Earned value calculation vs work contour

[Jan M.]

Hi,

I've been experiencing some problem with MS Project calculation of earned
value. For certain tasks of my project, there was a discrepency between BCWP
as shown in the earned value table and BCWP in the timescale. Whenever I
exported data to Excel to graph CPI and SPI, values in the graph didn't match
the values of the earned value table, which is quite annoying.

I thought the problem was due to my using of physical % completed to
calculate earned value. It took me some time to realize that the problem was
caused by the tasks to which I had applied a work contour! So I made a very
simple project to look further into that. Here are the results of my
simulation:

Project description:
1 task
40 hours of work
1 ressource at 1$/h assigned to the task at 100%
Calendar 8h/day, 5 days/week
Work contour applied: Back loaded
Duration as calculated by MS Project: 8.33d
Earned value method of calculation: % complete.

After saving the baseline, I applied the task usage view and inserted the
columns "% complete", "% work complete" and "physical % complete". What I
first noticed is that all "% fields" are tasks fields but only "% work
complete" is an assignment field (Thanks to Julie S. for the hint!).

I entered 4 hours of actual work on the first day(status date on the same
day): % complete as calculated by MS Project is 13%. On the assignment line,
BCWP reads 4$ (table and timesheet), but on the task line, BCWP reads only
0.93$ (table and timesheet). Which means that MS Project uses % work complete
to calculate earned value for assignments instead of % complete as it does
for tasks. The only reason I can come up with to explain 0.93$ of earned
value out of 40$ with 13% complete, is the distribution of work that is
skewed to the right due to the work contour. Having completed 10% of the
work, I would have prefered BCWP to be 4$ as shown on the assignment line.

Then, I made another simulation: I changed the earned value method of
calculation to physical % complete and copied % work complete (10%) in
physical % complete. As with % complete, BCWP reads 4$ on the assignment line
( table and timesheet). But now, BCWP reads 4$ on the task line in the table
and 0.67$ in the timesheet.

So, the discrepency I experienced in the above mentionned project was caused
by my using of work contour with physical % complete. To make sure I was
right, I made a simulation the same simulation without any work contour:

1) % complete as a method of calculation: BCWP is 4$ on the assignment line
(table and timesheet) and 7.27$ on the task line (table and timesheet). The
difference is "normal" since MS Project uses % work complete for BCWP
assignment and % complete for BCWP task. At least there is no discrepency
between earned value table and timesheet: values exported to Excel will match
the table in MS Project.

2) physical % complete as a method of calculation: BCWP is 4$ everywhere.

My recommendations:

1)Unless someone can prove me I got it all wrong, I advise against using
work contour for projects as far as earned value is concerned: earned value
calculation is biased by work contour.

2)I strongly suggest that Microsoft provides a method for calculating earned
value based on % work complete:value is earned as work is accomplished, not
as time passes by.

3)The same method of calculation should be used for assignments and tasks.

Thanks for going through all this.

Any comment someone?

Jan M.

 

[Steve House [MVP]]

I beg to differ on your point #2. Value is earned by accomplishing one's
goals in the time frame you have allotted to it. Percent Complete measures
duration, true. But it is not the simple passage of time that figures into
earned value. You're forgetting the status date. We have a task that is
slated for 10 days assigned to Joe Resource (who gets $1 per hour). It's
now at the end of day 5 and Joe has worked all 5 days. We are 50% complete.
As of this moment, BCWS = $40, BCWP=$40, and SPI = 1.0. But take that same
task, baseline duration 10 days. We are at the end of day 5 but Joe has
only worked 3 out of the 5 days we had scheduled him for, called in sick
perhaps. We are NOT 50% Complete but instead only 30% complete. BCWS is
$40, BCWP=$24, and SPI=0.60.

 

[Jan M.]

Hi Steve,

I agree with you except for one thing.

In your example, Joe only worked 3 days out of 5. Let’s suppose he was sick
on Tuesday and Wednesday. Actual work for those days would be 0 and MS
Project would reschedule 16 hours of work in the future. The new project end
date is scheduled to be 2 days later, but duration is still 10 days because
it’s based on working days, not elapsed days.

In your example(at the end of day 5):
% complete = 3 working days/10 days = 30%
% work complete = 24 h/ 80h = 30%
BCWS = 40
BCWP = 24
SPI = 0.6
CPI = 1

Those numbers are correct.

Now what if Joe was not sick but had something more important to do and only
managed to work the following hours on the project:
Monday:2h, Tuesday:3h, Wednesday:3h, Thursday:8h, Friday:8h.

Again, there are 24 hours that were performed during five days and the
project end date is still delayed by two days. But duration is now 12 days
instead of 10.

New values are:

% complete = 5 working days / 12 days = 42%
% work complete = 24h / 80h = 30%
BCWS = 40
BCWP = 33.73 (based on % complete)
SPI = 0.84
CPI = 1.41???


In both cases Joe managed to put in the same 24 h of work during the very
same 5 day period. Since we did not revise the remaining work, I can see no
reason for BCWP not to be the same at the end of day 5! How do you explain a
CPI of 1.41 in the second example, since Joe is not going to work less or
more hours than the 80 hours that have been planned?

For more fun, apply a back loaded work contour to the assignment and enter
the actual values as per example one: no work on Tuesday and Wednesday. Take
a look at BCWP, CPI and SPI and see how funny they have become!

Now try this: apply the Task usage view, insert % complete and % work
complete columns, then simulate the above examples. You will find that at the
assignment level, BCWP is calculated with % work complete. But at the task
level, BCWP is calculated with % complete. Every time % complete is different
from % work complete, assignment and task won’t get the same BCWP even with
only one resource assigned to the task. That doesn’t make any sense to me.

My point is that I would like MS Project to show consistency in the way
earned value is calculated. Personally, I do think that % work complete is a
better estimate than any tricky formulas based on duration (as demonstrated
by example 2) and I would like Microsoft to add this feature to the software.

Jan