The Community Blog for Business Analysts

HSantanam
HSantanam

Project Risk Management: Estimating Techniques

EMV process is used in the Decision Tree analysis, which visually maps out activity decision paths. As an example, if a deliverable for a project has two suppliers, and it is known that any delay in the crucial deliverable will result in additional resource idling/alternative costs of $1,000 per day, and that supplier "A" price is $20,000 and supplier "B" total price is $22,000, and that using "A" has a 10% risk of being 3 days late and that "B" has a 5% chance of being 2 days late, then the Decision tree would look like this:

Supplier Selection 

--- {Supplier A

---{Risk of being late: .10 x $1000 x 3 = 300

---{Total EMV: $20,000 + 300 = $20,300 

--- {Supplier B

---{Risk of being late: .05 x $1000 x 2 = 100

---{Total EMV $22,100

Sorry, the boxes didn't copy over from my MS-OneNote - but I think you get the concept.

As you can see, Decision trees are very numerical, but in real life, numbers don't tell the whole story -  perhaps supplier "A" has other benefits such as better support (hopefully that has been factored into the overall supplier risk rating).

 

That's all for now folks - hope you found this primer on risk management methods to be useful..or at least interesting.

 

 

Business projects carry risk - for time, cost, scope and quality - but they can all be expressed in monetary terms, for in business everything is Dollars (or Euros or Yen or other currency). Project managers and risk managers for businesses need to quantify risks and estimate costs for budgets, forecasts and management reporting.  How do they do that? Here are some ways:

 

SWOT (Strengths, Weaknesses, Opportunities and Threats) analysis is often used to describe at a high level the risks faced by a project. Strengths and Weaknesses deal generally deal with internal and project specific areas and Opportunities and Threats are more external or market facing environmental issues.

 

The risks are detailed in the Risk Management knowledge area of the PMBOK guide from PMI.

 

The risk register is a list of all the major threats to a project, and is an output of the Identify Risks process, and is then used as an input to other processes to further qualify, quantify and assess threats.

 

Qualitative Risk and Quantitative Risk assessment methods are used to assess the priority, urgency and impact of risks; as with many other things, some subjectivity is involved in deciding probability of risks.  One of the tools and techniques for performing Qualitative Risk analysis is the Probability-Impact matrix, which rates the possible threats on their likelihood of occurring and then the impact if they did happen.  Quantitative analysis then assigns monetary and impact measurements to the threats.

 

The PERT or three-point estimate is widely used and works this way:

3 likely scenarios - Pessimistic, Optimistic and Most Likely estimates - are gathered from experts, and then a weighted average is used:

 

(Pessimistic Estimate + (4 x Most Likely Estimate) + Optimistic Estimate)/ 6

Note: the 4 for the Most Likely Estimate is to give it a 4x weight in the weighted average - hence the term - as per the other two.

Example: Estimates for a software application range from 10 weeks(Pessimistic) to 7 weeks (most likely) to 5 weeks (Optimistic).

So the three-point estimate would be: (10 + (4x7) + 5)/6  => (10+28+5)/6 => 43/6 => 7.15 weeks.

 

Monte Carlo simulation is a complex modeling technique, and not one that many Business analysts and PMs are likely to have to work through.

 

Sensitivity Analysis looks at various project objectives and  measures how uncertainty would impact each objective. It is also known as a Tornado diagram due to its funnel like shape.

 

Expected Monetary Value (EMV) is used to capture cost and benefit of an uncertain outcome, based on statistical probability. As an example, if  a person has to pick a winning athlete in a four way race (and assuming all 4 are more or less equal in their abilities), and it costs $1 to bet and a $2 payoff (if he guesses correctly), then EMV for this outcome could be expressed thus:

 

1 in 4 chance of winning (25% or .25), 3 of losing (75% or .75):

(.75 x 0) + (.25 x 2) =  0 + .5 = 0.5 (50 cents)

 

Net EMV =  -1 + .5 = -0.5 (the -1 is the cost of the bet).

