Trends for cost and fixed price contracts

[vibhuti6]

Filters used:

• Defense contracts (agency code = 97)
• Small businessses only
• Exclude disadvantaged small businesses
• Fixed price contracts: type_of_contract_pricing_code in (A, B, J, K, L, M)
• Cost price contracts: type_of_contract_pricing_code in (R, S, T, U, V)
• Exclude bundled contracts
  • Multiple contracts are sometimes assigned the same task/budget
• Exclude disadvantaged small businesses
• Top level product or service codes used: B5, C1, C2, F0, F1, F9, H1, H2, H3, H9, J0, J9, K0, N0, Y1, Z1, Z2
  • They broadly correspond to infrastructure/construction contracts
  • Easy to characterize the change in deadline as a "delay" as compared to other procurement contracts, e.g., contracts for acquiring a specific product
• Fiscal years = [2008, 2009, 2010, 2011]
• Product codes showing similar trends from FY 2008 - FY 2010: H2, B5, F9
  • Note: Plots are based on first two letters of a four-digit product code -- there may be more similarity if we look at the entire 4-digit codes
• Example: Delays in fiscal year 2008 = End date as of Sept 30, 2008 - End date as of Oct 1, 2007

Past Trends	Infrastructure contracts

[vob2]

Would it be too much work to look at the project delays as % of the originally forecasted completion date?

Would it be possible to present some measure of variability around these averages. E.g., confidence bands or box plots? The variation for some of the graphs is really huge and I wonder if this is due to a few outliers.

[vibhuti6]

Hi Vlad, I will try to add the measures that you suggested, and get back to you on this issue. Thanks.

[vibhuti6]

Hi Vlad, I believe the large variation in these graphs comes from very few contracts being assigned cost pricing instead of fixed price for the same task. See the graph that I plotted for all the 4-letter PSC codes in this sample.

I also obtained the box plots as below. Note that the following graphs only take into account first two letters of a PSC code.

[vibhuti6]

And here are the graphs for percentage delays.

• Percentage delay = 100*(Eventual completion time - Initial completion time)/(Initial completion time)
  • Eventual completion time = Last reported end date - First reported start date
  • Initial completion time = First reported end date - First reported start date

Percentage delay	Infrastructure contracts

[vob2]

Hi Vlad, I believe the large variation in these graphs comes from very few contracts being assigned cost pricing instead of fixed price for the same task. See the graph that I plotted for all the 4-letter PSC codes in this sample.

I also obtained the box plots as below. Note that the following graphs only take into account first two letters of a PSC code.

Thanks, Vibhuti! I see. So, if I understood correctly, for a particular PSC code the use of contracts is imbalanced. It is either primarily fixed price or primarily cost plus and rarely both. Then, either we would have to aggregate across PSC codes, which is probably not going to work or not use fixed vs cost plus as the differentiation between treatment and control. I guess this is more ammunition for looking at small vs non-small business instead as we already decided to do.

[vibhuti6]

Yes, that's correct. Also, note that in these graphs I have aggregated different types of fixed price and cost contracts into one group each. They each have several categories, see the screenshot below.

So my sense is that it may be more difficult to find a suitable match if we look closely. For example, finding a contract under "fixed price award fee" and another under "cost plus award fee" for the same PSC code.