CSSM Procedure

By default, the CSSM procedure computes the estimates of all the components that are specified in the MODEL statements (you can print these estimates by using the PRINT= option in the respective TREND and COMPONENT statements, or you can output these estimates to a data set by specifying it in the OUT= option in the OUTPUT statement). However, in many cases it is desirable to obtain the estimates of additional linear combinations of the state elements and the regression effects. The CSSM procedure provides two statements, the COMPONENT statement and the EVAL statement, that are useful for specifying virtually any desired linear combination of the elements of the state vector and the regression effects in the observation equation. After a desired linear combination is specified, you can print or output its estimate as you would for a component that is used in the MODEL statement. This feature of the CSSM procedure is illustrated in many examples in the section Examples: CSSM Procedure. For example, in the second part of Example 14.4: Smoothing of Repeated Measures for Longitudinal Data, the COMPONENT and EVAL statements are used to define the contrasts between the growth profiles of cows that are receiving different treatments. Similarly, in Example 14.7: A Dynamic Factor Model for the Yield Curve, the EVAL statement is used to define the yield curve as a sum of the components that are used in the MODEL statement.