Professional Documents
Culture Documents
Management
Accountant
Part 1 : Financial Planning,
Performance and Control
4
Forecasting Techniques
• Regression Analysis:
– There are two main types of regression analysis:
• Simple regression analysis : which uses only one independent variable.
• Multiple regression analysis : which uses two or more independent variables.
– Regression analysis equations systematically reduce estimation errors and are therefore also
called least square regression.
– Regression analysis fits a line (the regression line) through data points—a line that minimizes
the difference between the line (prediction) and the data point (actual).
– The statistical formula that the regression is based on produces the least amount of error
between these two items.
120000
Annual Sales (Y)
100000
80000
Sales
60000
40000
20000
0
0 20 40 60 80 100 120
Marketing Budget (X) 6
Forecasting Techniques
𝑌 = 𝑎 + 𝑏𝑋
where:
• Y = annual sales; it is the dependent variable to be forecast.
• a = amount of Y when X = 0. This value is also called the Y intercept.
• b = slope of the line, also known as the regression coefficient.
It represents the “impact” X has on Y. For every 1 unit change in X.
• X = value for the independent variable (in this case, marketing costs) .
7
• Regression analysis also provides a number of objective benchmarks that allow users to evaluate
the reliability of the regression equation. Three common measures are :
R-squared, T-value, and standard error of estimate (SE).
– R-squared :
• R-squared (goodness of fit or coefficient of determination) is a value between 0 and 1 indicating the
degree to which changes in a dependent variable can be predicted by changes in independent
variables.
• It is the percentage of the variation in the dependent variable accounted for by the variability in the
independent variable.
• A regression with an R-squared value closer to 1 has more explanatory power than a regression with
an R-squared value closer to 0. Graphs of regressions.
• The R-squared value in previous example is 0.7127. This means that approximately 71.27% of the
variation in sales is accounted for by variation in marketing costs.
– T-value :
• T-value measures whether an independent variable (X) has a valid, long-term relationship
to a dependent variable. Generally.
• The T-value should be more than 2. A variable with a low T-value indicates little or no
statistically significant relationship between the independent and dependent variables,
and the variable should be removed from the regression because it can lead to
inaccurate forecasts.
– Standard Error of Estimate :
• The SE measures the dispersion around the regression line and allows the user to assess the accuracy
of the predictions, That is, a user can build a confidence interval around the estimate.
8
Forecasting Techniques
• Multiple Linear Regression :-
• In simple linear regression analysis was used to estimate the impact that the amount of money spent on
marketing would have on the company’s sales.
• Based on the results of the analysis, the R-squared is only 0.7127. That means only 71.27% of the variation in
sales can be explained by changes in marketing expenditures, and the remaining 28.73% (100% – 71.27%) is
explained by changes in other factors that are not included in the regression model.
• In forecasting sales, an organization needs to take into consideration not only its marketing efforts but other
factors, such as the economic conditions, its competitors’ actions, its pricing strategy, and so on.
• All of these other factors can be incorporated into a multiple regression model, where they become the
additional independent variables that can help to explain the other 28.73% in sales variation that cannot be
explained by marketing expenditures.
9
Forecasting Techniques
• Benefits and Shortcomings of Regression Analysis:
– Regression analysis gives management accountants an objective measure to use in evaluating the
precision and reliability of estimations.
– It is important to prepare a graph of the data prior to using regression analysis and to determine
whether any unusual data points, called outliers, are present.
– Regression analysis can be influenced strongly by outliers, which may result in an estimation line
that is not representative of most of the data.
– If present, each outlier should be reviewed to determine whether it is due to a data recording error,
a normal operating condition, or a unique and nonrecurring event.
– Regression analysis requires a collection of data points—preferably 30 or more—to be accurate.
– In the case of analyzing cost behavior, this linear assumption can be problematic if it is expected that
costs decline due to learning curves or when there are different relevant ranges of activity that
cause costs to shift.
– Regression analysis also assumes that past relationships between dependent and independent
variables will hold into the future.
– When using regression analysis as a forecasting tool, it is important to evaluate or make adjustments
for changes in the relationship between the variables over time. For example, clear knowledge that
one part of a variable cost element is increasing makes the b coefficient less reliable for future
periods.
– When using the results of a regression analysis to make any prediction, it is important to remember
that the dependent variables used for the prediction must fall within the range of the data set used
to establish the regression line.
