Implementation

The PensionMetrics software is written in MATLAB but does not require that the user have MATLAB or any knowledge of MATLAB; the coding (which is proprietary, but licensed to clients) can also be compiled in other languages.

There are different ways in which the model can be delivered to the client, depending, for example, on the software and other platforms involved, and the extent to which PensionMetrics might be integrated with client’s existing tools. In our experience, clients have very different software setups and delivery will require further discussion to determine what is best for your needs. Possible means of delivery include: putting a stand-alone GUI directly on the client’s website, using servers that make use of the MATLAB Production Server (MPS) software, compiling to Java, .net, Python, an App for an Apple or  Android product (e.g., ipad or smartphone), etc.  With some clients, we use third-party experts to help implement PensionMetrics into their own systems.

We can meet the following additional client requirements:
  • The client can design and own their own GUI
  • The interface will integrate with the stochastic model in a way that will enable real time flexing on the inputs to see changes in the resulting outputs (e.g., by either including calculation engine within the tool or a full set of indexed data tables, or some combination)
  • The range of inputs and corresponding outputs will be recordable by the client as client specific data
  • The key parameters underlying the stochastic model (and any future changes) will be proposed by us, subject to agreement, and if requested, changed by the client
Our due diligence processes can be made compliant with any requirements set by the client.  We wish to ensure that all the 'technical' issues are sound and pass due diligence in terms of:
  • how PensionMetrics works
  • how PensionMetrics is calibrated
  • backtesting
  • how it fits with client’s systems  
  • model updates
Our default maintenance cycle is as follows:
  • The key parameters will be reviewed at least every 12 months
  • An updated model will be released once a year, which will incorporate updated calibrations, model refinements and innovations
  • If software errors are identified, a corrected version of PensionMetrics will be released immediately