|
|
| Welcome
to Inside AXIS, our newsletter aimed at giving you,
our client, tips and information that can help with your daily
work. Articles can be quickly accessed using the menu bar on
the left hand side of your screen. You can also scroll down
the screen to read the articles, including useful links. |
| |
|
|
|
|
|
|
|
AXIS now offers the ability to include Hedging Strategies
when modeling Variable Annuities and Seg Fund Annuities.
This module addresses a major need of insurance companies
offering guarantees related to Equity Fund Performance who
need to assess hedging strategies designed to mitigate the
financial risks in these guarantees and to price and reserve
for these guarantees.
After several years of research and development, GGY has designed hedge simulation functionality that is fully integrated with the Annuity Module and with the entire financial projection process in AXIS. At any reinvestment date, AXIS will calculate Greeks and choose an optimum selection of hedges based on user selected parameters using those Greeks. Furthermore, when AXIS is used to calculate reserves or capital on methods that allow credit for an established hedging strategy (AG 43 and C3 Phase II in the US), AXIS can simulate that hedging strategy within each of the scenario projections required to calculate these reserves. And with the Embedded Stochastic Processing approach, these reserve and capital calculations, including hedging, can be projected to multiple future dates.
AXIS is not the first actuarial modeling system to address hedging. But in most cases hedging functionality available has been dedicated to producing hedge analytics such as Greeks, and any projection capabilities have been limited and inflexible. AXIS breaks new ground in detail, power and flexibility of its hedge projection feature set.
Since Greek calculations require an embedded risk neutral scenario calculation, any projection of reserves and capital using dynamic hedging will be a stochastic-in-stochastic calculation, nested inside the main projection. Clearly the performance demands are immense. But GGY is facing up to this challenge with a variety of existing tools and planned enhancements, including of course the GridLink module which enables large farms of servers to be efficiently utilized and managed for such complex calculations. With the pending release of AXIS, yet another distribution approach will be added to AXIS GridLink to help efficiently spread the workload in nested stochastic calculations over the available processor cores.
GGY has grown substantially over the past few years, now numbering almost 90 staff in its Toronto main office and in various “home offices” around North America. Some of these new additions bring amazing academic credentials and relevant work experience in the Variable Annuity market, hedging theory and practice, and scenario generation. Look for more announcements on the Hedging front as we continue to improve the speed and functionality of our Hedge Projection Module and announce new companion modules including Daily Hedge Analytics support.
-
Trevor Howes |
| |
|
|
|
|
|
|
|
In the October version of AXIS we will be releasing a Beta
version of the Policy Audit Report. This report will allow
users to produce a single output file per policy (Excel or
PDF) containing various existing reports as well as specific
Policy Audit Report information.
The report is selected as part of a Dataset Formula where
specific policies can be earmarked to produce user defined
reports. These reports are a compilation of various AXIS
Summary sheets, where each Summary sheet will generate an
Excel tab or PDF page. As per existing Summary sheet
functionality, the sheets are customizable e.g., allowing
access to calendar year rows and formulas and the use of
free form text.
The new functionality enables embedding of any model
point output within a summary report, whether Calendar Year
Report or Policy Year Report. In addition, specific Audit
Report output is available including:
- Policy ID
- A listing of the seriatim fields on the record and their
values
- Assumptions used in various object filters, e.g., Cell
screen, Reserve method, Other benefit, Reinsurance terms,
Investment accounts
For further information please contact David Fihrer at (416) 250-3444.
-
David Fihrer
|
| |
|
|
|
|
|
|
|
AXIS has provided support for FAS 97 and SOP 03-1 for a
number of years now, but as our base of US clients has
continued to grow over the past few years, we have added
additional resources focused on providing a more robust,
transparent system for performing US GAAP valuation and
projecting out US GAAP financials for FAS 97 products.
Actuaries looking to perform valuation work using AXIS
will find a complete set of tools for calculating and
analysing the current month’s DAC, URR and SIA balances. A
US GAAP cohort is assigned to each seriatim record, allowing
complete flexibility in how business is grouped. There are a
number of options for how historical cohort information is
entered, allowing this information to be either handled
directly inside AXIS or processed externally and fed into
AXIS. There are vendor defined reports available which allow
the user to drill into each projected margin or deferral by
cohort, cell or even model point. At the cohort level, we
have added additional vendor defined reports which decompose
each balance down into all of the pieces, including present
values, accumulated values and any K-factors.
Our support of SOP 03-1 has evolved over the past couple
of years as well. Users can now run stochastic SOP 03-1
projections directly within their base FAS 97 run. This
allows for a single, integrated run that combines
deterministic and stochastic results. Since it is integrated
into a single run, all of the analysis that can be performed
on the deterministic results is available for the stochastic
SOP 03-1 projections. This includes the ability to drill
down to the model point level on projections as well as a
complete breakdown of the liability balances into their
components. Another benefit of this integrated run is that
the SOP 03-1 liability is properly reflected in the margins
used for DAC and URR, including any circularity that may
arise with cohorts that have both an SOP 03-1 liability and
an unearned revenue reserve.
With the above functionality now in place for valuation,
our next focus is on projecting GAAP financials into the
future. This functionality will allow future assumptions and
balances to be unlocked as the pricing projection emerges
independent of the valuation projections. This will apply to
both the deterministic projections, as well as any
stochastic SOP 03-1 reserves required in the future. We are
planning on completing these enhancements in the first half
of 2011.