10
Forecasting Techniques
• Time Series Analysis:
– Time series analysis is an regression analysis which time is used as independent variable.
– Four components combine to provide the overall pattern in a time series analysis :
Trend, Cycle, Seasonality, and irregular variation
(172:185:192)
𝑀𝑜𝑣𝑖𝑛𝑔 𝐴𝑣𝑒𝑟𝑎𝑔𝑒 = = $183 Million 11
3
Forecasting Techniques
Weighted Moving Average :
Assumes that the most recent data has more power of prediction than data that came before.
Each prediction has a weight as a fraction of a whole so that all the weighting equal 1.
Example : if the weight for the pervious data is : 0.2 & 0.3 & 0.5
Exponential Smoothing
Uses a weighted average of past times series, selecting only one weight – that of the most recent set
of data.
14
• Cumulative Average-Time Learning Model
• Calculates cumulative total time by multiplying the incremental unit by the cumulative average
time per unit.
15
Forecasting Techniques
• Benefits of Learning Curve Analysis :
– Companies use learning curves to make a number of decisions, including setting prices.
A company may set a price lower than the initial costs of production in order to gain
market share, based on the assumption that as learning increases, production costs will
decrease.
– Some companies use learning curve analysis when evaluating performance, because it
expects an individual’s productivity to increase with the individual’s learning curve.
– Learning affects quality and improves productivity
• Limitations of Learning Curve Analysis :
– The learning curve approach is not as effective when machinery performs repetitive
tasks, such as robotics. It is most appropriate for labor-intensive contexts that involve
repetitive tasks, long production runs, and repeated trials.
– The learning rate is assumed to be constant in the calculations, but actual declines in
labor time are not constant. The analyst needs to update projections based on the
observed progression of learning.
– Conclusions might be unreliable because observed changes in productivity actually may
be due to factors other than learning, such as a change in the labor mix, the product
mix, or some combination of the two.
16
Forecasting Techniques
• Expected Value :
• When a company is trying to forecast its sales for the upcoming year, the
results can vary depending on the economic conditions.
• IF the economy is booming, the company’s sales may be higher, but if the
economy is in a recession, the company’s sales will likely be lower.
• Thus, future sales are considered to be a random variable because their
outcome is uncertain.
• After creating sales forecasts based on various economic conditions and
assigning a probability for the likelihood of each economic scenario, the
company can lay out the sales forecasts and probabilities for the various
scenarios in a table.
Expected Value (EV) = 𝑆 𝑋 𝑃𝑋
EV = expected value
Σ = sum of the variables that follow in the equation
S = amount associated with a specific outcome
Px = probability associated with a specific outcome
17
Forecasting Techniques
Economic condition Sales Forecast Probability
Boom $ 3,000,000 0.10
Average $ 2,000,000 0.80
Recession $ 600,000 0.10
18
• Benefits and Limitations of Expected Value
– The expected value technique helps an organization determine the average
outcome of an event when faced with uncertainties.
– These averages often help the organization decide whether it should undertake
certain actions.
• For example:
– An organization may be presented with an investment option with a 60% chance of
making a profit of $300,000 and a 40% chance of incurring a loss of $500,000. The
calculation of the expected value of this investment reveals that the expected
profit is:
– Expected Profit : ($300,000 × 0.6) + (−$500,000 × 0.4) = −$ 20,000
– In other words, this investment is expected to lose, on average, $20,000.
– Given this information, it would not be a good idea for the organization to pursue
this investment.
– However, the expected value calculation is only as good as the estimated potential
outcomes for each scenario and the probability assigned to each scenario.
– IF any of these assumptions is unreliable, then the expected value that is calculated
cannot be trusted in making sound decisions.
– Expected value analysis assumes that the decision maker is risk neutral.
– IF the decision maker is either a risk taker or risk averse, then the expected value
model would not be appropriate. 19
Forecasting Techniques
• Sensitivity Analysis. :
– Decision maker use sensitivity analysis to help decide what changes in a given situation
(called “state of nature”) are most likely to produce a particular outcome (called
“payoff”).
– Sensitivity analysis can be used to study impacts on qualitative as well as quantitative
“state of nature” and “payoff”.
– Sensitivity analysis is a type of “What if analysis” because it is conducted by changing a
specific variable and determining how sensitive the outcome is to changes in input.
21
End of Topic
Mr. Tamer Bedir
tamerbedir_81@yahoo.com
00966541553318
22