a roadmap letter of January 2025 set up each bank’s IFRS 9 team. Loans and other credit facilities come under it by January 2028, and investment securities by January 2029.
The guidance reads like a specification for a model. It sets five years of credit data as the floor, at least three scenarios where probabilities are weighted, ECL estimates every quarter, and an annual validation by someone who did not build the model. A workbook can meet that on paper, but it rarely survives the first quarterly cycle.
The legal basis comes from the Financial Reporting Council. The Financial Reporting Act 2015 treats banks as Public Interest Entities, and an FRC gazette dated 2 November 2020 requires them to follow IFRS. So banks must apply IFRS 9, and the practical task is running ECL each quarter on a model that management can explain.
Circular 06 converts the roadmap into a floor of requirements, and several of them decide what software has to do. It sets a floor of five years of historical credit data, refreshed every quarter, and at least three weighted scenarios with documented forward-looking inputs. It also lets a bank develop its model, buy one or take one from an overseas parent, and it expects management to explain whichever it picks.
Rust covers these points as software. It fits PD, LGD and EAD curves over each instrument’s life, runs scenarios from one model, shows every calculation step, and compares initial and final ECL for the quarterly statement. Your team still sets the assumptions, signs off overlays and answers to the board.
Circular 06 also says a validator must not have built the model, so Prima’s Rust team will not validate its own setup. A bank using Rust should plan an independent ECL model validation early, because the validator must be notified to Bangladesh Bank before appointment.
What your team needs: A repeatable run that gives the same result from the same inputs.
How Prima helps: Rust runs the PD, LGD and EAD curves as a configured process, so quarter-end stops depending on one workbook.
What your team needs: One data set of at least five years that refreshes every quarter.
How Prima helps: Rust loads past and current portfolio data and runs validation checks before go-live. Your team records the reason for any proxy and the date it will be replaced.
What your team needs: Three or more weighted scenarios with documented inputs, rerun when conditions change.
How Prima helps: Macroeconomic factors feed every scenario run, so a new assumption set means changing a setting rather than rebuilding tabs.
What your team needs: A method managers can describe to Bangladesh Bank, even when software or consultants did the build.
How Prima helps: Rust shows each calculation step, and training covers the method, so the explanation stays with your own staff.
What your team needs: Records that an independent validator can follow from a result back to its source.
How Prima helps: Every figure traces back through the calculation. Prima will not validate its own Rust setup, so you appoint a separate validator.
What your team needs: A clear view of what a bought model must show.
How Prima helps: Circular 06 allows internal builds, bought models and parent models. A bought model must fit your data and portfolio, and the scoping step tests that fit on your own loan data.
Track A: Curves, Scenarios and ECL Movement
| Feature | What It Means for Your Bank |
|---|---|
| Curve modelling for PD, LGD and EAD | Default probability, loss severity and exposure at default come from one structure over each instrument's life, not from linked workbooks. |
| Forward-looking scenario runs | Several macroeconomic scenarios run through the same model, and the weighted result updates without a rebuild. |
| Instrument lifecycle coverage | Measurement runs from initial recognition to the reporting date, which suits the January 2028 start for credit facilities and January 2029 for securities. |
| Opening-to-closing ECL movement | Initial and final ECL values sit side by side and show why allowances moved during a period. |
| Configurable dashboards | Cash flow and ECL results appear in views that risk, finance and audit teams can adjust. |
| Calibration options | Calibration choices widen as your portfolio and your model mature. |
Track B: Controls, Hosting and Support
| Feature | What It Means for Your Bank |
|---|---|
| Cloud or in-house hosting | Your IT team picks the hosting model that fits the bank's core systems and data policy. |
| Role-based access | Each user sees only the data and functions that fit their role, which limits accidental exposure of loan records. |
| Multi-factor sign-in | Access needs more than one proof of identity, adding a second barrier against unauthorised use. |
| Training and support | Your staff learn the ECL mechanics and first-time adoption steps, so the method stays with them after go-live. |
| Parallel run support | Rust runs beside current provisioning for a full reporting cycle, so differences between ECL and the regulatory provision show early. |
| Step 1: Data Readiness Check | Prima's team reads your loan data, portfolio structure and current provisioning method, then measures how far your history reaches against the five-year minimum. | OUTPUT: A data-readiness note and a scoped plan. |
| >>> | ||
| Step 2: Environment Setup | Rust's curve structures, calibration and dashboards are set up for your portfolios and your hosting choice. | OUTPUT: A Rust environment tuned to your bank. |
| >>> | ||
| Step 3: Data Load and Checks | Past and current loan records load into Rust and pass validation checks before anyone relies on the results. | OUTPUT: A checked data set with open issues closed. |
| >>> | ||
| Step 4: Side-by-Side Run | Rust runs next to your current provisioning through a full reporting cycle at minimum, so differences between ECL and the regulatory provision show early. | OUTPUT: A comparison of Rust results with today's figures. |
| >>> | ||
| Step 5: Handover and Support | Your team becomes the system owner, with Prima's training and support through the first live runs. | OUTPUT: A live quarterly ECL process, with training and support agreed. |
Corporate, SME and retail books where loan volume strains a workbook. Circular 06 covers both funded and unfunded credit facilities.
