financial statements, the most logical and consistent course of action is to assume that entire fixed-income portfolio of DFAIC is taxable.
In contrast to the artificial data problems introduced by fictional data, there were other issues raised by the reported DFAIC data that, if the option were available, could be resolved by talking to management. Prominent among those issues were (1) DFAIC's 18% allocation to cash and short-term investments (industry average is 4%) and; (2) reported changes in case reserve adequacy. These are the types of data analyses that are required at the front end of the DFA process to assure the robustness of the ultimate DFA findings.
Step 3: Model Parametedzation
Model parameterization refers to how the asset and liability variables are assumed to behave over the forecast horizon. Assumptions concerning the general economic and capital market environment as well as the specific assets and liabilities of DFAIC need to be parameterized. These assumptions can have a substantial impact on the evaluation and the recommended strategies. In the modeling world this risk is referred to as "parameter risk." The impact of parameter risk can be investigated and better understood through sensitivity testing.
Economic and capital market assumptions are an important part of any quantitative assessment of the potential rewards and risks associated with alternative strategic business decisions. These assumptions need to reflect both recent conditions and longer-term relationships inherent in the economy and capital markets. The simulations based on these assumptions should comprise a reasonable set of future paths that, while consistent with histodcal observations, reflect a prospective view of economic and capital market expectations and uncertainties.
The model that we used to generate our DFA economic and capital market simulations (FIRMTM Asset Model) differs from traditional mean/variance models in that economic variables, including interest rates and inflation, are explicitly modeled using accepted and rigorously tested stochastic processes. Capital market returns are then generated on a consistent basis with the underlying economic environment. This type of model has the following advantages over traditional mean/variance models:
the explicit modeling of both economic and capital market variables;
the ability to incorporate mean reversion in yields, providing for control over the serial
correlation of capital market returns over time;
multi-period simulation capabilities; and
additional flexibility in modeling asset categories such as mortgage-backed securities
and other securities with embedded options.
The economic and capital market parameterization process involved identifying and selecting asset classes that best represented the homogeneous groups of invested assets available to DFAIC. The twelve asset classes we defined and modeled were: