What have banks achieved so far with their IFRS 9 credit risk models?

The new standard is a game changer. IFRS 9 credit risk modelling and its implementation have been complex processes. Reforming the whole approach from an incurred loss to an expected loss required a clear and robust strategy; proper and timely implementation; and accurate calculations. The banks now have a clearer view on the impairment model, the additional provisions, the impact on the capital and P&L volatility, and able to build a strategic adjustment to their operating business model.

I see the IFRS 9 credit risk models as a comprehensive framework within the structure of an organization. It is being run through a policy and procedures outlining processes, responsibilities, systems, feeding channels, etc and this is a work in progress. There is still more to improve, stabilise and fine tune the model. We have seen the impact on the balance sheet and the costs from impairment increase, but this has set the ground for higher financial stability in my opinion.

IFRS 9 credit risk models have become a management tool which serve to the purpose. The achievement is beyond the banks only, no more “too little, too late”. It should build trust to public and internally solve important problems by strengthening risk framework and provisioning. On the other hand, it has also provided the means for the banks to understand how to optimize their business models for this higher risk coverage against an expected loss event. Business strategies need now to be adjusted, or change, and the sooner a complete optimization the better for a bank.

What is the importance of the auditing process in the improvement and stabilisation of IFRS 9 credit risk models?

IFRS 9 is a new standard; we had an idea of what to expect as discussed with consultants and auditors, but not exactly in what way it would impact and by how much. We built a strategy and we would have not been able to produce the results we generate now through the model if we did not cooperate closely with the consultants and the auditors. In addition to the recommendations on the development process itself, it has been a lot of work with the set of historical data, policy and procedures, the system and making all the pieces effective and operational, and compliant to the regulation. We were improving as moving forward.

Audit committee is the structure to oversight the process, to make sure we follow the strategy correctly. Crucial has been the recalibration for us as a bank. After the first tests of the new model were submitted to the auditors, they were questioning the soundness of the model, identifying risk calculation misstatements, pointing out judgements of subjectivity and making sure that all of it is as per policy. I will be honest, I was taking it as much of scepticism at first, and had the model improved as per the deficiencies pointed out, but that is their job. It has been their valuable contribution into the whole process. It has been undoubtedly a crucialcontribution to what is the final implementation.

The auditors run their own tests, sensitivity analysis, examining and assessing reasonableness of events and bank’s control functioning for any potential subjectivity or bias, and the results be disclosed. That is how we can then understand how all of it would affect in the long-term use of the model. This way, the whole framework was improved for us at local level; we followed most of the recommendations. I recall and compare the IFRS 9 credit risk model we have in use now with the one at first testing phase; we are definitely using a more accurate, an overall better one now.

What considerations will a smaller bank need to make in relation to the data they hold and use to strengthen IFRS 9 models?

It is a ‘no data’, a ‘problem’ and ‘no excuse’ issue. The smaller banks were estimated to be hit harder by the change of the standard and the result is more of a simplistic approach they follow. In some cases, it is a problem of lack of historical data for a certain type of portfolios. And this is a common cause in my opinion. If there is no data for an accurate calculation under IFRS 9, the option available is to use benchmarking or proxy data, but that is not the best solution, only an acceptable one.

A business cycle data can be considered for the moment as the IFRS 9 is mandatory and not optional, but it is never too late to start building and collecting the data you need. Similar to the standard itself and the credit risk models it requires to be applied, it is a continuous process that can be improved with time. Te right data can be collected as we go. We started a few years ago in improving data collection processes and though not where it needs to be, data pool and quality have improved. Small or big bank, it can be a cultural problem too.

Most of EU counties are solid on this, and going East becomes a problem. We did not have the data required for the IFRS 9 and following a simpler segment-based calculation approach, we have used data from similar behaviour portfolios. It is a challenge, the lack of data or its poor quality would in certain cases produce incorrect expected credit loss results and lead to significant amount of provisions.Improvement comes with investment in processes, systems, people, but that is the way forward if we want to build models that last. More costly for the small banks combining it with the IFRS 9 provisions increase, it has to be a full move forward, investments, the right processes, people and effort to influence a change of culture.

What will be the next steps to take in order to better IFRS 9 credit risk models once the auditing process is complete?

After the implementation of the IFRS 9 the banks have seen their balance sheet change, understanding the P&L volatility and the key areas influenced too, the need for adjustments in the business model is a necessity. In addition to data capture and interpretation, the back testing, benchmarking, sensitivity analysis, are a continuous process which with the assistance of the auditors too leads to further fine tuning of the IFRS 9 credit risk models and the integration with the existing models.

Last year reporting was first time under IFRS 9 and going forward it needs time to improve the model. Being as complex as it is, improving the IFRS 9 credit risk models requires a whole structure of factors influencing the accurateness of the model functionality; people, knowledge, data, external factors, systems, etc.

Banks could adjust their business strategy and revise lending network;however, it is essential to improving the result based on the weight of influence of an event, financial situation, forward looking factors correlation, forecasts, etc. Strengthening the monitoring systems and training the relationship managers toofor mitigating and taking action steps for preventing migration to Stage 2 is important. Digitization for preventing manual intervention, data interpretation, people’s knowledge, etc can provide a key step in understanding potential issues related to stages classification, migration in between Stage 1 and 2 regardless of loan’s days past due.

I see these as a whole structure holding a network of channels feeding the IFRS 9 model pool like tools, processes and policies all contributing to the enhancement of the model. I can bring my own experience when after the first testing of the model we were trying to understand what had gone wrong; human error, the next time; lack of data, and so on with the mitigations to the portfolio, the auditor’s recommendation, etc and now we have a much better model.It does still need more to do to improve, and we can do this over time.

What would you like to achieve by attending the 7th Annual Advanced Credit Risk Modelling Under IFRS 9?

I look forward to give my insight on the challenges we have faced, especially on the data quality, the changes we needed to make right away on the data collection processes and working with the auditors to build a reliable model.It is important to me to hear about others experiences and discuss the challenges faced and future steps.

All of us working on this have faced a challenge, a different one of their own which has arisen from different areas, because of different purposes, etc and each has their own experience to share, their knowledge and expertise to give, a different approach that may work elsewhere too. That is why this conference is important, as what we can all get away with, will make us look and aim the exact next step to refine the model we use. This is an opportunity to discuss about the whole implementation process and the challenges faced until final use with the colleagues and acquiring knowledge applicable for further improvement of the IFRS 9 credit risk models.

At the beginning of 2018, banks went live with the credit risk models which are currently feeding IFRS 9 figures. By this point, the initial models and methodologies for calculating the expected credit loss on credit products and the guidelines for building this methodology are well understood. But while banks have built the models that IFRS 9 necessitated, the bulk of actionable insights regarding credit risk models under IFRS 9 are to be gained now. IFRS 9 models are endlessly running and supplying new information on risk and performance, which banks must be ready to take advantage of. This year banks are looking to optimise the data they are gaining from their models, stress tests, and audits in order to lend their models greater stability and robustness, thus better managing IFRS 9 volatility. 

With this in mind, this marcus evans conference will deep dive into how banks are advancing credit risk models under IFRS 9 by confronting self assessment through validation, audit and data insights.