As always, if there is any feedback, either questions or
comments, please contact our help desk and we will be happy
to discuss.
- Sean Hayward |
| |
|
|
|
|
|
|
GGY is pleased to announce the third series of videos on recent enhancements and features added to AXIS:
Visit
http://www.ggy.com/support/training/videos/playvideo.asp to watch and learn more about these
and other exciting new developments - and stay tuned for more videos that describe continuous improvements to AXIS.
-
Karim Sajan |
| |
|
|
|
|
|
|
|
We develop the AXIS Actuarial System to meet the needs of our current and potential clients but I thought you might like to know how much we use it for our own purposes. We are not an insurance company, and we don’t do consulting, so you might think we never sit in the position of users. First, AXIS was originally designed for me and my former colleagues at National Reinsurance to do reinsurance pricing, valuation and modeling. Well here’s a quick list of the uses we make of AXIS today:
- We use AXIS when we sit on Industry Committees looking at new developments such as Principles Based Reserving and Capital or Hedging.
- We build models of a sample of each potential client’s assets and liabilities as a part of our Marketing.
- We build models to test every bit of new functionality we add into the system.
- Some of our actuarial students use AXIS to replicate the calculations they are studying for their exams, and use our help text to better understand a wide range of study material.
- Some students have used AXIS to build examples for their
DMAC (Fellowship) projects.
- Our Client Support staff use AXIS to validate spreadsheets sent in by our users, and help clients use AXIS to model new or complex features.
- We make enormous use of AXIS GridLink in our Quality Control and testing processes.
As you can see, we use AXIS for many things, and I can’t describe them all here but I would like to concentrate on just one mode where we are certainly power users. So I’ll just tell you how we use GridLink here at GGY, since it has become such a vital part of our operation. We use it for a variety of purposes:
- Developer Testing
- Nightly Testing
- Release Testing
- Improving Scalability
- Client Support
- Outsourcing
Developer Testing
Our source code sits in a Version Control System
called Microsoft Team Foundation Server (TFS). When a
developer starts a new job, she (or he) will get a
current version of the code from TFS and make the
necessary changes on her own workstation. Before that
code can be checked back into TFS we need to have the
changes checked, tested and documented, and new cells
must be added to our library of test data to make sure
that new functionality will be tested routinely
thereafter in our nightly testing procedures. Part of
the process is that the developer must run through a
large AXIS dataset to make sure her changes have not
affected the results of any other functionality in AXIS
and that speed has not been compromised. She could do
this test on her own workstation but that would take a
lot of time and the results of those tests would be
quite unstructured. So we built a special GridLink farm
of six machines for developer testing. Each developer
can send both code and data to this farm for testing
under special calibrated and documented conditions. This
test farm will automatically compare the before and
after code results, highlighting differences in
calculations, messages and speed, which can then be
addressed by the developer before the new code can be
accepted. While this testing is going on, the developer
is free to work on the next job on her own workstation.
Nightly Testing
Our Quality Control Team does regression testing
every day on a much larger basis. They will test the
current version of AXIS compiled from all the latest
files in TFS, and they do so on a much larger set of
test data. For this process we have set up a larger test
farm which looks more closely and in an even more
systematic way, feeding a large database of regression
test results for later referral. The idea here is to
make sure that the code changes made by multiple
developers do not combine to introduce errors. We are
comparing each night’s run to the run from the night
before. Version Testing
Using the same GridLink farm that we use for the
nightly testing, we also run a suite of regression tests
before any new version can be released. Here the
comparison is between the last published release of AXIS
and the one we are about to release. This process can
take a week or more as we run a huge set of datasets,
some sent in for this purpose by our clients, all of
which must be fully reconciled. Improving Scalability
When we first developed distributed processing, we
supported up to 8 processor cores running in parallel on
a single job. Today we support 256 cores working on a
single job, and multiple jobs working simultaneously on
one farm. That has taken an enormous amount of testing
and development work to eliminate scalability
bottlenecks. In fact, we are working now on a further
increase to 512 cores on a single job. For our
scalability testing, we need access to a large farm, and
we have built one of 256 cores for this purpose.
Astonishingly, this rack-mounted server farm is just 28”
high, but it takes a lot of dedicated air conditioning
to keep it running smoothly. By combining our various
farms together, and by using other available computers,
we can currently assemble a farm of 512 cores on an
occasional basis, but you can also expect us to increase
the size of our standard test farm continuously to meet
the demands of testing. Client Support
Clients often send datasets to GGY for testing
purposes – either to see how fast they would run on a
farm of a particular size, or to investigate an issue.
We can run such datasets on our large farm to give the
clients the analysis they need. Outsourcing
Users may rent time on our large farm on a first come
first served basis. Our network administrators will run
selected jobs on client datasets uploaded over our
secure connection, and will make the results available
for download when they are done. And Now for
Something Completely Different
In preparation for the launch of our new AXIS
EnterpriseLink Module (E-Link), we are setting up
various areas of our company to be the first E-Link
users. We will start with our Client Support specialists
and move on to the Quality Control Team, so that before
it gets to market, it will have been thoroughly tested
and we will have built up a lot of practical experience
with it. - Phil Gold |
| |
|
|
|