Portfolio management is a critical aspect of a reinsurer’s performance. Fundamentally, a sound portfolio management strategy has to maximize premium versus loss, within a given capacity.

While the risk-return approach is very well documented and applies to the pricing of a single policy, optimisation of premium versus capacity is still relatively new, even in specialty reinsurance, where the capacity limit is often the main constraint to writing more premium. The capacity limit is a value set by management. It is driven by compliance constraints on capital and by business priorities. It sets a limit that exposure to certain events or risks cannot exceed.

This article aims to quantify the opportunity of adopting a proactive portfolio management strategy, relying on making the best use of data analytics. Such an approach enables risk professionals to select policies with the highest marginal impact, in terms of premium, while keeping exposure within a predefined capacity.

While this analysis focuses on Energy reinsurance, we believe the same approach is suitable and relevant for other business lines across the specialty risks’ space, where accumulation needs to be cautiously managed.

Quantifying the Impact of Optimised Exposure Management

This study analyzes the impact of three potential levers offered by data analytics and technology, enabling risk professionals to improve the total premium of their reinsurance book within a given capacity constraint. This study does not aim to optimize a portfolio, given its fundamental risk quality or its probabilities to potential loss events, although these parameters can also be added.

The three identified levers are the following:

1. More certainty: decrease the capacity buffer by improving the confidence in risk accumulation figures.
2. Portfolio optimization: select better policies by benchmarking them in terms of premium versus maximum exposures.
3. Others: identify trapped exposure and unlock untapped capacity by understanding the class-specific methodologies used by cedant to display and transfer their risks.

Based on this study, the combined impact of these three levers on a typical energy reinsurance book can add up to 17% of the net premium for a given referenced reinsurance portfolio, with equal Capacity.

Firstly, the article defines the parameters used for the analysis that supports these numbers. Secondly, it establishes a reference case (where no lever is applied), and thirdly, it quantifies the marginal impact of each lever sequentially.

Optimisation Parameters

In the Energy class of risk, (re)insurers ensure that their exposure to any risk (i.e. assets such as platforms, wind farms, etc.) remains within the capacity set for their class.  The exposure to each risk then needs to be calculated, with the top one, the “max exposure”, being compared with the given capacity. Ideally, it should be as close as possible to the capacity without exceeding it. (Re)insurers can play with their shares in each policy in order to modify their exposures and premium.

Multiple challenges currently prevent risk professionals from qualifying, quantifying and optimising their book’s exposures. The root cause of these challenges is limited and unstructured risk submission data.

Digitalization platforms powered by machine learning and artificial intelligence, such as Allphins, offer tangible solutions, enabling underwriters to access details, complete and transparent views of their exposure in real-time for pre-placement analysis. Using such tools allow underwriters to perform the following:

1. Recognise every single risk within any schedule of value
2. Calculate the exposure to any event for any a given portfolio and for any transfer structure – such as the “installation total loss” scenarios, the highest of which is the max exposure
3. Generate customizable probable maximum loss scenarios
4. Analyse the premium and exposure impact of any underwriting decision

This analysis looks at how technologies and data allow us to optimize premium while maintaining the max exposure within capacity while quantifying the premium impact.

Reference Case

In this reference case, we look at what the industry can do without the use of data analytical tools and models.

We built a fictional representative book, which effectively works as our control group. The underlying assumptions used to construct this book are based on our experience of the market and the many interviews we conducted with market professionals. This book of policies has an asset type distribution, maximum exposure limits, a regional split,  and transfer structures (combination of excesses of losses layers at different levels and of quota shares) that are representative of the market.

The use of spreadsheets with limited and disorganised data is widespread in the industry. Furthermore, exposure figures derived from spreadsheet-based analysis are prone to errors and approximation. To account for this uncertainty, exposure managers typically add 10 to 20% on top of their aggregate exposure figures, which translates into an additional capital conservative buffer. For the purpose of this analysis, we assume a reference case where a 15% buffer is used.

Note that the “share” is the percentage of a policy that the reinsurer underwrites. For this reference case, we assume the share in each policy layer is 5%. Later we will see how these shares evolve to optimize the book.

The table below describes the features of our representative book, as well as the hypothesis underlying the reference case:

Lever 1: Gaining More Certainty in Exposures

The first opportunity to generate additional premium is to gain more certainty and clarity about the exposures and thus to reduce the capital conservative buffer. This allows reinsurers to write more risks while maintaining the same capacity.

