Models require an organizational structure. The Excel spreadsheet has a structure of workbooks containing sheets (or pages, or tabs), which in turn, contain cells referenced by row numbers and column letters. Building a model with VORSIM requires you to select an Excel-like organizational structure for your model. However, within the Excel framework, you have many choices for organizing your model structure, variables, data, and equations.
A model actually consists of many Excel Workbooks such as the definition, equation, data, and indicator workbooks. The final model itself is a workbook. The Sheet is the highest division of model organization within workbooks. You may want all information about a country to reside on a sheet in workbook. For example, the model DEMO has three sheets (the Eastern region, the Western Region, and a Market Clearing Mechanism) in the equation and data workbooks. Within sheets, subdivisions used in are called Categories. DEMO has three categories, which appear in each equation and data sheet (BoLts, NUTs, and Washers). In the equation workbook for example, all of the equations and parameters for the Eastern region appear on the E sheet, those for the Western Region on the WR sheet, etc. Within each sheet you will find variables and equations for the categories that have the codes BL, NUT, and W. Codes used for Sheets and Categories are chosen for their mnemonic convenience. Model Variable types are designated by lower case letters. Two lower case letters identify variables if they differ by Category and one to five lower case letters if unique to a sheet. Model variable names are concatenations of sheet, variable, and category codes. Sheet codes are added before the variable code type and category codes are put behind the variable code type. Sample variable names would be WRqpBL, MCMt, or WRnew.
In the model DEMO, regions are the main model division (sheets) while products are a subdivision on sheets. In DEMO, the variable types are used to distinguish economic concepts such as production, sales, prices, etc. The variable nomenclature allows each of these concepts to be associated with each category within a sheet. Look at the sheets E or WR in the workbook Demoeqp.xls and see how variable and equation codes are repeated for variables and categories.
DEMO could have been organized differently, even if the same equations were used. E and WR could have been categories while BL and NUT could have been sheets. Then the workbook DEMOeqp.xls would have had sheets BL with variables such as BLqsE. Rather than a region sheet, there would have been a product sheet containing product information about all regions. This illustrates the first major model design choice to be made, the selection of major model divisions (Sheets) and subdivisions (Categories).
Having made the first design choice, the second set of model design choices concerns variable types to be included in the model (as opposed to indicator variables that can be calculated from the solution of model variables (they do not influence model variables)). Codes for these Variable types along with sheet and category codes form the basis for model variables. A model is a set of equations that explain a set of model variables. The equations may be simultaneous and the solutions may differ, depending upon values of exogenous variables. Your model must include variables and equations that change during solution and your data must include exogenous variable values. These “core model” variable, sheet, and category choices MUST be made in the VORSIM model definition process. However you don’t want more variables than you need if you want a model that solves quickly and does not take up too much disk storage space. You can always add sheets, categories, or variables that expand your model later on in the model building process. Experience also dictates the usefulness of adding plenty of policy variables that can be changed to generate alternative solution scenarios.
So when design a model using VORSIM, you have to think of the structure and the equations you want. Then these decisions have to be translated into choices of model codes during the model creation operation. Although this seems complicated, these kinds of choices have to be made in any modeling activity. The model examples on the VORSIM CD are helpful in this regard.