Step2:Wechooseaprobabilitydistributionforeachofthe
variables.Thereareanumberofchoiceshere,rangingfrom
discreteprobabilitydistributions(probabilitiesareassignedto
specific outcomes) to continuousdistributions (the normal,
lognormal,orexponentialdistribution).Inmakingthischoice,
the following factors should be considered:
- Therangeoffeasibleoutcomesforthevariable(e.g.,
therevenuescannotbelessthanzero,rulingoutany
distributionthatrequiresthevariabletotakeonlarge
negative values, such as the normal distribution). - Theexperienceofthecompanyonthisvariable.Data
onavariable,suchasoperatingmarginshistorically,
mayhelpusdeterminethetypeofdistribution that
best describes it.
While no distribution will provide a perfect fit, the
distribution thatbestfits the data should be used.
Step 3:Next, theparametersofthedistribution chosen for
eachvariableareestimated.Thenumberofparameterswill
varyfromdistributiontodistribution;forinstance,themean
and the variance have to be estimated for the normal
distribution,whiletheuniformdistributionrequiresestimates
of the minimum and maximum values for the variable.
Step 4: One outcome isdrawn from each distribution;the
variableisassumedto takeonthatvalueforthatparticular
simulation.Tomaketheanalysisricher,wecanrepeatthis
processeachyearandallowforcorrelationacrossvariables
and across time.
4