STEM newsletter

How do scenarios work?

30 October 2003

A business model is not an end in itself but a means to simulate the economic performance of investments. However well calibrated, results are inevitably sensitive to uncertainties affecting future parameters including competition, demand, cost and technology trends. The real value of a model is as a bespoke platform for exploring differing outcomes for a given network solution, and STEM’s integrated scenario engine manages the process of comparing a consistent set of results across multiple scenarios.

Varying an input across scenarios

Consider a simple wireless local loop model where both market size and network scale are driven directly by the deployment of base stations. This is typically modelled with a Location element, representing the base-station roll-out plan, the size of which will be critical driver for both revenue and cost outputs.

Market size and network scale driven by the deployment of base stations

In order to introduce scenarios for the first time:

  • Select the input you wish to modify, in this case the Location Sites input.
  • Select Create Scenarios… from the dialog Variants menu.

Creating scenarios for the selected input

STEM displays a tabular dialog with three columns for each of the newly created scenarios, the first row of which is labelled to match the original input, with the same value copied into each column.

Newly created variant data table

By default, the three scenarios are labelled ‘Variant 1’, ‘Variant 2’ and ‘Variant 3’, and linked via ‘Dimension 1’ (which is explained later). These elements appear as icons and can be renamed in the usual way, as shown below.

Alternatively, if you de-select Automatic Names from the Options menu, then you would be prompted for these names as the Variants are created, and before the Variant Data table is displayed. You would press if perhaps only two Variants were required. Additional Variants can be created from the toolbar, or from the Edit menu in the Variant Data dialog.

In this example, we shall name the Variants ‘Slow’, ‘Moderate’ and ‘Fast’, and then enter roll-out periods of ten, five and two years respectively.

Slow, Moderate and Fast roll-out compared

This simple structure provides a concise description of the scenario space, and the full results set can generated with a single command in the Editor:

  • Select Scenarios… from the File menu. The Scenarios dialog is displayed.
  • Click Save and Run All.

Running scenarios

The appropriate Variant Data are pasted into three parallel, generated copies of the model which are immediately run in sequence. The Results program then offers a selection from these three scenarios when you draw any graph so that, for example, you can compare the NPV between the three scenarios.

Comparing NPV across multiple scenarios

Varying another input with the first

The Slow roll-out is easy on investment but weak on customer acquisition. In order to address this shortcoming, the operator might offer an introductory tariff discount for the early years, but only in the Slow scenario.

The same procedure is used to identify Service Rental Tariff as a parameter for these scenarios, and this time STEM asks you to choose a Dimension for this new parameter. We select the existing Dimension 1 because the tariff is correlated with the roll-out, and so the tariff input is added to the second row of the existing Variant Data table.

Adding an extra Dimension parameter

Varying uncorrelated inputs

Another angle you might wish to explore would be sensitivity to variations in cost, independent of the Roll Out scenarios described above. When you create these scenarios, select, so that the cost variants become the first row of a new table.

Now you can see the significance of the Dimension element as a descriptor for each correlated set of input variants. Thus we might rename Dimension 1 as ‘Roll Out’ and Dimension 2 as ‘Cost’, with new variants ‘Low’, ‘Medium’ and ‘High’.

When you come to run scenarios, STEM will have generated 3 × 3 = 9 possible combinations, with labels such as ‘Moderate / Medium’. You can maintain a selection of the particular scenarios of most frequent interest for comparison, the results for which can be generated with the Run Selection button.

Running a selection of scenarios

Technology variants

One common question is whether different structures can be compared with scenarios. The answer is a firm yes! The standard approach is to use separate Multiplier Transformations to drive the different technologies, and then to create ‘one/zero’ scenarios for the multiplier inputs in order to route demand to the appropriate elements.

Comparing technologies using scenarios

Note: if there are just two technologies, it is more concise to drive only the first multiplier as a variant, with an invariant complementary formula for the second multiplier.

Behind the scenes

Of course the Dimension element is more than just a label. As well as grouping the relevant Variant elements, it also identifies the respective input parameters which will be replaced with Variant Data when scenarios are generated. Dimension parameters can be accessed directly from the Dimension icon menu:

The Dimension Parameters dialog

If you prefer, additional parameters can be defined by editing one of these parameters and then clicking on the desired input – see Internal references and reference mode in the Essential STEM Editor Skills section of the STEM User Guide. The original value of this input will be copied to all variants for the new parameter.

Parameter references can be removed from the Parameters dialog, or more commonly from the original input data dialog, by selecting Remove from the dialog Variants menu.

© Implied Logic Limited