Cost Analysis

1
Cost Analysis

• May 21, 2001

2
Variable Costs
3
Variable Costs
4
Fixed Cost
5
Fixed Cost
6
Step Costs
7
Cost Behavior
With respect to a COST DRIVER
8
Be careful! There is a difference between COST
DRIVER and ALLOCATION BASIS.
9
Mixed Cost
10
Mixed Cost Example
11
Mixed Cost Example
Variable Portion
12
Mixed Cost Example
13
Mixed Cost Example
Y mX b
14
Mixed Cost Example
15
Mixed Cost Example
16
Mixed Cost Example
17
Cost Analysis
The art of identifying (and quantifying) the
behavior of a cost, with respect to a cost driver.
18
Cost Analysis Techniques

• High-Low
• ScatterPlot
• Regression
• Engineering
• Account Analysis

19
Cost Analysis Techniques

• High-Low
• ScatterPlot
• Regression
• Engineering
• Account Analysis

Use historical data
20
High-Low Analysis
21
High-Low Analysis
22
High-Low Analysis
23
High-Low Analysis
24
High-Low Analysis
RISE
25
High-Low Analysis
RUN
26
High-Low Analysis
Y mX b
RISE / RUN
27
High-Low Analysis
HIGH
LOW
28
High-Low Analysis
29
High-Low Analysis
Difference --------------- 794
RISE
30
High-Low Analysis
Difference -- 299 ------ 794
RISE
RUN
31
High-Low Analysis
Y mX b
RISE / RUN
32
High-Low Analysis
Y (794/299)X b
RISE / RUN
33
High-Low Analysis
Y (2.655)X b
RISE / RUN
34
High-Low Analysis
Y m X b 8084 2.655
(514) b 8084 1365 b
6719 b
35
High-Low Analysis
Y m X b 7290 2.655
(215) b 7290 571 b 6719
b
36
High-Low Analysis
Y m X b Cost 2.6555 (units)
6719
37
High-Low Analysis
38
High-Low Analysis
Cost 2.6555 (units) 6719
39
Scatterplot Analysis
40
Scatterplot Analysis
41
Scatterplot Analysis
Y mX 5000
42
Scatterplot Analysis
43
Scatterplot Analysis

• Y m X
  b
• m (276) 5000
• 6.00 m

44
Scatterplot Analysis
Cost 6.00 (units) 5000
45
Regression Analysis
46
Regression Analysis
X-coefficient 5.74 Intercept 5250 R-square
.57
47
Regression Analysis
Cost 5.74 (units) 5250
48
Problems with Regression
49
Problems with Regression
50
Problems with Regression
R-square 0.57
51
Problems with Regression
R-square 0.57
52
Main Points

• High-Low quick and dirty
• ScatterPlot indispensable
• Regression mathematical (but!)

53
RELEVANT RANGE
54
Relevant Range Problems
55
Relevant Range Problems
56
Multiple Regression
57
Problems with Multiple Regression

• Multi-collinearity
• Always reduces R-squared artificially
• Counter-intuitive results
• Descriptive vs. Predictive
• Hold-out sample technique

58
Engineering Analysis

• Misnomer
• Includes true engineering methods study
• Also includes contractual determinations etc.
• Does not rely on historical data
• Only method available if there is reason to
  believe the cost structure has changed

59
Hold Account Analysis until next time
60
Quick Review

• Cost Analysis --- Behavior --- Structure
• High-Low Analysis
• Scatterplot Analysis
• Regression Analysis
• Multiple Regression Analysis (!)
• Engineering Analysis
• Account Analysis ()