CHAPTER 6

Forecasting
 Forecasting Applications
 Qualitative Analysis
 Trend Analysis and Projection
 Business Cycle
 Exponential Smoothing
 Econometric Forecasting
 Judging Forecast Reliability
 Choosing the Best Forecast Technique
 Method:
 Collection of data
from a range of sources:
▪ Market research
▪ Past sales data
▪ Market growth data
▪ Specialist analyst data
▪ Secondary data, e.g. SWS, DTI, BSP
 Demand estimates for products and services are
the starting point for all the other planning in
operations management.
 Management teams develop sales forecasts
based in part on demand estimates.
 The sales forecasts become inputs to both
business strategy and production resource
forecasts.
Forecasting is a tool used for
predicting
future demand based on
past demand information.
Demand for products and services is usually uncertain.
Forecasting can be used for…
• Strategic planning (long range planning)
• Finance and accounting (budgets and cost controls)
• Marketing (future sales, new products)
• Production and operations
What is forecasting all about?

Demand for Mercedes E Class We try to predict the

future by looking back
at the past

Predicted
demand
looking
back six
Time months
Jan Feb Mar Apr May Jun Jul Aug

Actual demand (past sales)

Predicted demand
What should we consider when looking at
past demand data?

• Trends

• Seasonality

• Cyclical elements

• Autocorrelation

• Random variation
 Process of predicting a future event
 Process of predicting the values of a
certain quantity, Q, over a certain time Sales will be
$200 Million!
horizon, T, based on past trends and/or
a number of relevant factors.
 Underlying basis of
all business decisions
 Production
 Inventory
 Personnel
 Facilities
 Examples from student projects:
 Demand for tellers in a bank;
 Demand for liquor in bar;
 Demand for frozen foods in local grocery warehouse.
 Short-range forecast
 Up to 1 year; usually less than 3 months
 Job scheduling, worker assignments
 Medium-range forecast
 3 months to 3 years
 Sales & production planning, budgeting
 Long-range forecast
 3+ years
 New product planning, facility location

4-13
 Medium/long range forecasts deal with more
comprehensive issues and support management
decisions regarding planning and products,
plants and processes.
 Short-term forecasting usually employs different
methodologies than longer-term forecasting
Short-term forecasts tend to be more accurate
than longer-term forecasts.

4-14
Examples from student projects:
Demand for tellers in a bank;
Traffic on major communication switch;
Demand for liquor in bar;
Demand for frozen foods in local grocery
warehouse.
Example from Industry: Unilever
70,000 items;
25 stocking locations;
Store 3 years of data (63 million data points);
Update forecasts monthly;
21 million forecast updates per year.
 Scheduling existing resources
 How many employees do we need and when?
 How much product should we make in anticipation of
demand?
 Acquiring additional resources
 When are we going to run out of capacity?
 How many more people will we need?
 How large will our back-orders be?
 Determining what resources are needed
 What kind of machines will we require?
 Which services are growing in demand? declining?
 What kind of people should we be hiring?
 Macroeconomic Applications
 Predictions of economic activity at the national or
international level, e.g., inflation or employment.
 Microeconomic Applications
 Predictions of company and industry performance, e.g.,
business profits.
 Forecast Techniques
 Qualitative analysis.
 Trend analysis and projection.
 Exponential smoothing.
 Econometric methods.
 Qualitative methods Quantitative methods

Rely on subjective
Rely on data and
opinions from one or
analytical techniques.
more experts.
 Types of Forecasts
 Qualitative --- based on experience, judgement, knowledge;
 Quantitative --- based on data, statistics;
 Methods of Forecasting
 Naive Methods --- eye-balling the numbers;
 Formal Methods --- systematically reduce forecasting errors;
time series models (e.g. exponential smoothing);
causal models (e.g. regression).
 Focus here on Time Series Models
 Assumptions of Time Series Models
 There is information about the past;
 This information can be quantified in the form of data;
 The pattern of the past will continue into the future.
 Qualitative (Subjective): Incorporate factors like the
forecaster’s intuition, emotions, personal experience, and value
system; primarily subjective; rely on judgment and opinion
 based on judgments, experience, and knowledge of individuals
or groups
 These methods include:
 Jury of executive opinion
 Sales force composites
 Delphi method
 Consumer market surveys
 In-house judgments
 Expert opinion
- Jury of executive opinion (Business Executives)
- Solicitation of views of individual to forecast sales
- Delphi Method
 Opinion Polls and Market research
- who are the consumers
- why the consumer is buying/not buying
- how the product is used
 Survey of Spending Plans (macro-type data)
- Consumer intentions – changes in consumer attitudes and
effects on spending
- Inventories and sales expectations
- Capital expenditures surveys
  Consumer interviews
 Range from stopping shoppers to speak with them
to administering detailed questionnaires
 Potential problems
▪ Selection of a representative sample, which is a sample
(usually random) having characteristics that accurately
reflect the population as a whole
▪ Response bias, which is the difference between responses
given by an individual to a hypothetical question and the
action the individual takes when the situation actually
occurs
▪ Inability of the respondent to answer accurately

