Specialisms / IFRS 9 / Model Development

Model Development

Whether its building models from first principles or enhancing existing models, we always employ industry standard approaches that are commensurate to our client’s size and ambitions.

Our approach is based upon estimating IFRS 9 ECLs using the standard components of PD, EAD and LGD, with the application of discounting and compliant stage allocation applied.  


The probability of the account moving to default status within the forward-looking period. The PD is typically the most sophisticated component in the ECL framework due to the accessibility to its data drivers, the stronger trends with idiosyncratic and exogenous risk and the higher predictability and robustness from modelling it using statistical techniques.


The exposure or balance at the point of default. The approach to EAD is dependent upon the type of loan: amortising loans are modelled by simulating the pay-down of the capital balance of the loan and revolving is based on modelling the expected balance in the event of default, often modelled in relation to the limit.


The loss incurred given the account has defaulted. The approach to LGD is split by whether the loan is secured by collateral with the expected value of the collateral driving the magnitude of the loss. Modelling LGD can be challenging as it relies on sufficient loss cases although pragmatic workarounds can be deployed.

Stage Allocation

Stage Allocation is the assignment into accounts into 1 of 3 stages depending on the performance of the account.

Stage 1: Performance accounts (not stage 2 or 3.)

Stage 2: Under performing accounts that are either more than 30 days past due or an equivalent or have experienced a significant increase in credit risk which is informed by material shifts in lifetime PD from origination.

Stage 3: Non-performing or accounts in default.


The ECL is calculated by combining the PD, EAD and LGDs with the stage 1 accounts measuring ECLs over a 12-month period and stage 2 and 3 over a lifetime period with the losses discounted using EIR.

The ECLs are generally estimated under multiple economic scenarios with the final ECL being the probability weighted average of all scenarios.

We have incorporated our regulation and credit risk expertise into an affordable, fully integrated and automated IFRS 9 solution. Crucially, we deliver this capability through a secure, hosted platform, which is used to run the IFRS 9 models on an ongoing basis.

Horizon manages all aspects of IFRS 9 Expected Credit Loss models

Horizon models each component of Open Rate, Average Balance, PD, EAD and LGD, providing dynamic reporting for each of these components, and calculating Stage Assignment, Lifetime PD predictions, Expected Credit Loss and Provisions.

Modelling is based on your own data, coupled with a series of parameter settings that give you control over the way the models respond to economic factors. The outputs can be split by segment and into a range of economic scenarios, and the use of economic data means that the system can also be used in stress testing.

All backed by our rich heritage in credit risk

While Horizon is designed for use by analysts within your Risk or Finance function, our consultants are on hand if you need additional support on data definition or IFRS 9 model creation. Horizon is fully transparent and is backed by a methodology that has proven successful in multiple implementations.

Crucially, our IFRS 9 solution is designed with data security in mind – using all our extensive experience of handling sensitive data on behalf of our clients.