ICRFS-Plus™ Training videos for new users
If for any reason you are unable to view the training or demonstration videos, please contact our support staff at support@insureware.com 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.
The training videos should be used for hands on training.
We suggest you run the videos on a separate computer using a data projector, and train as a group.
The only way you will learn all the new concepts and be able to
exploit all the immense benefits is by using the system. Experiential
learning is imperative.
It is important that you study the videos in sequential order set
out below.
Table of Contents
1. Introduction to ICRFS-Plus™ 10 and modelling modules
3. ICRFS-Plus™ 10 modelling wizard and other features
- 4.1 MPTF Concepts discussed in the context of two lines of business: LOB1 and LOB3
- 4.2 MPTF Three layers are considered along with outward reinsurance
- 4.3 MPTF Functionality is revisited and the method of constructing an optimal model is given
- 4.4 MPTF Example: CompA versus Industry
- 4.5 MPTF Example: Credibility modelling - extending a short triangle
- 4.6 MPTF Diagnostic modelling and triangle decomposition
- 4.7 MPTF Coefficient of variation analysis: M1 1Mxs1M 2M
5. ICRFS-Plus™ 10 and NAIC Schedule P
4. MPTF modelling
The MPTF module of ICRFS-Plus™ is used to design a composite model for multiple lines of business, multiple segments and multiple layers. It has many benefits including assessment of level of diversification, design of optimal outward reinsurance and credibility modelling.
4.1 MPTF Concepts discussed in the context of two lines of business: LOB1 and LOB3
The starting models in MPTF are typically the models designed in PTF using the Wizard. Correlation, linearity, linear regression, weighted least squares and normality are intimately related concepts. Two lines of business LOB 1, LOB 3 are first studied in PTF. The individual PTF models are run in MPTF. We find two types of correlations, namely, process correlations and parameter estimates correlations. These induce correlations between any two pair of cells between the two lines of business and hence correlations between all aggregate forecast distributions between the two lines of business. Level of diversification is assessed and is compared to the case assuming the two lines are completely unrelated.
We also study a case involving 40 hypothetical lines of business.
Click here to view.
4.2 MPTF Three layers are considered along with optimal outward reinsurance
Three layers 1M, 1Mxs1M and 2M are first studied in PTF. The structure of each model is the same and as expected process correlation is very high. This example is used to explain the difference in modelling approach for triangles having high process correlations compared to triangles having insignificant process correlations. We also find that outward reinsurance where each individual loss is limited to 1M ($1,000,000) is not more capital efficient than outward reinsurance where each individual loss is limited to 2M ($2,000,000).
We study a composite model for gross data and data net of reinsurance and find that the outward reinsurance program is far from optimal from the point of view of the cedant.
Click here to view.
4.3 MPTF Functionality is revisited and the method of constructing an optimal model is given
The study involving 1M, 1Mxs1M and 2M is revisited and used to explain the various buttons (icons) functionality. We also show how to create the optimal MTPF composite model from the individual PTF models (for each data set) using a simple five-step procedure.
Click here to view.
4.4 MPTF Example: A Company versus Industry shows major differences
The TG COMPA contains paid losses for a company writing relatively low exposures. The paid losses are extremely volatile yet calendar year trend is stable. This TG was studied in Video 6. As a result of high process variability the estimate of the calendar year trend is not significant, equivalently, it is not credible. We use the optimal PTF model for industry data, TG Maa951, that has very little process variability and much larger trends, to create a composite model for the two data sets. The COMPA component of the composite model has a significant (credible) calendar year trend. Which trends are credibility adjusted depends on the information in each triangle and their inter-relationships! The industry and the company experiences are very different, both in terms of trends and process variability.
Click here to view.
4.5 MPTF Example: Credibility modelling - extending a short triangle
Company B is a triangle that has very few accident years. Company A is the same line of business but has larger dimensions. We use the “information” in Company A to design a credibility model for Company B.
Click here to view.
4.6 MPTF Diagnostic modelling and triangle decomposition
MPTF modelling of LOB 1 and LOB 3 is revisited in order to discuss cross data sets constraints. In PTF both the CL and SM are diagnostic models for testing calendar year and accident year trend patterns respectively after removing the trends in the other two directions. In MPTF the CCI model is a diagnostic model for testing process correlations after removing for each data set development year trends and average inflation. Development year is regarded as a covariate.
Another significant application of MPTF is the decomposition of one triangle into two and designing a composite model for the two. For example, suppose we have a (complete) triangle with accident years 1980-2004 and the mix of risks written changes between accident year 1990 and 1991. Calendar year inflation, for calendar years 2000-2004 for certain years may not be the same for accident years 1980-1990 and accident years 1991-2004. This also applies to a merger where on company does not have the same experience as the other.
Click here to view.
4.7 MPTF Coefficient of variation analysis: M1 1Mxs1M 2M
MPTF modelling of 1M, 1Mxs1M and 2M is revisited to show that coefficient of variation of total reserves for individual losses limited to $1M is the same as the coefficient of variation of total reserves for individual losses limited to $2M. This means that the % capital required above the mean to achieve a specific percentile is the same in each case. It is explained that this is not counter intuitive.
Another example is shown involving layered paid losses.
Click here to view.