AXIS Module - Scenariotools

The AXIS ScenarioTools Module is a package of tools within AXIS that allows you to generate scenarios, analyze the scenarios, and analyze the results projected when you run the scenarios. A number of batch-based scenario generators are available in the ScenarioTools module. Some are designed to generate interest yield curves; others generate equity rates, while one generates yield curves and equity rates. Also available is an option to generate values on simple user-defined distributions.

A new framework for scenario generator has been added to AXIS. This new framework allows scenarios to be generated on the fly during stochastic processing where a batch-based scenario generator would not be accessible. This is particularly useful for the determination of principles-based reserves and capital requirements. This new framework allows the user to select from a number of available stochastic models for the scenario generator.

Batch-based Scenario Generators available are:

• Yield Curve generators:

• Correlated Monte Carlo Simulation for Yield Curves (Ravindran 1998) which generates spot yield curves using a risk neutral simulation model.
• Two Factor Lognormal Model (Boyle and Tan 1998) which is also used to generate spot yield curves.
• Calibrated Correlated Monte Carlo Simulation which is used to generate spot yield curves that are calibrated to the current and implied market prices.
• Prescribed CALM Scenarios which generates the 7 CALM prescribed scenarios consistent with the Canadian Standards of Practice for prescribed valuation scenarios.

• Equity generators:

• Correlated Monte Carlo Simulation for Equities (Ravindran 1998) which generates equity growth rates using a risk neutral simulation model.
• Correlated Lognormal Equities - Real World which is used to generate equity growth rates in real world approach.
• Regime Switching Lognormal Model - Real World (Hardy 1999) which is used to simulate growth rates assuming a discrete process that switches between 2 regimes randomly.

• Dual generator:

• Stochastic Asset Model for Actuarial Use (Wilkie 1995) which is used to generate both the spot yield curves and equity income and growth rates using an actuarial model.

• User-Defined generator:
  • Allows you to specify distribution types and parameters for one or more columns in the scenarios using one of the four distribution types:
    • Uniform
    • Normal
    • Lognormal
    • Triangular

Stochastic Models (known as Market Models in AXIS) which are available for the new scenario generator framework include:

• Lognormal Model

• Generates equity growth rates for either risk neutral or real world measure.

• Regime-Switching Lognormal Model

• Generates equity growth rates assuming a discrete process that switches between 2 regimes.
• Supports fitting model parameters to a set of historical data.

• Integrated Hull-White Lognormal Model

• Generates both the yield curves and equity growth rates.
• Allows for correlation between the random processes in the Hull-White and Lognormal models.

• User Defined Stochastic Model

• A model framework which allows the user to program his/her own model using AXIS script.

More models will be made available under the new scenario generator framework over time.

Valuation Scenario Formula Tables

In lieu of Market Models, Valuation Scenario Formula Tables (VSFT) can be selected within the new scenario generator. Each formula table will generate one scenario, and up to 50 VSFT can be selected. VSFT, which is based on AXIS script, is used to generate a deterministic future scenario where its starting point may be the current scenario or the valuation scenario at some revaluation date. Built-in input variables, input functions and output functions are provided to assist the user in writing his/her own formula tables. Scenario generators for the prescribed scenarios used in the Canadian Asset Liability Method (CALM) and the New York Regulation 126 (NY7) are examples of deterministic scenario generators.

Scenario Analyzer and Results Analyzer

The "Scenario Analyzer" produces a summary of selected statistics on a group of generated scenarios, which are useful in comparing different generates, and confirming that the sample parameters are close to the model parameters. Reported measures include the mean, standard deviation, skewness, kurtosis, rho, min, max, percentile, and CTE. The ScenarioTools module also includes access to the "Results Analyzer". This tool allows you to produce statistics on various Summary Report results calculated by AXIS under multiple scenarios. Similar to the Scenario Analyzer, the statistics available in the Results Analyzer include: mean, standard deviation, skewness, kurtosis, rho, min, max, percentile, and CTE.