Years, Trials & Accounting Results
The parameters of this topic specify the time period for which investment returns are simulated and the number of trials in the simulation and allow you to select a monthly, rather than annual, return base for the simulation. Furthermore, if monthly returns are generated, you have the option to generate separate accounting returns, as would be appropriate if the (accounting) Measurement Date is not the same as the (funding) Valuation Date.
Years in simulation indicates the (whole, not fractional) number of future years (from 2 to 100) that the Capital Market Simulation is to cover. Generally, this is the same as the number of years in the Stochastic Forecast. If you have more years in the Capital Market Simulation than in the forecast, ProVal will use the earlier years of the simulation in the forecast run, ignoring results of any years in the simulation after the last year of the forecast. Use of a Capital Market Simulation with fewer years than the number of forecast years, however, will result in an error message and aborted run when the forecast is executed.
Number of trials indicates the number of scenarios of future experience in the Capital Market Simulation. The sample error of the simulation gets smaller as the number of trials increases. If the standard deviation of investment returns of your asset classes is large, you should select a large number of trials to reduce the sample error.
You may choose to run any number of trials from 10 to 25,000, inclusive. Generally, the number of trials is the same as in your stochastic forecast. If you have more trials in the Capital Market Simulation than in the forecast, ProVal will use the first trials of the simulation in the stochastic forecast run, ignoring results of any other trials in your simulation. Use of a Capital Market Simulation with fewer trials than the number of forecast trials, however, will result in an error message and aborted run when the forecast is executed.
Each future scenario is selected randomly, using the Monte Carlo simulation technique. The Random number seed generates the trials needed for this technique. The seed may be left at the Fixed value that appears when you enter the dialog box or you may enter a different number. If you run your simulation again with the same parameter settings and do not change the value of the seed, then this simulation will be reproduced exactly. Alternatively, you may enter a different number (a whole number between 1 and 999,999,999, inclusive) or select Based on time when run, in which case ProVal selects the seed and, after the run is executed, stores it in the (inaccessible) “Fixed value” text box. Note that if you select the Based on time when run option for this parameter and you run your simulation again with the same parameter settings, you will not replicate your prior results.
Check the Base simulation on monthly returns, and generate accounting returns box if you wish the Capital Market Simulation to generate monthly, rather than annual, returns and then combine the monthly returns geometrically to produce the final annual returns for each month of each year in the simulation. In addition, the simulator will generate separate funding and accounting returns, assuming the specified number of months between the measurement and valuation dates. Enter the number of months of difference as an integer between 0 and 12. If you enter 0, denoting that the measurement date is the same as the valuation date, simulation results for funding and accounting results will not be the same as if the Base simulation on monthly returns, and generate accounting returns box had not been checked, because monthly returns will be produced, instead of the annual returns generated when there is no check in this box. Note that if, based on input to the Asset & Funding Policy, the Measurement Date does not follow the Valuation Date by the same number of months as specified for the months between the measurement and valuation dates parameter, ProVal will generate accounting returns based on the number of months specified by the months between the measurement and valuation dates parameter, rather than by the settings of the Measurement Date and Valuation Date parameters (and a warning will be issued). Standard output variables are available, under the Stochastic Forecast Output command, to view funding returns. To view separate accounting returns, however, you must first define custom stochastic variables for the accounting values and include these custom variables in your Stochastic Forecast.