Blog

Internal Model Validation

top banner cropped 2

With 2017 well underway, insurers are launching the next phase of model development and are starting to prepare for validation season.

At Dynamo Analytics we believe that a well-designed validation process will enable insurers to embed their models in their critical business decisions, as well as satisfying the regulatory requirements. It should be a process targeted at helping your Board and senior management understand and manage model risk and model quality. In order to do this we designed our Model Certification framework, which provides a consistent, robust and proportionate review process across all actuarial models. Over the past four years we have used this framework to not only help insurers demonstrate the suitability of their internal models relative to Solvency II requirements, but also to help them make smarter business decisions.

We offer:

INDEPENDENCE Our independence from the model allows us to more easily and openly identify model limitations and make recommendations.

INSIGHT Our fresh perspective on your model will bring new insight and ideas to help fuel future model development.

COST-EFFECTIVE SOLUTION Model validation is typically cyclical. Using external resources can be more cost-effective and free up internal resources for business critical roles.

EXPERTISE Our team has a diverse range of actuarial and modelling experience. We can provide subject matter expertise and suggest new techniques and improvements.

TECHNOLOGY We have developed our own technical modelling platform, Psicle®, in which we build robust, repeatable, auditable validation tools and side models.

With most London Market insurers now in their third or fourth validation cycle, capital model validation is ready to advance – it’s what athletes call moving from the “associative” phase to the “autonomous” phase. Advancing model validation means:

  • Answering those niggling questions that are concerning the Board;
  • Staying ahead of the curve with the latest actuarial techniques and emerging risk approaches; and
  • Automating validation processes to allow resources to focus on the results.

graph
MODEL CERTIFICATION FRAMEWORK
Our Model Certification framework consists of five high level, overarching statements. Each of these statements are supported by sub-statements, which give more granularity on aspects of the model to be assessed. With the help of a suite of validation tools, we answer each of the validation statements to help us, and you, gain comfort that the model is operating as expected, and is fit-for-purpose.

WORKING WITH US
Dynamo often bases validation team members at our client’s office for periods of time over the validation cycle. This makes the challenge and feedback straight forward and efficient. We can conduct model runs on-site or remotely. We have a strong team of highly motivated consultants and associates across the UK, South Africa and the Nordics, enabling us to be more cost effective than our competitors.

We believe there are five stages to model validation, so whether you want to start with deciding on this year’s scope, or with designing your annual training programme, we can help.

long image

FLEXIBILITY
We offer a flexible suite of validation services:

  • Ranging from high level reviews to full technical validation;
  • Our services can include model runs to take the pressure off you;
  • We can also use our software Psicle® to rapidly design and build tools that suit your model;
  • Our final deliverable can range from high level feedback to a full validation report, which expresses an independent opinion on the SCR.

TIMELINE
Over the next few months, insurers will start to plan the 2017 validation exercise, ensuring it fits effectively within business planning, model development and regulatory timeframes. Whatever your validation requirements, we’d be delighted to tailor our services to suit you, so please get in touch to find out more about our 2017 validation offering.

Stretching the bounds of Actuarial Science: PART 1 – Anesu Shuro

Traditionally, actuarial skills have been used in insurance for the purpose of valuations, capital modelling and pricing. The kind of statistical skills that actuaries learn in training can; however, be used very efficiently in a lot of fields. One of these skills being is that of predictive modelling using tools like multivariate analysis. These skills, your generic statistician would possess which would make one wonder why someone would pay expensive actuarial fees instead of using data analysts and other statisticians that might prove cheaper.

One thing that makes actuaries more suited to answering some of the questions that companies face is their ability to structure the way they approach problems. One of the many tools Actuaries learn in training and exams is that of the actuarial control cycle (ACC). The ACC is sets out a framework for solving problems including understanding the problem; how to choose an appropriate technique to solving the problem, how to monitor and update assumption and consideration of external factors i.e. the environment and professional issues. This and other tools ensure a structure is maintained in solving problems and creates better clarity for clients and reviewers on the work done.

Actuaries are also trained in communicating uncertainty and highly technical concepts to a non-technical audience. This ensures that companies understand the results, limitations and implications of the results. Actuaries also have a board that ensures structured reporting, communication and ethics are practiced during actuarial work

Here I have highlighted some of the applications of predictive modelling including some that Dynamo has been involved in.

Banking

Modelling can be done around customer relationship management (CRM) in order to optimise marketing and operations. This included projecting customer receptiveness to different market offers; propensity to repay debt; identifying opportunities of cross-selling and up-selling and retention offerings to deliver to specific customers at given times; churn and risk management; and, using decision methods like declarative rules and decision tables.

Telecommunication

Modelling can be done on usage data, network performance and device sales.  Again predictive modelling in churn management can be used.

Healthcare

There is extensive data that can be used in healthcare which can be used in predictive modelling in order to reduce costs and provide better healthcare. This includes claims and costs data, pharmaceutical and R&D, clinical data (from medical reports) and patient behaviour and sentiment data from stores. This data can be used in clinical analysis, financial analysis, supply chain analysis and fraud and HR analysis

General

In any subscription market CRM is a big area especially if companies have “big data” that is of good quality, Churn is one of the biggest areas of modelling in subscription markets.

These are just but a few of the applications of predictive modelling and predictive modelling is just one of the many applications of actuarial skills. In the next parts of my blog I will highlight some of the wider fields that actuaries are involved in and that we at Dynamo are quite passionate about.

Protecting Sensitive Data – Human De Jager

The world as we know it has become increasingly digital.  Daily life has been streamlined by little bright screens and mundane tasks have become (mostly) quick and effortless.  Banks are moving away from physical cash and don’t even want their patrons to necessarily go into the bank, their fee structures are even set out in such a way that it’s cheaper to go online than to go to your nearest branch.  I for one am more than happy to stay at home and do my financial chores and admin rather than to go wait in a queue.  These little luxuries do not however come without its problems.

Most of these services require your personal information and you therefore have to trust a company’s capability to protect it.  Yes, a company may have an excellent track record and the best staff to keep the bad guys out of your personal matters, but they do sometimes get it wrong.  Apple’s cloud was hacked 2 years ago and various celebrities had very compromising photos thrown across the internet for everyone to see.  Ouch.  What is worrying is that Apple is one of the leader’s in the digital field and I am sure that this will not be the last time we see one of the big companies have similar blunders.

It therefore may be wise to think twice with who you are entrusting your digital life with.  The best way may be to go analogue (dis-enable internet connectivity) or, if possible, not to go digital at all.  After all, something can’t be taken from you that does not exist in the first place.