As part of risk management, a project risk analysis is performed to assess the impact of risks and the probability of the risk occurrence. In this article we will look at the differences between qualitative and quantitative project risk analysis.
To see how project risk analysis fits within risk management, please refer to this article.
Qualitative Risk Analysis
Qualitative project risk analysis is a subjective look at risk. We could use historical data to make subjective estimates of risk level or rely on "expert" estimates.
Qualitative risk analysis is often used as the first step in risk analysis. It is often used to identify risks that should be further analyzed using quantitative methods.
Qualitative Risk Measures
There are many techniques for performing qualitative risk analysis, but what is at the heart of most of these techniques is assigning a likelihood of the risk occurring and the level of impact if the risk occurs.
Let's say our organization uses the following categories of likelihoods:
Let's also say that we use the following impact categories:
In a qualitative sense, the product of likelihood and impact results in a risk level. For example, an unlikely risk with a minor impact will result in a low risk level. A likely risk with a major impact will result in a high risk level. In the case of a rare risk with a catastrophic impact will result in a relatively high risk level, even though its likelihood is low.
Quantitative Risk Analysis
Quantitative project risk analysis quantifies risks. In other words, we are using numerical values to determine risk levels. Quantitative risk analysis is more time consuming because of the time to build models of the item at risk as well as generating inputs to the model.
When we refer to a model, it depends on the situation. Financial models are often created in a spreadsheet. In engineering projects, a model may also be a spreadsheet, but could be a CAD model, a physical prototype, or a test rig. For our discussion, we will limit ourselves to spreadsheets, but keep in mind that analysis could be done by other means.
Some Quantitative Methods
A model that uses single point estimates as inputs is used to calculate an outcome. Often the model is calculated using worst case, most likely, and best case scenarios to get a feel for the range of outcomes.
The limitation with analytic models is the limited information we gain. We can find the total range of outcomes, but not the probability of occurrence.
Decision trees are useful for enumerating potential outcomes in a graphical manner. Risk can be incorporated by the use of chance nodes that have outcomes based on probability. Decision trees are also useful where decisions must be made over time, depending on previous outcomes.
Each node in the tree has a cost or payoff, and nodes emanating from a chance node has a probability of occurrence. From this information we can calculate expected value or expected utility of the tree.
Monte Carlo Simulation
A stochastic model has random variable inputs. Each random variable is assumed to follow a probability distribution, and each time the model is calculated, each random variable is sampled from its underlying distribution.
The advantage of Monte Carlo simulation is that with enough iterations, effectively every outcome is simulated. Therefore, we see the frequency and impact of every outcome.