IRR measurement is a complicated and challenging process.  Our last article discussed types of IRR measurement processes, and this installment will delve into measurement processes, challenges, and best practices.

Successful IRR measurement requires that those responsible for the process understand the IRR models’ calculation methods and assumptions. One common source of IRR measurement inaccuracies is failing to update the IRR models to reflect changes in the bank’s business model due to changes in business strategies, products and services, mergers, or acquisitions.

Data Gathering and Aggregation

Accurate IRR measurement depends on the data inputs for the models being accurate, complete, and current. IRR measurement data would typically include detailed information regarding:

  • Balances and interest rates of fixed income instruments and portfolios
  • Fixed-rate loans, including interest rate levels, maturity dates, prepayment characteristics and loan type.
  • Variable-rate loans, including: rate indexes, reset periods, margin levels, caps and floors, and prepayment characteristics.
  • Material off-balance sheet positions.
  • Rate sensitive noninterest sources of income.

A complete assessment of IRR exposure will often require detailed information regarding factors that can influence loan prepayment speed, such as rate history, origination dates, and geographic location of the loans.

It is not unusual for banks to have to source this data from multiple systems that support commercial and consumer lending, investments, and deposits. This is an area where effective data management practices are critical to the integrity of the data lineage and accuracy, and the bank’s ability to retrieve it in a timely manner.

The data for similar instruments or portfolios with similar risk characteristics are often aggregated and stratified before being input into the IRR models to increase efficiency.  For example, adjustable-rate mortgages may be stratified by rate index, cap levels, and reset frequency. Complex instruments, such as structured investments are typically input individually.

Developing Stress Scenarios

Regulators expect a bank’s stress scenarios to cover a meaningful range of outcomes that will sufficiently identify all risk types that would apply to its particular business model.  In most cases that would include repricing, basis, yield-curve, and options risk. Interest rate change scenarios would not just reflect parallel changes in rates along the yield-curve, but also changes to the shape and slope of the curve. The amount of the change should be sufficient to cover severe, yet plausible changes relative to current levels of interest rates and ranges from 100 to 400 basis points are typical. Depending on the bank’s risk profile, stress scenarios should include:

  • Instantaneous and severe rate shocks.
  • Short-term rate changes.
  • Long-term rate changes.
  • Prolonged periods of stable rates.
  • Relative changes between key market rate indexes.
  • Changes to the shape of the yield-curve (including steepening, flattening, and inversion).
  • Negative levels of interest rates.

Interest rate scenarios are typically developed using one of the following methods:

  • Deterministic Approach: An approach that establishes standard scenarios for the timing and level of interest rate changes based on estimates of the likelihood of the rate changes for risk analysis. These scenarios are often periodically supplemented by stress scenarios.
  • Stochastic approach: This approach generates exposure estimates for multiple random rate scenarios and calculates an expected value from the distribution of estimates.

Developing IRR Model Assumptions

IRR models use assumptions about how an instrument’s maturity and repricing characteristics will vary under different market environments. These assumptions should be reasonable and consistent with the bank’s experience and likely customer behavior. Management is expected to document, monitor, and periodically update these assumptions.

The documentation should provide an understanding of how the model was built, validate the assumptions, facilitate the periodic review of assumptions and provide a rationale for how they were derived.

Typical sources of assumptions include historical trend analysis, internally or third party developed prepayment models, independent third-party estimates, and input based on the experience of appropriate business units.

NMD Deposit Assumptions

Because deposits that have no specified maturity (NMDs) represent a major portion of most banks’ source of funding, assumptions used for modeling their IRR are particularly critical. The key assumptions include deposit price sensitivity (the expected change in the deposit rate relative to market rates) and decay, or runoff, rates (level of deposit withdrawals over a given time period). NMD assumptions are typically based on analysis performed by the line of business supported by behavioral data and pricing methodologies, historical trend analysis of deposit data, and/or industry analysis of data from multiple firms.

Banks should use different assumptions for stable and nonstable deposits and recognize that rate-sensitive and high-cost deposit sources (such as brokered deposits) will have higher decay rates than deposits from other sources.

Prepayment Assumptions

Prepayment assumptions, particularly those related to mortgage loans and securities, are also particularly critical to IRR modeling as they significantly affect their expected cash flows. Understanding the price sensitivity (duration) and its impact on the earnings and price volatility of mortgage loans and securities is a critical component of IRR measurement.  Prepayment speeds change in different rate environments- for example, prepayments generally increase in a low-rate environment as borrowers refinance to lock in lower rates. However, prepayments do not always follow predictable patterns and can be quite erratic.

Assumption Governance

Bank management should evaluate key IRR assumptions for reasonableness at least annually and more frequently when market conditions or the competitive environment are changing rapidly, or there are material changes to the bank’s business model.

The review will generally start with a sensitivity analysis to determine the bank’s exposure under different sets of assumptions. This will help management to identify the most critical assumptions and determine the conditions under which they cause the model parameters to break down.

Computing Risk Levels

Calculating Risk to Earnings

Net Interest Income (NII) is estimated by deriving projected average rates using the bank’s assumptions for future interest rates and multiplying that by projected average balances based on the repricing, maturity, and growth rate assumptions of existing positions. Interest sensitive fee income such as mortgage servicing fees and closing points are usually added to NII.

Marked-to-market gains or losses from trading and dealer positions are often performed separately. All expected future cash flows are then discounted back to present value and Net Present Values are calculated using different rate scenarios.

Calculating Risk to Capital

The appropriate method for calculating a bank’s economic exposure depends on the complexity and maturity schedule of its assets, liabilities, and off-balance-sheet instruments.

Banks with more complex structures will typically employ stochastic models, while banks with less complex structures may use deterministic models.

Banks will typically quantify their economic value exposure using models that perform a series of present value calculations that discount the cash flows derived from their current positions and assumptions for a specified interest rate scenario.

Banks with exposure to options and embedded options will usually employ stochastic models that capture the effect of the value of options increasing as rates approach the strike price and the probability of the option going “in-the-money” (see Part 2 of this series for a discussion on options risk).


Interest rate risk measurement is a complex and technically challenging subject that is critical to the financial stability of every bank and something that senior management and the board need to understand and manage accordingly.  Bank management needs to have a thorough understanding IRR models, the assumptions that they use and how IRR is calculated.  Future articles will take a closer look at IRR monitoring and control systems and model risk management.

Doran Jones can provide the regulatory compliance and risk management expertise combined with extensive technological knowledge and experience to design and implement a cost-effective solution that will increase efficiency and lower risk by upgrading your risk and compliance systems or identifying and remediating gaps in existing processes.

Contact us to learn how a strategic partnership with Doran Jones can provide you with cost-effective solutions by leveraging our expertise with these and other critical risk and compliance functions.