As we discussed in our first two blogs, International Financial Reporting Standard (IFRS) 9 was put in place as a result of the financial crisis of 2007-2008 to better understand an organization’s expected credit losses (ECLs). To help insurance companies take appropriate steps to comply with the new requirements, our first blog looked at the challenge of calculating impairment, while our second blog discussed the need for methodologically robust tools for assessing ECLs. We now focus on the importance of efficiency and transparency.

The Need for Efficient and Transparent Analytical Tools

As insurance companies get ready to handle IFRS 9, they should consider analytical tools that are robust, efficient, and transparent. ECLs are typically generated by the estimation of three elements: (1) Probabilities of Default (PD) – a view on the probability that a bond issuer will not pay when due, (2) Loss Given Default (LGD) – a view on post-default recovery prospects, and (3) Exposure at Default (EAD) – the total value claimed by the holder of the bond against the issuer. Balancing the need for efficiency as well as transparency is often a struggle for modelling practitioners. Our approach has been designed to generate risk parameters in an efficient manner, whilst maintaining transparency.

Efficiency is Paramount

The investment books of insurance companies often contain tens of thousands of bonds. Given the need to run risk assessments at each financial reporting date (e.g., quarterly), having an efficient process is critical. Our approach was designed to enable the use of automated pre-filled inputs, manual user-filled inputs, or a mix of the two. Users value these options, often using pre-filled inputs for those countries, sectors, or bond issuers where exposure is relatively low and user-filled internal inputs where exposures are relatively high.

As discussed in our second blog, our methodology utilises the public rating of a bond issuer before assigning a long-term or through-the-cycle (TTC) PD. This TTC PD is subsequently adjusted to a point-in-time (PIT) view via our Credit Cycle Projection Overlay using macroeconomic and market data, which are country and sector-specific. (See Figure 1 below.) The six inputs used in the Overlay are pre-populated from three distinct sources for over 115 countries,¹ including forecasts from S&P Global Ratings. Alternatively, the Overlay can also “ingest” internal forecasts for any of the required six factors for the countries of choice. The user simply specifies which countries and sectors are to be analyed using internal forecasts and enters the data when prompted.

Figure 1: Credit Cycle Projection Overlay

^{S&P Global Market Intelligence. For illustrative purposes only.}

The output of the tool is a single, consistent table displaying PIT PD term structures for all countries selected, segmented by sector and the rating of the issuer. This enables the output table of PDs to be used in whichever manner best fits the user’s needs, as highlighted in Figure 2 below.

Figure 2: Sample Output of the Credit Cycle Projection Overlay

^{S&P Global Market Intelligence. For illustrative purposes only.}

The LGD risk parameter is often over simplified. Our approach employs multiple fundamental and sectoral inputs in order to derive meaningful LGD estimates. In most cases,  these inputs are sourced from S&P Global Market Intelligence’s fundamental databases that provide deep and broad information for over 1 million public and private companies globally [1]”.² This enables the automatic generation of LGD term structures by simply using the bond identifier (e.g., ISIN). The LGD term structures are bond, sector, and country specific, as shown in Figure 3 below. This automation is currently available for key sectors, including corporates, banks, insurers, and sovereigns.

Figure 3: Calculating LGD Risk Parameters

^{S&P Global Market Intelligence. For illustrative purposes only.}

Transparency is a Necessity Too

The need for transparency is often overlooked, or given less importance than efficiency and robustness. This can cause problems post implementation when auditors and internal/external stakeholders ask specific questions about the underlying models. It can be difficult to defend ECL outputs if a “black box” is used, as there is a direct correlation between the transparency of a model and the ease to which it can be audited and defended.

Our approach (including models, documentation, and training) is considered to be a “glass-box”, as users can view all sections and inputs of the models, and pre-filled data is fully referenced from public sources. This enables insurers to easily report ECL with confidence and explain changes from previous reporting periods to non-technical parties in a straightforward fashion.

To learn more about our robust, efficient, and transparent IFRS 9 offering, contact us here

¹ As of November 2019.

² [1] Data as of February 2020