Week 2

Question 1

Cost estimation

Answer 1

The development of a relationship between a cost object and its cost driver to predict the cost.

1) Helps to predict future costs
2) Identify the key cost drivers
3) Cost drivers and cost-estimating relationships are useful in planning and decision-making

Question 2

Six steps of cost estimation:

Answer 2

1) Define cost objects
2) Define cost drivers
3) collect consistent methodology and accurate data
4) graph the data
5) select and employ estimation model
6) assess the accuracy of the cost estimate using for example absolute mean percentage error (MAPE)

Question 3

High-low method:

Answer 3

easy to apply, but less accurate than regression analysis. Uses algebra to determine a unique estimation line between high and low points in data.

The advantage is that it requires the accountant to prepare and study a graph of the data, so that outliers are eliminated. Otherwise the analysis is sensitive to outliers if those are at the both ends.

Question 4

Regression analysis:

Answer 4

a statistical method for obtaining the unique cost-estimating equation, that best fits a set of data (minimizing squared estimation errors). Also called the least squares regression method.

Dependent variable - cost to be estimated
Independent variable - cost driver

Evaluating a regression analysis:
R-squared - a number 0-1, which measures how much a change in the DV can be attributed to a change in IV
t-value - a value that measures the degree to which each IV has a stable relationship with DV, must be >2
Standard error of estimate - a measure of how dispersed actual observations are around the calculated regression line, providing a measure of accuracy. This allows to measure CI, as 1 SE falls in 67% CI, 2 SE falls in 95% CI
p-value - a measure of the risk that changes in the DV associated with changes in IV are only a result of change. Must be <0.05

Question 5

Time-series and cross-sectional regression:

Answer 5

Time-series regression - is the application of regression analysis to predict future amounts, using prior period’s data.

Cross-sectional regression - estimates cost for cost object based on info on another cost object and variables, where the info for variables is taken from the same time.

Question 6

Implementation problems (nonlinearity):

Answer 6

Linear regression can be adapted to deal with nonlinearity:
1) Seasonality: most common methods are:
Use of a price change index to adjust values of each variable to some common time
Use of a trend variable which takes on the values 1,2,3 for each period
Replacement of the original values of each of the variables with the first difference.

2) Outliers: which may be resolved using a dummy variable

3) Data shift: which can be adjusted using a dummy variable to indicate the periods before and after the shift.

Question 7

Learning curve analysis:

Answer 7

A systematic estimation of costs when learning and improvement in production is present. The associated learning rate is the percentage by which the average or total production time decreases as output doubles.

Y = a*X^b
a - time required for 1 unit
X - cumulative output
b - learning index = ln(learning rate)/ln(2)

Limitations:
1) More relevant for labour-intensive contexts that involve repetitive tasks
2) Learning rate is assumed to be constant
3) The estimated learning curve could be unreliable, as the observed change in productivity in the data could be associated with different factors.