Counting Employees

[mattgoff]

Due to some bad element naming in 2006, our "FTE" measure is actually "Headcount." We don't have a true FTE measure, leading to overstated heads in some situations, e.g. a termination and replacement in the same month, transfer between two depts, and some others. I've been asked to try to come up with a new measure to correct this; here are the options I've brainstormed so far. I'd love to hear how others approach this problem.

Fractional FTE (kludgey name since "FTE" already exists as something else)
Measure % of month worked. So,

• if someone's first day worked was August 5, fFTE = (31-5+1)/31 = .8701
• if someone's last day worked was August 5, fFTE = 5/31 = .1613

Pros: Useful elsewhere in TM1 (will probably create this either way, already used internally in TM1 to calculate start/term month salary).
Cons: Fractions abound.

Half Month Test
If person works > 1/2 the month, they could as a head.

• if someone's first day worked was August 5, adjFTE = 1
• if someone's first day worked was August 20, adjFTE = 0
• if someone's last day worked was August 5, adjFTE = 0
• if someone's last day worked was August 20, adjFTE = 1

Pros: Sticks to integers.
Cons: More likely to understate count, e.g. term Aug 14, start Aug 16 = 0 heads

Full Month for Starters
Only count as a head once a new starter works a full month.

• if someone's first day worked was August 5, adjFTE = 0 (in Aug)
• if someone's first day worked was August 20, adjFTE = 0 (in Aug)
• if someone's last day worked was August 5, adjFTE = 1
• if someone's last day worked was August 20, adjFTE = 1

Pros: Sticks to integers.
Cons: Some built-in error if term/replacement span two months. e.g. term Aug 1, replacement start Sep 1, Aug shows a head but Sep does not. Lags actual headcount.

Full Month for Terminations
Don't count as a head in the month an employee terminates

• if someone's first day worked was August 5, adjFTE = 1 (in Aug)
• if someone's first day worked was August 20, adjFTE = 1 (in Aug)
• if someone's last day worked was August 5, adjFTE = 0 (in Aug)
• if someone's last day worked was August 20, adjFTE = 0 (in Aug)

Pros: Sticks to integers.
Cons: Some built-in error if replacements span two months. e.g. term Aug 31, replacement start Sep 1, Aug doesn't shows a full head but Sep does not.

[tomok]

I must admit I've never seen FTE defined as you are doing it. FTE, at least as far as I have always seen it, is for coming up with a way to combine full time employees with part-time employees to come up with a total. In other words, people that work the normal work week (40 hours or whatever) with those that work a fraction of that. I've never seen FTE defined in terms of when a person started or were terminated/transferred. Having said that, I wouldn't make it so complicated. If you have the data I would say anyone that started before the 15th be counted as a 1 and anyone that started after would be a 0.

[mattgoff]

I must admit I've never seen FTE defined as you are doing it. FTE, at least as far as I have always seen it, is for coming up with a way to combine full time employees with part-time employees to come up with a total. In other words, people that work the normal work week (40 hours or whatever) with those that work a fraction of that.

Agreed, we're doing that too, but it's pretty straightforward since we have the number of hours scheduled coming in from HRIS.

This issue (how to treat people who start and terminate) came up at the same time. We have some areas (e.g. sales account coordinators) who turnover relatively quickly and who are also replaced relatively quickly. There was a situation where a bunch of ACs (not sure of the number, let's say 5) left and were replaced in the same month, and FTE (really headcount) looked really strange for that department in that month as a result. Just looking at the cube, FTE increased by five for one month and then returned to normal. So, it was asked if there was a way to develop another measure of FTE that would avoid that sort of distortion.

Matt

[Martin Ryan]

I've see both ways, but agree tomok's is more common. Depends on what your dimension is. If it's employee, as in specific person, then I think the fractional method makes the most sense. If it's position then you either have it or your don't, in which case the simplest integer method makes the most sense.

One way I've seen it done recently is similar to your fractional method
FTE: as in tomok's description. I.e. hours per week worked / 40
Start date: <some date>
End Date: <blank, or some date>
Adjusted FTE: weights the FTE figure if the user has started or finished during the month

If I was reading a report I think this is what I would expect to be seeing.

Martin