A negative EMV is a risk and a positive EMV is a benefit. However, don't use this as a reason to blow your savings at the casino :).


This entry was published on Nov 17, 2010 / HSantanam. Posted in Project Management, Estimation. Bookmark the Permalink or E-mail it to a friend.
Like this article:
  4 members liked this article

Related Articles

COMMENTS

Marc Thibault posted on Friday, November 19, 2010 9:56 PM
The decision tree analysis example is unrealistic. Each of the vendors also have some probability of being 0,1, 2, 4, etc days late. The decision parameters will be dead wrong if all of the possibilities aren't taken into account.

PERT and CPM formulations suffer from the flaw of averages and give ridiculously optimistic estimates.

Tornado sensitivity analysis gives you the sensitivity of a chosen variable that's valid only when all of the other variables are at their expected values. The real world isn't that simple. Proper sensitivity analysis will wiggle all the variables, to model real-world possibilities and probabilities, and give you a scatterplot of the chosen variable against the result under all circumstances.

Monte Carlo Simulation isn't really complex--just more laborious and precise. It seems complex because you use vectors instead of scalars, but the computer does the work. It's actually simpler than PERT. --And a great deal more realistic; it's the only way to get estimates that reflect the uncertainty in the model inputs. The high rate of failure in our projects is not the fault of incompetent PMs; it's due to estimating techniques that give wrong numbers that are flat-out not achievable.

The one useful thing about PERT is that it emits numbers that are more likely to get the project approved than would a realistic estimate. If having a project is more important than finishing it, you may want to avoid Monte Carlo techniques.

Marc Thibault
HSantanam posted on Saturday, November 20, 2010 9:57 AM

Absolutely right...I was only illustrating the use of these widely used estimating techniques. An issues that I have with these techniques is that in the end, some of the premises and basis used in these calculations are experts' opinions, and can easily be challenged - for example, who's to say that vendor "A" has a 10% probability of being late instead of say, 15% (unless the experts actually knew from prior experience that vendor A was always 10% late).
Same thing with the probabilty-impact matrix - there is some subjectivity and guessing built into it, which is the way most human activities work, I think... but maybe Monte Calro analysis might show more accurate results.
I like your last line of the comment - "If having a project is more important than finishing it, may want to avoid Monte Carlo techniques" - true!
HSantanam
Marc Thibault posted on Saturday, November 20, 2010 2:46 PM

Human estimators can be trained to produce good estimates. I recommend reading Douglas Hubbard.

Where there isn't enough history to sample, an expert estimate of the probability distribution is a great deal better than throwing some nominal curve at it, or worst: a single-number best guess.

You're right that these estimates can be challenged, but only by improving on the assumptions that go into the estimate--by getting more or better measurements. Estimating is a process that can be documented and validated.
Marc Thibault
Only registered users may post comments.

Modern Analyst Blog Latests

As we start a new year many of us will take the time to reflect on our accomplishments from 2012 and plan our goals for 2013. We can set small or large goals. goals that will be accomplished quickly or could take several years. For 2013, I think Business Analysts should look to go beyond our traditional boundaries and set audacious goals. Merriam-...
Recently, I was asked by the IIBA to present a talk at one of their chapter meetings. I am reprinting here my response to that invitation in the hope that it will begin a conversation with fellow EEPs and BAs about an area of great concern to the profession. Hi xx …. Regarding the IIBA talk, there is another issue that I am considering. It's p...
Continuing the ABC series for Business Analysts, Howard Podeswa created the next installment titled "BA ABCs: “C” is for Class Diagram" as an article rather than a blog post. You can find the article here: BA ABCs: “C” is for Class Diagram Here are the previous two posts: BA ABCs: “A” is for Activity Diagram BA ABCs: “B” is for BPMN

 



Blog Information

» What is the Community Blog and what are the Benefits of Contributing?

» Review our Blog Posting Guidelines.

» I am looking for the original Modern Analyst blog posts.

 




Copyright 2006-2024 by Modern Analyst Media LLC