Calibrating deposit models: Using historical data or forward-looking information?  

After a long period of negative policy rates within Europe, the past two years marked a period with multiple hikes of the overnight rate by central banks in Europe, such as the European Central Bank (ECB), in an effort to combat the high inflation levels in Europe. These increases led to tumult in the financial markets and caused banks to adjust the pricing of consumer products to reflect the new circumstances. These developments have given rise to a variety of challenges in modeling non-maturing deposits (NMDs). While accurate and robust models for non-maturing deposits are now more important than ever. These models generally consist of multiple building blocks, which together provide a full picture on the expected portfolio behavior. One of these building blocks is the calibration approach for parametrizing the relevant model elements, which is covered in this blog post. 

One of the main puzzles risk modelers currently face is the definition of the expected repricing profile of non-maturing deposits. This repricing profile is essential for proper risk management of the portfolio. Moreover, banks need to substantiate modeling choices and subsequent parametrization of the models to both internal and external validation and regulatory bodies. Traditionally, banks used historically observed relationships between behavioral deposit components and their drivers for the parametrization. Because of the significant change in market circumstances, historical data has lost (part of) its forecasting power. As an alternative, many banks are now considering the use of forward-looking scenario analysis instead of, or in addition to, historical data. 

The problem with using historical observations 

In many European markets, the degree to which customer deposit rates track market rates (repricing) has decreased over the last decade. Repricing first decreased because banks were hesitant to lower rates below zero. And currently we still observe slower repricing when compared to past rising interest cycles, since interest rate hikes were not directly reflected in deposit rates. Therefore, the long period of low and even negative interest rates creates a bias in the historical data available for calibration, making the information less representative. Especially since the historical data does not cover all parts of the economic cycle. On the other hand, the historical data still contains relevant information on client and pricing behavior, such that fully ignoring observed behavior also does not seem sensible.  

Therefore, to overcome these issues, Risk and ALM managers should analyze to what extent the historically repricing behavior is still representative for the coming years and whether it aligns with the banks’ current pricing strategy. Here, it could be beneficial for banks to challenge model forecasts by expectations following from economic rationale. Given the strategic relevance of the topic, and the impact of the portfolio on the total balance sheet, the bank’s senior management is typically highly involved in this process.  

Improving models through forward looking information 

Common sense and understanding deposit model dynamics are an integral part of the modeling process. Best practice deposit modeling includes forming a comprehensive set of possible (interest rate) scenarios for the future. To create a proper representation of all possible future market developments, both downward and upward scenarios should be included. The slope of the interest rate scenarios can be adjusted to reflect gradual changes over time, or sudden steepening or flattening of the curve. Pricing experts should be consulted to determine the expected deposit rate developments over time for each of the interest rate scenarios. Deposit model parameters should be chosen in such a way that its estimations on average provide the best fit for the scenario analysis. 

When going through this process in your organization, be aware that the effects of consulting pricing experts go both ways. Risk and ALM managers will improve deposit models by using forward-looking business opinions and the business’ understanding of the market will improve through model forecasts. 

Trying to define the most suitable calibration approach for your NMD model?

Would you like to know more about the challenges related to the calibration of NMD models based on historical data? Or would you like a comprehensive overview of the relevant considerations when applying forward-looking information in the calibration process?

Read our whitepaper on this topic: 'A comprehensive overview of deposit modelling concepts'