• | Maximum forecast horizon: Choose the maximum length the forecast sample can take on. If your model has exogenous variables, then you need to make sure that the data for these variables cover the forecast sample. For more information, see the Data Construction File. |
• | Number of observed variables paths per parameter value: The predictive distribution is computed by simulating paths for the observed variables. Similarly, the distribution for the sample moments (mean, standard deviation, autocorrelations) is also computed by simulating data for the selected sample. The value chosen from this control determines how many such paths YADA will compute per parameter value. Integer values from 1 to 1,000,000 are supported. |
• | Method for controlling conditional forecasts: YADA supports two methods for dealing with conditional forecasts. The first is Values for shocks, where a group of shocks are linked to a group of conditioning variables. The shocks are then manipulated over the conditioning sample to ensure that the conditioning assumptions are fulfilled. The second method is Distribution of shocks, where the shocks over the conditioning sample are drawn from a distribution that guarantees that the conditioning assumptions are met; see . |
• | Adjust prediction paths (sample mean = population mean): When is simulating paths for the predictive distribution for a given parameter value, the population mean (theoretical mean) can differ for a finite number of paths from the sample mean. If you check mark this option, then the paths are adjusted such that the sample mean for each time period T+h is equal to the population mean. |
• | Use conditioning data for state distribution: A value for the state variables for period T (last period before the forecast sample) is required when simulating paths for the observed variables over the forecast sample. By default YADA uses draws a value for the state vector from a normal distribution with mean equal to the smooth estimate of the state and covariance matrix equal to the covariance matrix for the smooth estimate. When running conditional forecasts, this distribution does not take the conditioning assumptions into account. To make sure that YADA utilizes the conditioning assumptions for this distribution you need to check mark this option. |

Additional Information
• | A more detailed description about forecasting is found in Section 12 of the YADA Manual. |
Page url:
http://www.texlips.net/yada/help/index.html?forecasting.htm