Week 10

• A decision tree is a diagram that illustrates all alternatives under all possible conditions

  It helps work out the Expected Value in more complex situations
• Expected Net Present Values (ENPVs) have to be used when events and payoffs occur over a number of years
• Sometimes managers can estimate what the possible outcomes for the future can be, but have no idea of what the probabilities of each of these outcomes is

  We have two techniques that help with this:

  Sensitivity (what if?) analysis

  Maximin/Maximax/Regret  decision rule criteria
• Sensitivity analysis

  Answers ‘what if’ questions, e.g. what will be the change in profit if sales decline by 1%?; which variables have the biggest impact on profit?
• Maximin: Assumes that the worst possible outcome will occur. Therefore the largest payoff under this assumption should be chosen.  FOR PESSIMISTS!
• Maximax: Assumes that the best possible outcome will occur. Therefore the largest payoff under this assumption should be chosen. FOR OPTIMISTS!
• Regret: Assumes that, if the chosen alternative does not prove to be the best decision, the decision-maker will regret not having chosen another alternative. The amount of regret is the difference between the payoff of the chosen alternative and the payoff of the best alternative. The decision rule of a commonly used regret criterion is to minimise the  maximum possible regret (known as minimax regret).
• Generally businesses have a decision to make, and will identify various outcomes of their decision (how much profit they might make, for example)The outcomes of the decision will depend on factors outside their control, often described as ‘states of the world’ (or sometimes scenarios)
• You need to write out a matrix showing the outcomes of their decision under different states of the world (known a a ‘payoff’ matrix)
• There is a degree of risk and uncertainty in nearly all business decisions.
• We can use a range of techniques to help us deal with risk and uncertainty.
• Expected values are useful for repeated decisions/event. They are less useful for one-off decisions.
• Decision trees are a useful way of displaying complex situations with a range of alternatives and possible outcomes.
• If it is not possible to assign probabilities to possible outcomes, maximax, maximin or regret decision criteria can be used. Sensitivity analysis is also a useful tool.