Islamic banks are scheduled banks too, so the same guidance reaches them. Map your product set to the staging rules before configuration starts.
Circular 06 lets a bank take a model from an overseas parent or Head Office. The branch still needs managers who can explain it locally.
Teams whose ECL logic lives in linked workbooks and depends on a few people.
Banks that hold under five years of credit data and need a plan for documented proxies.
IFRS 9 reaches the bonds and receivables insurers hold. Circular 06 is addressed to banks, so confirm which rules apply to you.
Circular 06 addresses scheduled banks only. Confirm which guidance applies to you before choosing ECL software.
Trade receivables under the simplified approach usually need a provision matrix, not a full ECL engine. Rust is built for lending books.
Five years of data, quarterly runs and an annual validation must be in place by January 2028. A bank that finds a data gap in 2027 has little time to rebuild history, and a software rollout adds a parallel run on top.
An unexplained model draws regulatory questions as well as audit findings, since managers have to describe how it works, whoever built it. So a rollout should train your own staff from the first step.
Overlays need board approval and a documented basis. Teams that add them by hand in a workbook will struggle to show that trail, so settle where overlays live before the first quarterly run.
A demo uses a sample of your portfolio, so you see IFRS 9 ECL software in Bangladesh working on figures you recognise: PD, LGD and EAD curves, scenario runs and ECL movement. Book a session and we will scope it against your reporting calendar.
Rust is Prima's IFRS 9 platform for expected credit loss. It fits default, loss and exposure curves, runs macroeconomic scenarios and compares initial and final ECL values, while your bank keeps the method, assumptions and sign-off that Circular 06 puts on management.
No clause requires a specific tool. Circular 06 lets a bank develop its model, procure one, or take one from an overseas parent or Head Office. The bank answers for the result in each case, so software helps most where loan volume or quarterly timing strains a workbook.
Loans and other credit lines, funded or not, come under the guidance by January 2028. Investment securities and the remaining IFRS 9 instruments follow by January 2029. Bangladesh Bank may add directives, so check its circulars before you fix a plan.
A bought model must suit the bank's own data, portfolio and credit experience, and management must be able to explain how it works. The guidance also asks banks to weigh how well the model links to core banking systems for data extraction, and to control any manual transfer.
Use at least five years of historical credit data, refreshed and reviewed every quarter. Proxies may stand in for missing records, briefly and with a written reason. So a bank with a short history should start the data work first.
A bank that weights probabilities should consider at least three scenarios, for example worst, normal and best case. Forward-looking inputs need documentation, and a change of source from one period to the next needs a stated reason. Rust runs several scenarios through one model, so extra runs add little effort.
Each quarter a bank files a statement of changes in ECL allowance within 30 days, plus a bridge between ECL allowance and regulatory provisions. Rust's comparison of initial and final ECL supplies the ECL side of those figures. Finance still prepares the filing.
No. Validators must not have helped build the model, the bank must tell Bangladesh Bank before it appoints one, and results go to the regulator within 30 days. So a bank that uses Prima to configure Rust should appoint a separate validator.
Either. Your IT and compliance leads decide which one meets the bank's data policy, and Prima sets Rust up to match.
An Excel model often rests on two or three people and takes time to defend at audit. Rust runs the same curve logic as a configured process and shows how each figure was produced.
Rust limits access by role and asks for multi-factor sign-in. Prima also holds ISO 27001 certification.
Timing turns on how big the portfolio is, how ready the data is and which method you use today, and the first step sets a plan against your own data. A scoping session covers your portfolios, hosting choice, data history and reporting calendar.
Share your portfolio size, hosting preference and reporting calendar in the form below. A Rust specialist will reply with a suggested next step.
We book a call at a time that suits you
We hold an introductory meeting
We send a written proposal