7-22
 Delphi method – calls on the expertise and
insights of a panel of experts to help with
forecasting – seen as being more reliable than
data analysis only
 Could be drawn together from around the world
as there is no need to have people together at
the same time
 In-house judgements – Use the expertise and
judgements of those involved in the business in
aiding and making judgements
 Focus groups - a group of
individuals selected and assembled
by researchers to discuss and
comment on, from personal
experience, a topic, issue or
product
 User groups – similar to focus
groups but consisting of those who
have experience in the use of a
product, system, service, etc.
 Panel surveys – repeated
measurements from the same
sample of people over a period of
time
 Usually based on judgments about causal factors
that underlie the demand of particular products
or services
 Do not require a demand history for the product
or service, therefore are useful for new
products/services
 Approaches vary in sophistication from
scientifically conducted surveys to intuitive
hunches about future events
 The approach/method that is appropriate
depends on a product’s life cycle stage
Time Series: models that predict future demand based
on past history trends

Causal Relationship: models that use statistical

techniques to establish relationships between various
items and demand

Simulation: models that can incorporate some

randomness and non-linear effects
 Makes use of all the statistical data collected by the firm
and by other firms/organisations to help inform decision
making
 Surveys
 Sales data
 Impact on sales
 Primary data – collected by the firm themselves
 Data collected by others and used by the firm
 Advantages and disadvantages:

 Data from several years can give accurate guides to

future performance
 Statistical techniques can make the data
informative and useful
 All depends on the quality of the data and the
accuracy of the techniques used to analyze the data
 Based on the assumption that the “forces”
that generated the past demand will
generate the future demand, i.e., history will
tend to repeat itself
 Analysis of the past demand pattern provides
a good basis for forecasting future demand
 Majority of quantitative approaches fall in the
category of time series analysis
 Secular trends show fundamental patterns of
growth or decline.
 Constant unit growth is linear.
 Constant percentage growth is exponential.
 Cyclical fluctuations show variation according to
macroeconomic conditions.
 Cyclical normal goods have εI > 1, e.g., housing.
 Seasonal variation due to weather or custom is
often important, e.g., summer demand for soda.
 Random variation can be notable.
There is information about the past;
This information can be quantified in the
form of data;
The pattern of the past will continue into
the future.
  Quantitative Forecasting – generally utilizes significant
amounts of data or statistics as a basis of prediction.
a) Naïve method – eyeballing the numbers; projecting
past data into the future without explaining future trends
b) Causal or explanatory – attempts to explain the
functional relationships between the variables to be
estimated and the variables that account for the
changes.
 Qualitative Forecasting – based on judgments,
experience, and knowledge of individuals or groups

7-32
 The Business Cycle is a rhythmic pattern of
economic expansion and contraction.
 Economic indicators help forecast the
economy.
 Leading indicators, e.g., stock prices.
 Coincident indicators, e.g., production.
 Lagging indicators, e.g., unemployment.
 Economic recessions are periods of declining
economic activity.
 One-parameter Exponential Smoothing
 Used to forecast relatively stable activity.
 Two-parameter Exponential Smoothing
 Used to forecast relatively stable growth.
 Three-parameter Exponential Smoothing
 Used to forecast irregular growth.
 Practical Use of Exponential Smoothing
Techniques
 Advantages of Econometric Methods
 Models can benefit from economic insight.
 Forecast error analysis can improve models.
 Single Equation Models
 Show how Y depends on X variables.
 Multiple-equation Systems
 Show how many Y variables depend on several X
variables.
 Tests of Predictive Capability
 Consistency between test and forecast sample
suggests predictive accuracy.
 Correlation Analysis
 High correlation indicates predictive accuracy.
 Sample Mean Forecast Error Analysis
 Low average forecast error points to predictive
accuracy.
 Data Requirements
 Scarce data mandates use of simple forecast
methods.
 Complex methods require extensive data.
 Time Horizon Problems
 Short-run versus long-run.
 Role of Judgment
 Everybody forecasts.
 Better forecasts are useful.
 Q  a  bP  cM  dPR
 In linear form
 b = Q/P
 c = Q/M
 d = Q/PR
 Expected signs of coefficients
 b is expected to be negative
 c is positive for normal goods; negative for inferior goods
 d is positive for substitutes; negative for complements

