Financial Planning Analysis

Six Sigma – iSixSigma Forums General Forums Methodology Financial Planning Analysis

Viewing 6 posts - 1 through 6 (of 6 total)
• Author
Posts
• #53742

Heeb
Participant

I’m working on examining financial planning deviance, planned spend vs actual spend, in a large company. I have questions around measurements and applicable tools.

We have per-project metrics on monthly/yearly plans and actuals. Yearly plans are adjusted every quarter. Plans vary greatly in value, from thousands to millions of dollars.

1. I’m looking for a good indicator for planning effectiveness. I can calculate % deviance and \$ deviance at a monthly/YTD level. A good indicator must take both into account. I’m not sure if I can just do (% deviation) * (\$ deviation), and I’m unsure what sort of statistical tools I could use on it after that transformation.

2. What sort of statistical tools can be leveraged in this situation? So far I’ve mainly been using the test of equal variance.

Thanks!

0
#191286

MBBinWI
Participant

The problem with looking strictly at deviance from plan is that if you under estimate, you have a problem and have to ask for more money (so nobody under estimates the project) and if you over estimate and are being evaluated on the deviation from the plan, suddenly any overage finds a way to be spent.

Instead of trying to track deviation from plan, I would suggest that you track payback period. With that, you encourage thrift in project fund expenditures, yet still support doing the right thing for the customer (otherwise, sales will flounder and the payback will take longer).

As for statistical tools, what question are you looking to answer? Not sure how a test of equal variances would assist what you described.

0
#191288

Heeb
Participant

Hi MBBinWi,

This financial planning aspect is new to me. My experience is primarily in production, so I appreciate your input.

MBBinWI wrote:

The problem with looking strictly at deviance from plan is that if you under estimate, you have a problem and have to ask for more money (so nobody under estimates the project) and if you over estimate and are being evaluated on the deviation from the plan, suddenly any overage finds a way to be spent.

Surprisingly, they have a lot of variation, particularly at the month level. Even at the year level, there are spends between 2%-1300% of plan. Of course these extreme deviations occur on smaller projects, so looking only at % deviation is misleading. Much of this variation cancels out, so when you look at the aggregate result it’s not so bad. I’ll be meeting with the sponsor more to understand what good should look like.

As for statistical tools, what question are you looking to answer? Not sure how a test of equal variances would assist what you described.

My data contain items such as project/program manager, labour/material spend, region, etc. I was hoping to identify if any of these xs are significant.

Unfortunately it’s non-normal. Tests of equal variance on \$/% deviation show, for example, that most PMs have very similar variance. To date I’ve visually compared \$ vs % results to identify areas of interest, but I wish I could get a good indicator.

One additional aspect I was planning to look at was how significantly yearly plans are altered each quarter, although there could easily be business/strategic reasons for those changes.

Instead of trying to track deviation from plan, I would suggest that you track payback period. With that, you encourage thrift in project fund expenditures, yet still support doing the right thing for the customer (otherwise, sales will flounder and the payback will take longer).

Thanks. I will look into payback period, though my sponsor is mostly interested in improving the reliability of the forecasts. I’m not sure how concrete payback period would be as many of these projects focus on internal capabilities and sustainment.

0
#191294

Donald Rybarczyk
Participant

After reading your post I cant help but think this is not a statistical problem but rather a project management problem. I’d suggest using Earned Value Management.

Earned value management (EVM) is a method for evaluating project performance. It relies on three primary measurements to calculate a snapshot or to date cost and schedule performance index. Furthermore, it enables users to forecast the same indexes at the end of the project.

The three primary EVM measurements are
1) Planned value (PV), or the value of the work to be completed. PV answers the question How much work did you plan to have done by now? To keep things simple, lets say that the configuration phase of a software implementation project requires 1,000 hours of work at a blended cost of \$100 per hour, for a total PV of \$100,000.

2) Earned value (EV), or the value of the work actually completed to date. EV answers the question How much work is actually done? If only 80% of the tasks required to complete the configuration phase are complete, the EV is \$80,000.

3) Actual cost (AC), or the actual cost incurred to date on the particular tasks being measured. AC answers the question What is the cost of the work you have already completed? For this example, lets estimate that the AC is \$90,000.

After gathering the data, two critical performance indexes can be established:

1) Cost performance index (CPI). The CPI gives a project manager an indication of whether he is ahead of or behind budget as it relates to cost. To calculate the CPI, divide the EV (earned value) by the AC (actual cost). In the example, the CPI is \$80,000 divided by \$90,000, which results in a CPI of .88.

2) Schedule performance index (SPI). The SPI gives a project manager an indication of whether he is ahead of or behind schedule as it relates to work actually completed. To calculate the SPI, divide the EV (earned value) by the PV (planned value). In the example, the SPI is \$80,000 divided by \$100,000, which results in an SPI of .8.

What do the numbers mean?
When the CPI and the SPI equal 1.0, the budget and work schedule are on plan. If the index is greater than 1, the project is moving ahead of plan. In the case of the CPI, the project is costing less than planned, and in the case of the SPI, the project team is accomplishing more work than was planned to be accomplished to date.

In instances when the indexes are less than 1, the project can be categorized as below plan, meaning that it costs more and that the team is accomplishing less work than was planned to be accomplished to date. Using the example above, where the CPI is .88 and the SPI is .8, the project is costing more and taking more time than planned. Based on this information, the projects manager, executive sponsor, and steering committee should put together a plan that catches up the work not yet completed, containing costs or readjusting the budget or schedule budget accordingly. In addition, the project manager can estimate what it will take  in time and dollars  to complete the project, based on the indexes to date and by making assumptions about the project going forward.

Hope this helps.. and i didnt misunderstand your question :)

0
#191331

MBBinWI
Participant

Hey, rybarczyk: I’m afraid that EVA suffers from the same issues that I offered up in my first response.

0
#191333

MBBinWI
Participant

Sorry that it’s been awhile. What you have currently is an open-loop system. An estimate is created (out of thin air?). Being under estimated has dire consequences so I still bet that few projects are underestimated. So, if they are over-estimated and you track deviation from estimate, and nobody wants to look bad, then the estimate will be spent, whether truly needed or not.

Hopefully you can still help your sponsor to understand that he/she is trying to control an open loop system with closed loop methods. You either need to change the system controls to closed loop form (payback period – where the time to recover the investment is tracked) or find open loop control methods (unfortunately in this case, that’s more of the sort that you fire the people who are deemed not to perform adequately).

0
Viewing 6 posts - 1 through 6 (of 6 total)

You must be logged in to reply to this topic.