ICRFS-Plus™ Demonstration videos and powerpoint slide show
ICRFS-Plus™ is a tour de force of interactive software design and computational speed.
An ICRFS-Plus™ corporate database, which is not difficult to create, enables complete executive oversight. This means that that you will be able to find, with just a few mouse clicks, models and reports for any segment of your business in any country, the actuary modelling that segment of the business, capital allocation by LOB and calendar year, reserve risk charge and underwriting risk charge for the aggregate of LOBs, whether outward reinsurance is effective in respect of reducing retained risk, and more. Creation of an ICRFS-Plus™ database from triangles stored elsewhere or unit record transactional data is also seamless and effortless using COM scripts.
One great benefit of ICRFS-Plus™ is that you can manage and measure all your long tail liability risks with a single composite model. Only one model for each company!
Click here to view an powerpoint presentation overview (about 14Mb) of ICRFS-Plus™
A single composite model measures the reserve, underwriting and combined risks for each LOB and the aggregate.
One double click loads the model and reveals pictorially the volatility structure of each long tail LOB in your company and their inter-relationships (correlation structures). All the critical financial information such as risk capital allocation by LOB and calendar year, and Tail Value-at-Risk for different time horizons can be computed in a matter of seconds. A company-wide report can be created effortlessly with a single report template.
In respect of Solvency II Capital Requirements (SCR), Market Value Margins (Risk Margins) and Technical Provisions (Fair Value of Liabilities), for the aggregate of multiple LOBs, video chapter 5 provides Insureware's solution to the one year risk horizon.
View the videos below to experience the numerous unique benefits and applications afforded by a unique paradigm shift.
Some of the (real) case studies modelled in the videos are also discussed briefly in the ICRFS-Plus™ brochure.
These videos are arranged in logical order so it is important that you view them that way.
If for any reason you are unable to view the training or demonstration videos, please contact our support staff at firstname.lastname@example.org and we will arrange to send you a copy of the videos on CD-ROM. You will be able to run the videos from the CD.
- 2.1 Triangle Group CTP- The model building process
- 2.2 Triangle Group ABC-Major shifts in calendar year trends and testing for model specification error
- 2.3 Triangle Group COMPA-volatile paid losses with stable calendar year trends
- 2.4 WC California, Reserve Upgrade Myths and Case Reserve Estimates (CRE) versus Paid Losses
- 2.5 Triangle Group LR High - Mack Method gives results much too high!
- 2.6 Three layers 1Mxs1M, 2M and 1M, pricing and aggregate distributions
- 2.7 Uneven sampling periods and updating the database
- 2.8 How can we tell that data modeled by Murphy et al is not real?
2. Applications of the PTF and ELRF modelling frameworks
2.1 Triangle Group CTP-The model building process
A model is identified (built or designed) in the PTF modelling framework for the real data CTP, first manually, in order to illustrate some intuitive statistical concepts. Validation analyses are used to confirm the stability of trends, and to extract additional information from the data. The warp speed modelling wizard identifies the same model in less than one second, which was designed manually in a few minutes. The cumulative data are modelled in ELRF and as a result of condition 1 being satisfied, link ratios have no predictive power whatsoever.
2.2 Triangle Group ABC-Major shifts in calendar year trends and testing for model specification error
The triangle group ABC has major shifts in calendar year trends and very little process variability. The PTF model is identified by the warp speed wizard. Analyses of the cumulative data are also conducted in ELRF that includes the Mack Method. These methods do not capture the calendar year trend shifts as shown in the residual graphs. In ELRF modelling framework calendar year trends cannot be estimated or controlled for going forward (in any prediction)..
In the PTF modelling framework we identify (build) a model that replicates the volatility in the data. This way, model specification error is removed (almost) completely. A further test can be conducted to ensure absence of model specification error.Three triangles are simulated from the identified PTF model and it is shown that it is impossible to distinguish in respect of variability between the real triangle and the simulated triangles. QED.
Also see demonstration videos 9.1-9.5 to see how the bootstrap can be used to show that the Mack method can give false indications whereas the identified PTF model replicates the volatility in the real data.
2.3 Triangle Group COMPA-volatile paid losses with stable calendar year trends
The triangle group Compa has extremely volatile paid losses. A good model measures this volatility and projects it into the future based on explicit assumptions. Even though the data are extremely volatile, the reserve distributions (beyond the evaluation period) estimated nine calendar periods earlier are the same as the ones estimated at the current evaluation period, because the calendar year trend is stable. Moreover, the distributions of the (161) observations left out in the last nine calendar periods are forecast accurately nine calendar periods ago using only 91 observations!
Three triangles are simulated from the model and it is shown that it is impossible to distinguish in respect of variability between the real triangle and the simulated triangles.