7-43
 Q  a  bP  cM  dPR
 Estimated elasticities of demand are computed
as
ˆ P
 Ê  b
Q

ˆ M
 EM  c
ˆ
Q

 ˆ PR
ÊXR d
Q
7-44
 When demand is specified in log-linear form, the
demand function can be written as
Q  aP M P
b c d
R

 To estimate a log-linear demand

function, convert to logarithms
lnQ  ln a  b ln P  c ln M  d ln PR
 In this form, elasticities are constant

Ê
bˆ ˆ
ˆ
Ec Êˆ
d
M X
R
7-45
 Single Equation Model of the Demand For
Cereal (Good X)

QX = a0 + a1PX + a2Y + a3N + a4PS + a5PC + a6A + e

QX = Quantity of X PS = Price of Muffins

PX = Price of Good X PC = Price of Milk
Y = Consumer Income A = Advertising
N = Size of Population e = Random Error
Multiple Equation Model of GNP

Ct  a1  b1GNPt  u1t
I t  a2  b2 t 1  u2t
GNPt  Ct  I t  Gt

Reduced Form Equation

a1  a2 b2 t 1 Gt
GNPt    b1 
1  b1 1 1  b1
 To estimate demand function for a price-
setting firm:
 Step 1: Specify price-setting firm’s demand
function
 Step 2: Collect data for the variables in the firm’s
demand function
 Step 3: Estimate firm’s demand using ordinary
least-squares regression (OLS)

7-48
 Linear Regression
 Simple Moving Average
 Weighted Moving Average
 Exponential Smoothing (exponentially
weighted moving average)
 Exponential Smoothing with Trend (double
exponential smoothing)
 Linear regression analysis establishes a
relationship between a dependent variable
and one or more independent variables.
 In simple linear regression analysis there is
only one independent variable.
 If the data is a time series, the independent
variable is the time period.
 The dependent variable is whatever we wish
to forecast.
 Regression Equation
This model is of the form:
Y = a + bX
Y = dependent variable
X = independent variable
a = y-axis intercept
b = slope of regression line
 Constants a and b
The constants a and b are computed using
the following equations:

a=
  y-  x xy
x 2

n  x 2 -(  x)2

n  xy-  x  y
b=
n  x 2 -(  x)2
 Once the a and b values are computed, a
future value of X can be entered into the
regression equation and a corresponding
value of Y (the forecast) can be calculated.
 Simple Linear Regression
At a small regional college enrollments have grown
steadily over the past six years, as evidenced below. Use
time series regression to forecast the student enrollments
for the next three years.

Students Students
Year Enrolled (1000s) Year Enrolled (1000s)
1 2.5 4 3.2
2 2.8 5 3.3
3 2.9 6 3.4
 Simple Linear Regression

91(18.1)  21(66.5)
a  2.387
6(91)  (21) 2

6(66.5)  21(18.1)
b  0.180
105
Y = 2.387 + 0.180X
 Simple Linear Regression

Y7 = 2.387 + 0.180(7) = 3.65 or 3,650 students

Y8 = 2.387 + 0.180(8) = 3.83 or 3,830 students
Y9 = 2.387 + 0.180(9) = 4.01 or 4,010 students

Note: Enrollment is expected to increase by 180

students per year.
 Multiple
Multiple Regression
Regression Analysis
Analysis

Multiple regression analysis is used when there are
two or more independent variables.

An example of a multiple regression equation is:
Y = 50.0 + 0.05X1 + 0.10X2 – 0.03X3

where: Y = firm’s annual sales ($millions)

X1 = industry sales ($millions)
X2 = regional per capita income ($thousands)
X3 = regional per capita debt ($thousands)
 Coefficient of Correlation

x y x2 xy y2

120 9.5 14,400 1,140 90.25

135 11.0 18,225 1,485 121.00
130 12.0 16,900 1,560 144.00
150 12.5 22,500 1,875 156.25
170 14.0 28,900 2,380 196.00
190 16.0 36,100 3,040 256.00
220 18.0 48,400 3,960 324.00
1,115 93.0 185,425 15,440 1,287.50
 Assumptions of Time Series Models
 There is information about the past;
 This information can be quantified in the
form of data;
 The pattern of the past will continue into
the future.
 Business forecasting techniques provides an
important tool in the decision making process.

