John Harbor considers how the goals of transparency and efficiency can be achieved via the effective implementation of business intelligence and stochastic modelling solutions
In September 2004, at an Xchanging seminar, Dane Douetil, CEO of Brit Insurance and Brit Syndicates and Chairman of the LMA Market Processes Committee, delivered a speech entitled 'Leading Change'. On the topic of 'Why do we need market reform?' he proposed that it was "not just about cost savings, nor data integrity. It was about quality, real-time information at the underwriting coalface leading to better decisions and also about increasing the accuracy and speed of claim reserving leading to a less volatile market."
Similarly, data availability, quality and transparency have been highlighted by regulators as critical risk factors when attempting to evaluate business metrics. This has more recently been reiterated by John Tiner in his speech to the LeBoeuf, Lamb, Greene & MacRae Symposium in New York, where he stated that "experience needs quality data such as on claims, underwriting experience and also requires certainty in policy wordings."
The topic of contract certainty is not to be covered here. Instead, I plan to illustrate how the aims of transparency and efficiency within business operations can be achieved, by all businesses operating within the financial services environment.
One of the largest areas of improvement provided by IT developments is in the use of data analysis and organisational decision-making tools within the business infrastructure. In many organisations the existing process of physically analysing information and making decisions based on MIS-produced static reports remains an inefficient resource drain. The alternative is to give stakeholders and business managers controlled, interactive access to the information they need to make the best decisions.
It is important to note that, in this context, reporting is not the same as analysis. A hard coded report from a management information system is static; it is often not possible to change the report structure or analysis type without employing the use of other IT-focussed resources within the organisation. Similarly, it is not immediately possible to drill through the reported data to a more detailed level because you have spotted an anomaly and wish to examine the reason. The ability to sort the 'information' from data using decision-supporting analysis tools can give an immeasurable competitive edge, critical in today's global markets.
Interactive information analysis in the form of business intelligence and stochastic modelling solutions can offer decision makers and analysts a centralised approach to global risk management. Many of these solutions originate from the retail sector where they have been used for decades to interpret individual buying behaviours and the effect that altered pricing, location, and even seasons can have on the individual. They are now being used increasingly within the financial services sector, designed to offer underwriters and brokers a centralised approach to risk pricing and management.
Business intelligence and stochastic modelling provides the means to flexibly, intuitively and instantly view, dissect and manipulate information to aid in the decision-making process. They offer users the ability to analyse data in a flexible way, to search for trends and patterns, analyse productivity and to aid business planning and forecasting. The technology significantly reduces the time required for decision makers to understand key metrics driving their business.
London has long been a centre for placing and managing speciality risks, but it is also becoming a centre for a very international marketplace.
Almost all of the major insurers, reinsurers and brokers have a major presence in London, and many use this as a hub to conduct business on a global scale.
A by-product of this globalisation trend has created disparate operating units which can then find it difficult to centrally manage their business data and processes. At the crux of this issue is the fact that each business unit often relies on separate IT systems storing disparate business data.
This inevitably leads to increased operational and financial risk. To combat exactly these types of business risks, technology has become more centralised, making it possible to create central data warehouses capable of storing huge quantities of information fed from various international sources. This provides a central repository for the collation of policy and asset information for each division or office.
Business intelligence solutions
Once a data store such as this exists, and is comprehensively populated, business analysts are able to examine and analyse this data in a far more structured and complete manner. This is where business intelligence solutions come to the fore. Sophisticated front-end solutions provide the means to flexibly, intuitively and instantly view, dissect and manipulate this information to aid in the decision-making process. For example, where each operating division of a large multi-national once had access to their own transactional database feeding standard reporting suites capable of creating any number of pre-configured reports, now each can have access to a co-ordinated central database. This is optimised for reporting and analytical purposes, using a solution that turns standard reporting into living business models that can be manipulated by the business users, and can provide information on key performance indicators such as exposure and utilisation.
An integral part of the business planning process is also to accurately project and understand claims reserving requirements. The analysis of development and deterioration of historic claims, from both an actuarial and statistical perspective, can offer an insight into the likely run-off of claims.
The actuarial profession in the non-life sector was little known even as recently as 20 years ago. Gradually the growth of statistical and actuarial techniques has become more readily accepted and today they are generally considered to be the norm. There can be no doubt that the accuracy of claims reserving has significantly increased over this period, with the majority of insurers and reinsurers using established methodologies or variants thereof. In general, statistical and actuarial techniques are used to calculate the likely required reserve for any class of business, or other grouping, and to assess the confidence the business can have in the sufficiency of any reserve set.
Clearly, however sound the methodology, the outcome will still be dependent on the quality of underlying data and the soundness of the assumptions that have been made. An advantage of the current practice of modelling claims on several different bases is the ability to project a range of outcomes. The problem is where to select from within the range of projected results and additionally the lack of transparency, ie, generally speaking, third party users of this information are unlikely to know whether an optimistic or a pessimistic view has been taken on future projections.
More recently, increasing focus has been given to stochastic modelling.
In essence this requires a model to be developed with assumptions made in respect of all uncertainties within that model. By the use of statistical distributions, not only can a range of outcomes be produced but factors can be applied as to the probability of the likely outcome. In other words, management can determine the level of confidence they wish to use to set their loss reserves, and also third parties will become aware of the policy of management and to where, within a range, ultimate loss reserves have been selected.
Not surprisingly this approach has attracted the interest of insurance regulators and the International Association of Regulators has endorsed the stochastic approach. For example, the Australian insurance regulator has already embedded the use of stochastic techniques within its regulation and by way of general guidance has suggested that companies should take the expected outcome and apply a 25 per cent confidence loading on top.
In other words, for Australian loss reserves, companies will be using the 75th percentile of confidence.
Loss reserving is clearly crucial to any insurer. However, loss reserves used in isolation have only limited value. The way in which loss reserving is broken down, the way in which companies can drill down and analyse what lies within their loss reserves, is not only crucial to the well-being of an insurer but should also drive the underwriting and, therefore, the rating process. For this reason it is highly desirable that the information systems used by insurers are able to dynamically feed the forecasting models, but of equal importance is that the outcomes from those models are fed back into those management information systems so that they can be analysed in a flexible way.
An integrated approach to systems and the use of data from the systems is key to ensuring that accurate and timely information is produced. Most importantly, the loss reserves are constantly available as the drivers for writing new business. In summary, therefore, as companies increasingly adopt stochastic modelling this will result in:
- greater insight into the reserving and projection process;
- allowing supported justification of setting reserves at any given level;
- projections returned dynamically to management information systems;
- ability to display different projection bases and business plan projections against each other for comparison; and
- satisfaction of regulators and auditors that the process for setting loss reserves is transparent.
It is clear from current practice and the approach of regulators to monitoring a business' systems and controls, that responsibility for delivering greater transparency and data quality does not fall to central bodies, but lands directly in the board room. The main facilitator for incorporating best practice data management strategies into the business process is the efficient use of analytical solutions. This leads to better decision-making and optimum financial management; the issues of data integrity and quality, and ultimately cost savings will naturally follow.
In short, business intelligence combined with stochastic modelling provides business managers with fast and easy access to analytical reporting, data visualisation and business forecasting, ensuring more informed business decision-making and risk mitigation.
John Harbor is a senior insurance partner at Moore Stephens in London.