ProVal has two types of forecast models that utilize the results of a Core Projection: the Deterministic Forecast and the Stochastic Forecast. Their essential difference is in the number of scenarios of future experience produced by the forecast. The Deterministic Forecast produces only one future scenario, whereas the Stochastic Forecast produces as many scenarios as there are trials specified. It is useful to perform a deterministic forecast prior to running a stochastic forecast, because the ability to determine, that is, to specify exactly, what the future experience will be enables you to check the reasonableness of results and spot possible coding errors or misconceptions. For example, if your Projection Assumptions are identical to your Valuation Assumptions for parameters they have in common and you apply no interest rate sensitivity, then, provided your Deterministic Assumptions match your Asset & Funding Policy for parameters they have in common, the absence of experience gains and losses over the period of the forecast would confirm that, very likely, the results are “good”.
To perform a Deterministic Forecast, you specify various deterministic assumptions that describe the experience of the plan’s assets during the forecast period and that indicate what future valuation interest rates ProVal should use when it interpolates the core projection’s results during the forecast. (For more information, see the Technical Reference article entitled Interpolation of a Core Projection's results.) You tell ProVal, (generally) separately for each forecast year, exactly what the economic experience and valuation assumption interest rates will be. Although you may wish to assume that the plan’s experience will be as anticipated in the actuarial assumptions used to perform the annual valuations throughout the projection period, experience assumptions need not be the same as valuation assumptions. In fact, one of the useful aspects of a forecast is to study the implications of experience deviating from assumptions. Such deviations, of course, will generate actuarial experience gains and losses, which, in turn, will affect annual plan costs, funded ratios and other results.
The Deterministic Assumptions dialog box contains the following parameters:
Name is a descriptive phrase to use when saving this set of deterministic assumptions.
The Populate with stochastic trial parameter lets you populate deterministic assumptions with experience assumptions, and with future valuation interest rates, from a particular trial of a Stochastic Forecast (that is, copy the trial’s assumptions into the Deterministic Assumptions set you are editing). For details of coding the trial population parameters, see the discussion at the end of this article.
For deterministic assumption sets in the U.S. qualified mode, ProVal needs to know whether, and how, to reflect the Pension Protection Act of 2006 (PPA). The Applicable law parameter determines the future valuation interest rates available for parameterization for Deterministic Forecasts. The available law choices are:
| Multiemployer | for multiemployer plans (as defined under ERISA) |
| PPA | for single-employer plans (as defined under ERISA), to apply PPA only |
| Pre-PPA | for single-employer plans, to apply the law as in effect prior to PPA only |
| Pre-PPA and PPA | for single-employer plan forecasts, to apply both the law as in effect prior to PPA and PPA law, with the transition occurring at the applicable effective date indicated in the Asset & Funding Policy. |
Note: The underlying Core Projection(s) must produce the necessary liabilities for your choice of Applicable law. Therefore, be sure that your selection here is consistent with the law selection of the Valuation Assumptions set for each Core Projection to be included in the Deterministic Forecast referencing this set of Deterministic Assumptions. For example, you may use a “pre-PPA” set of Deterministic Assumptions with either a “pre-PPA” Core Projection or a “pre-PPA and PPA” Core Projection but not with a “PPA” Core Projection. Furthermore, your choice of applicable law must be the same selection as made for the Asset & Funding Policy to be referenced by this set of Deterministic Assumptions.
Select a topic to edit contains entries for each category of information (topic) found under the Deterministic Assumptions command. To some extent, these categories parallel those entered for the corresponding Valuation Assumptions and Asset & Funding Policy, allowing you to specify deterministic assumptions that are the same as or differ from the assumptions of your valuation with respect to liabilities (Valuation Assumptions) and assets (Asset & Funding Policy).
There are four topics in the Economic Experience Assumptions group (reflecting what you assume will be the actual experience over the forecast years). Click the name of a topic to access its parameters. Future Valuation Interest Rates refers to future valuation interest rate assumptions used at the various forecast valuation dates. Dynamic Asset Allocation, if selected, permits Investment Return, Accounting Expected Return, and Canadian Pfad to vary by funded ratio.
Economic Experience Assumptions – heading is for descriptive purposes only; click indented topic name:
Investment Return, Inflation & Lump Sum Benchmark Yield (pension modes) or Investment Return, Inflation (OPEB mode)
Additional Contributions & 420 Transfers (U.S. qualified mode) or Additional Contributions (other modes)
Future Valuation Interest Rates
In Public mode, check Vary by group and specify a coded field that represents the funding groups. If checked, certain parameters throughout the Deterministic Assumptions can be entered by group. The parameters to enter by group will be ghosted and a spreadsheet icon 
will appear. Click on the spreadsheet icon to enter information by group. All information can be copied and pasted from an outside source. In a Deterministic Forecast, Deterministic Assumptions by group can be combined with both an Asset & Funding policy that varies by group and Core Projections that are subtotaled by the same coded field to produce results by group.
Populate with stochastic trial
Deterministic Assumptions can be populated by specifying a Stochastic Forecast, Asset Mix and telling ProVal how to select a stochastic trial. (Note: only executed Stochastic Forecasts appear in the list of Stochastic Forecasts unhidden in the current Project.) Tell ProVal how to Select trial by either trial Number or by indicating an Output Variable (e.g., nominal return), forecast Year and Percentile of interest. If you select the Variable option, ProVal will determine the trial number based on the Output Variable, year and percentile you chose to view. Otherwise, trial numbers can be determined by using the Stochastic Forecast Output command. (ProVal cannot determine trial numbers for stochastic forecasts run using version 2.27 or earlier. Therefore, to determine trial numbers for a Stochastic Forecast run in a version earlier than 2.28, simply rerun the forecast.)
There is a key difference between Deterministic Assumptions that are populated “by hand”, that is, by the user, and those that are populated by reference to a Stochastic Forecast. The difference is in the interpolation methodology. In general, when ProVal executes a Deterministic Forecast, it interpolates by benefit all the liabilities and normal costs for the desired economic scenario and valuation assumptions. It then sums across the benefits to get the total liabilities. To save space and time in a Stochastic Forecast (and in ProVal PS), however, ProVal sums across comparable benefits first and then does the interpolation. This difference in methodology generally will create a very small difference in the final costs and contributions. To eliminate this “noise”, ProVal remembers if a set of Deterministic Assumptions had been populated by reference to a Stochastic Forecast, whether or not it is subsequently modified by the user. If it had been populated from a Stochastic Forecast trial, the interpolation methodology for the Deterministic Forecasts is set to “collapse on benefit”, so that the numbers will match exactly what would have been produced by the Stochastic Forecast. If, for pedagogical reasons or because of intellectual curiosity, the user wants to see results using standard deterministic forecast interpolation methodology, the deterministic assumption set can be exported and reestablished in a new deterministic assumption set.