 Time Series analysis techniques – involves

consideration of historical data, and obtaining
estimates based on past values.

 Definition: A series of values taken over a time

period is referred to as a time series.
 A time series is a set of numbers where the
order or sequence of the numbers is
important, e.g., historical demand
 Analysis of the time series identifies patterns
 Once the patterns are identified, they can be
used to develop a forecast
Time series are collections of
observations made sequentially in 4.7275

time. 4.7083

5.5
4.6700
5.4
4.6600
5.3

5.2
4.6617

5.1
4.6517
5
4.6500
4.9

4.8 4.6500
4.7
4.6917
4.6

4.5
0 20 40 60 80 100 120 140 160 180 200
4.7533

4.8233
ECG Heartbeat Image

Stock Video
 Time series are analyzed to discover past patterns
of variability that can be used to forecast future
values (usually years, quarters, months, etc.)
 Decomposition - identify components that influence
the series.
 Trend
 Cyclical
 Seasonal
 Irregular
 This model assumes that a time series is made up of the
following four components:
 Trend - represents the long-run behavior of the data and
can be increasing, decreasing or constant.
 Cyclical – represents the ups and downs of the economy or
of a specific industry. It is a long term fluctuation and
conforms to the business cycle of slump, recovery, boom
and recession.
 Seasonal – relates to periodic fluctuations that repeat
themselves at fixed intervals of time; regularly occurring
fluctuations (usually one year or less).
 Irregular or Randomness – Variations that cannot be
explained and generally cannot be predicted; jumps in the
level of the series due to extraordinary events
 Trend – direction of movement of the data over a relatively
long period of time
 Cyclical fluctuations – deviations from the trend due to
general economic conditions (GDP, unemployment, etc.)
 Seasonal fluctuations – pattern that repeats annually
 Irregular – those that occur randomly and do not repeat
regularly and certainly cannot be predicted; represents
“noise” in a series since events never occur in a completely
regular, stable manner.
 Yt = f (Tt, Ct, St, Rt)

Yt = Actual value of the data in the time series at time t

Tt = Trend component at t
Ct = Cyclical component at t
St = Seasonal component at t
Rt = Random component at t

In additive form and multiplication forms:

Yt = Tt + Ct + St + Rt
Yt = (Tt) (Ct) (St) ( Rt)
 A time-series model shows how a time-ordered
sequence of observations on a variable is
generated
 Simplest form is linear trend forecasting
 Sales in each time period (Qt ) are assumed to be
linearly related to time (t)
Qt  a  bt

7-68
 Basic forces in trend: population change, price change,
technological change, productivity change, product life cycles
 Two basic purposes: project the trend and to eliminate it from the
original data.
 Trend analysis: independent variable (X) is time
 Method most widely used to describe straight line trends is least
squares method. Computes the line that best fits a group of
points mathematically.
 Assumes that the correct trend curve is selected and that the
curve that fits the past is indicative of the future.

Yˆ  bo  bX
 Trend is determined directly from all available data.
Seasonal component is determined by eliminating all the
other components.
 Trend is represented by one equation. A separate
seasonal value has to be calculated each period, usually
in the form of an index number. An index number is a
percentage that represents changes over time. Most
common calculation is ratio-to-moving average for the
multiplicative decomposition model.
  Use regression analysis to estimate
values of a and b
Qˆ t  aˆ  bt
ˆ
 If b > 0, sales are increasing over time
 If b < 0, sales are decreasing over time
 If b = 0, sales are constant over time

 Statistical significance of a trend is

determined by testing b̂ or by examining
the p-value for bˆ
7-71
 Q
Estimated trend
Q̂ 2009
12
line 
Q̂ 20047 
 
Sales


 

  

t

2006
2004
2005

2007
2000
1997

1999

2001
1998

2002
2003

2012
Time
7-72
7-73
 The linear component
Linear Component

40
35
R e la tiv e U nits

30
25
20
15
10
5
0
0 5 10 15 20 25 30
t, Time