From the above simulations, you could conclude that on an income basis the investment looks attractive, while on a cash flow basis it does not. We have, however, built a model that contains quite a few assumptions, and these assumptions are known to be uncertain. We could go in and change the assumptions one at a time and simulate the model to understand the implications. There is, however, an alternative to this, known as Monte-Carlo Simulation or Multivariate Sensitivity Simulation (MVSS). To use this, we will set ranges on the uncertain assumptions, then Vensim will simulate the model multiple times randomly selecting values for the uncertain assumptions.
Open the Simulation Control and create a sensitivity control file (e.g., financ01.vsc). Select the following parameters, distributions, and ranges (Chapter 15 of the User’s Guide has details on setting up and running Sensitivity Simulations):
average payable delay = RANDOM_UNIFORM(.07,.11)
billing processing time = RANDOM_UNIFORM(.03,.05)
building time = RANDOM_UNIFORM(.8,1.2)
fractional loss rate = RANDOM_UNIFORM(.05,.08)
interest rate = RANDOM_UNIFORM(.09,.15)
price = RANDOM_UNIFORM(.9,1.2)
production capacity = RANDOM_UNIFORM(2200,2600)
required investment = RANDOM_UNIFORM(1800,2200)
variable production cost = RANDOM_UNIFORM(.5,.7)
Be sure to choose the Multivariate option and set the number of simulations to 200.
In order to do a sensitivity analysis you will need to select a set of variables for which to view the results. Though Vensim normally stores all the results, this is not practical in the case of sensitivity analysis, and so we need to specify a shorter list. Chapter 15 of the User’s Guide has information on setting up save lists. For this model it is useful to save (financ01.lst) :
npv cash flow
npv income
net cash flow
net income
Now run the simulation. Vensim first does a normal simulation, and then the multiple sensitivity simulations. Once the sensitivity simulations start you will see a window showing each simulation as it is finished. This will typically run very fast, but if it is running slowly, you can stop before Vensim completes all 200 simulations by clicking on the Cancel button. The results, if you do stop early, are likely to be a little different from those reported here.
