Back in the driving seat: How driver-based modelling can help restore profitability
Investors are starting to flex their muscles again in the Square Mile. It seems that an average Return on Equity (RoE) of 7% across Europe1 – below the typical cost of equity – is not enough to satisfy shareholders, as the calls to return to pre-crisis levels of profitability grow louder.
A prolonged period of low interest rates and increased capital requirements mean the days of RoEs over 20% are a distant memory. Nevertheless, banks are looking for ways to respond to investor pressure and improve returns. In this environment, understanding what drives returns, and the impact should they change, is key.
Consequently, many Finance, and Financial planning and analysis (FP&A) teams in particular, are reviewing their existing forecasting methodologies, with a renewed focus on driver-based modelling.
In my opinion, this is a welcome move, but it’s not without its challenges.
Data quality, quantity and accessibility create limitations when analysing historical correlations. Underlying historical data is often not readily available and, where it does exist, is not always relevant. For example, historical data trends may be influenced by a specific business strategy, or one-off industry event that’s unlikely to be repeated.
Another common issue is overconfidence in models’ results, often driven by a lack of transparency regarding model assumptions and limitations. This can be exacerbated by a pervasive belief that model complexity equates to reliability. In such environments, managers can become complacent, underestimate model risk, and may overlook the valuable role of expert judgment in analysis and decision making. In the worst-case scenarios, this can result in business decisions made solely on the basis of modelling outputs.
Technology proves another challenge to FP&A teams developing driver-based models. While ultimately advances in tools and technology should help improve modelling capability, decisions on what products to choose are understandably daunting. Although Excel sits firmly in the comfort zone, it comes with limitations that can affect the accuracy (and significantly increase the operational risk) of modelling, such as version control and human error, which can negatively impact both the results and their credibility.
However, despite these difficulties, banks can reap the benefits of a modelling capability by ensuring the following:
Embrace new technology: Banks must seize the opportunity to take on new tools for modelling, or risk being left behind in an archaic world of Excel spreadsheets. Completing a formal vendor assessment is a good way to see what’s on offer. The trend to move towards open source technologies such as Python is making building models easier and cheaper. Alongside this, looking at visualisation software, such as Tableau, will help Finance and business teams better interpret results. The buzz around machine learning stretches to financial modelling too – with careful application it can be used to reduce manual processes and judgement in financial modelling, however its full application is likely to be some way off.
Build a Centre of Excellence: A centralised modelling function helps consolidate the numerous decentralised modelling capabilities which many banks currently house. By bringing together modellers from across functions, a Centre of Excellence enables modellers in Finance to draw on the skills and techniques used elsewhere, for example from Risk teams, whose models and modelling skillsets tend to be comparatively more advanced. Additionally, a Centre of Excellence can improve volume, scale and standardisation benefits across the bank’s modelling suite.
Take control: More than ever, Finance managers need to be able to trust their models. Better control doesn’t necessarily mean having more layers of review, but instead getting smarter about awareness of model limitations and sensitivities. Setting up a model review function is a good way to control model risk, while business training on model usage can help awareness. Critically, banks need a well understood and simple governance framework with clear roles and responsibilities to control the process and use of the results.
Invest in data: The well-known rule ‘rubbish in, rubbish out’ applies just as equally to driver-based models as others. Banks should undergo a regular data clean up and, where desired data is unavailable or irrelevant, consider purchasing external data. Storing data in a way that is easily available, logical and grouped based on desired reporting categories addresses a host of issues currently faced by modellers. This can be done through housing data in a central repository such as a data warehouse or data lake.
Recruit, train and motivate: It’s hard to find people who not only have the right modelling skill set, but also have Finance experience – they are scarce and in high demand. This is exactly where recruitment needs to focus attention and organisations will reap rewards if they invest in the right people now. When recruiting, consider this: the right people are more likely to have started with a science or quantitative background and learned Finance, rather than the other way around. In addition, clear career paths for modellers should be defined and training opportunities offered to ensure employees stay motivated.
Taking these steps and implementing a modelling capability that provides an objective view of how changing macroeconomic and operational drivers impacts performance can yield several benefits:
- Better engagement between Finance and the business on how to react to the expected economic headwinds and adapt future strategy
- Better identification of risks resulting from model drivers leading to more informed risk appetite setting
- Increased accuracy of projected numbers within planning, budgeting and forecasting processes
- More informed and well-evidenced regulatory submissions (e.g. regulatory stress tests)
- Development of technology and analytical skills critical to Finance’s future success
Perhaps most importantly, gaining a better understanding of how drivers will affect banks’ financial performance will lead to better management decisions on the road to improved returns. If banks are to return their shareholders’ value on their investment, driver-based modelling will be key.