One of the biggest challenges to achieving this is to be able to trust exposure and risk accumulation figures. We strongly advocate for a fact-based approach that analyzes every risk within a reinsurance submission data pack or schedule. In the absence of such analysis, risk professionals are forced to take an overly conservative buffer.

For example, name recognition is a significant challenge for the industry and matching the correct asset-risk pair is a complex task. The same asset can be addressed with multiple names. Using Allphins, underwriters can apply a risk identification algorithm, which recognises each asset using NLP (Natural Language Processing) and guarantees that no risk is missed in the aggregations.

Furthermore, the Allphins platform enables reinsurers to interrogate their book, apply probable maximum loss (PML), or perform sensitivity analysis.

Overall, this allows users to have a very strong certainty of all their exposures, in particular the top ones, based on an exhaustive analysis and auditable calculations. The need for a buffer is therefore drastically reduced.

For this study, we claim that by using Allphins, (re)insurance professionals can reduce the capital conservative buffer by 5% to 10%. The benefits from this reduction can be translated into an increase in share across the book.

The graphs below show that while the capacity remained unchanged, the reinsurer has been able to increase their shares in the layers. The increase in average share from 5% to 5.23% translates into a 4.55% increase in total net premium.

Lever 2: Portfolio Optimization

Similar to other financial structures, combining multiple policies creates opportunities for portfolio optimization, notably in terms of risk return.

To achieve this, the reinsurer needs to be able to assess the marginal impact of their policies in terms of total premium and top exposure, which can be done with Allphins. Users can overweight policies that generate the most premium for the least incremental exposure to the top assets.

Optimizing a portfolio involves defining one or multiple objectives within a set of constraints. For this analysis, the set objective is to maximize the book total premium under the following two constraints: (1) absolute change in any share in policy is limited to 1% (to account for commercial limitations) and (2) top aggregate exposure remains unchanged.

The analysis shows an additional increase of 5% in total net premium without changing the max exposure. The graphs below show that the exposure to some assets has increased (Asset 11) while the exposure to some top assets has decreased (Asset 2).

Lever 3: Others – Opportunities Hidden by Complexity

Each class of risk has particular characteristics that need to be addressed and taken into account: typical aggregations, how policy wording affects these, how the data is transferred etc. Each specificity needs to be understood to ensure the exposure is properly calculated. In Energy, one such particularity is the potential for double-counting of risks in a book.

For example, we shall assume two separate risks, “Sleipner East Complex” and “Sleipner A”, each with their respective exposures. A complex refers to a group of individual platforms that are linked by a bridge. Sleipner A is an individual platform that is physically part of Sleipner East Complex. Therefore, a separate entry line for Sleipner A can be redundant with the Sleipner East Complex line, and in certain cases, summing the two exposures is double-counting.

If one fails to distinguish the characteristics of Sleipner A and Sleipner East Complex, and understand the potential double-counting, the exposure values may be overestimated and may excessively consume the reinsurer’s capacity. Traditionally, identifying such double entry has been a significant challenge for the industry. Using Allphins, underwriters can apply a double-counting algorithm that enables clients to mute such risks instantly across the book. Other trapped exposures exist in energy related with the head of covers and which are not treated here.

After applying a double-counting algorithm on the Optimised Portfolio, the max exposure falls by 6% (see “double counting 1” on the graph below). To offset this decrease, we increase the shares across the book from 5.4% to 6.2% to meet the previous max exposure value (see “double counting 2” on the graph below). This enables the reinsurer to further improve the profitability of their portfolio.

Conclusion

This article demonstrates that using a data analytics model, such as Allphins, can help (re)insurers materially improve the profitability of their portfolio. Such a model cannot replace a professional’s judgment and experience, but instead, data and technologies can enable informed and effective decision-making.

Our analysis does not account for the fundamental risk quality per unit of exposure, and does not apply frequencies or probabilities to potential loss events. However, we suggest a simple framework to analyze the opportunities offered by a fact-based portfolio management approach.

This analysis has been applied to energy reinsurers, where capacity management is a pressing challenge. However, more and more insurers in other classes of risks, manage their books with “capacity” set per event, asset type, geography, sub-classes or industrial sector. This natural evolution of underwriting requires tools that provide class-specific data and features, and that can manage premiums and exposures in parallel, such as Allphins.

Mihir Rai, Data Business Analyst & the Allphins team

For more info: contact@allphins.com

‍