The residuals in ELRF look very bad. Not only do link ratio methods not capture the trend structure, the error structure is not symmetric. That is, the log normal distributions on a dollar scale are very skewed because the corresponding normal distributions on the log scale (in PTF) have a high variance. The data must be modelled on a log scale!
2.4 WC California, Reserve Upgrade Myths and Case Reserve Estimates (CRE) versus Paid Losses
A Californian Workers Comp portfolio exhibits stable calendar year trend for the last ten years, yet many companies lost much money in this business. A portfolio belonging to Frontier exhibits high calendar year trends for many years prior to its closure by the New York insurance department. Why would you close a company that has had the same high trend for many years? A total loss reserve upgrade myth is also debunked. Measurement of calendar year trends (and their volatility) is one of the critical assumption used in forecasting probability distributions of reserves.
Case Reserve Estimates and Paid Losses are modelled separately. From their respective model displays one can tell almost instantly when this company was sold. In the two years prior to the sale, whilst negotiations were conducted, the case reserve estimates decreased by almost 30% a year, yet the paid losses were always increasing. The reinsurers involved in providing aggregate stop loss cover for this transaction lost a lot of money! This kind of information cannot be extracted from the incurred data and moreover link ratios for the incurred data have no predictive power. They are also information free! For another real live example Case Reserve Estimates lag paid losses in respect of calendar year trend changes.
2.5 Triangle Group LR High - Mack Method gives results much too high!
Application of the Mack Method, equivalently, volume weighted average link ratios give mean forecasts for the cumulative paid losses array that are much too high. This is immediately seen by examining the residuals versus calendar years- they exhibit a very strong negative trend. This means that the calendar year trend measured by the link ratio methods (including Mack) is much higher than the trend in the data. Of course, methods in the ELRF modelling framework do not have descriptors of trends measured. However, it is the residuals that inform us about the difference between the trend in the data and the trend measured by the method.
To illustrate the ideas we simulate a triangle with a 10% calendar year trend and calculate residuals by setting the trend in the method to be 20%.
PTF modelling of the paid losses, case reserve estimates and number of claims closed arrays provide much information with respect to each loss development array and their interrelationships. Forecasting with the most recent calendar year trend, the lowest experienced up to date, in the paid losses produces much lower answers than Mack. The calendar year trends in the case reserve estimates and number of claims closed arrays provide some evidence that the future trend in the paid losses may reduce further. This means that the default forecast in the paid losses is on the conservative side (which is much lower than based on link ratios). Most importantly, in the PTF modelling framework, assumptions are explicit, can be controlled by the actuary, and can be related to past experience.
Also see demonstration videos 10.1-10.4 to see how the bootstrap can be used to show that the Mack method can give false indications whereas the identified PTF model replicates the volatility in the real data.
2.6 Three layers 1Mxs1M, 2M and 1M, pricing and aggregate distributions
For three layers, limited $1M, $1Mxs$1M, limited $2M, the trend structure is the same and residuals are very highly correlated. For the layer $1Mxs$1M the calendar year trend estimate is subject to high uncertainty that is it is non- credible, equivalently, insignificant. This is due to high process variability in the layer. The distributions of aggregate losses for calendar years, accident years and grand total are obtained by simulating from the log normal distributions (using their correlations) projected for each cell. Different prospective and retrospective adverse development cover programs are designed for the layer $1Mxs$1M.
In the next chapter we make use of the high process correlations between the three layers to design a composite model that relates the three layers (triangles).
2.7 Uneven sampling periods and updating the database
Loss development arrays can now be specified with more frequent valuation dates than accident periods. For example, arrays can be specified with annual accident periods and monthly development and calendar periods. This new feature results in two important applications: more regular reserve valuations and valuations of short tail lines such as disability and health insurance.
In this video, the uneven sampling period feature is demonstrated along with updating triangle groups, monitoring and forecast tracking. Once you have a model at year end 2006, say, there is no need to design a new model at year end 2007. Updating, monitoring and forecast tracking are effortless and seamless.
2.8 How can we tell that data modeled by Murphy et al is not real?
In the paper Manually Adjustable Link Ratio Model for Reserving by Murphy et al on the CAS web site, a dataset is modeled using the Mack method. Notwithstanding that the Mack method for these data is completely unsound we demonstrate effortlessly that the data are not real. Our submission on “Meaningful Intervals” explains this. The fact that the data are not real was subsequently corroborated by Daniel Murphy at the CLRS held in Washington 2008.
For additional information on ICRFS-Plus™ features - click here.
Solvency II Capital requirements for each LOB and the aggregate of all LOBs are only met by ICRFS-Plus™ in a sound statistical framework.