CHAPTER FIVE

Demand forecasting
 Demand Forecasting

The business environment is dynamic and flexible

allowing number of changes to occur in it.
The managers therefore have to operate under
conditions of uncertainty and should take decisions
relating to pricing and production.
Under the conditions of uncertainty the marketing
managers depend upon techniques like demand
forecasting.

09/27/2024 2
 Cont…

Demand forecasting is an attempt to predict

the future demand based on past data under
conditions of uncertainty.
In this context, the various methods of
demand forecasting will come to the rescue
of the decision maker.
Demand forecasting occupies importance
for production, planning, product scheduling,
and inventory planning and so on.
09/27/2024 3
 Importance of Demand Forecasting
Production Planning and product scheduling
Inventory planning
Capital planning
Marketing strategy
Manpower planning
Pricing strategies

09/27/2024 4
 1. Production Planning and product scheduling
A business firm cannot function in wilderness. It
has to take crucial decisions about what to produce
and how much to produce.
This in turn depends upon its estimates of future
demand for the product. If the forecasted demand is
likely to rise, the firm can plan expansion of its
production capabilities to meet the growing
demand at the right point of time.
 In the eventuality of declining demand it should
resort to product improvement, diversification,
design changes or even pursue an aggressive sales
promotion strategy.
09/27/2024 5
 2. Inventory planning
Demand forecasting is useful for the firm to
acquire the right quantum of inventory at the right
point of time, to meet the needs of the production
department and at the same time without
unnecessarily locking up the finances of the firm
in inventory accumulation.

09/27/2024 6
 3. Capital planning
Increased production requires increased capital
resources fixed as well as working capital. Availability
of demand forecasts helps the firm to mobilize the
capital resources in time.
4. Marketing strategy
Demand forecasting will be useful in devising
appropriate sales promotion or marketing strategies. If
the demand forecasts indicate a declining trend in sales,
it should resort to intensive sales promotion campaign to
sustain its sales. Demand forecasting will also help the
firm in setting sales targets to the sales personnel.
09/27/2024 7
 5. Manpower planning

A firm has to recruit and train the appropriate level of

work force. This calls for forecasting the demand well in
advance so that the required contingent of the labour
resources could be obtained.
6. Pricing strategies
Devising and setting the optimum pricing depends upon
the forecasted demand. If the forecasts indicate a
declining share in the market demand then it has to slash
the prices to sustain demand. Conversely, if the forecasts
indicate increased demand for the product over a longer
period it can charge higher prices subject to the other
considerations.
09/27/2024 8
 Methods of Demand Forecasting
1. Consumers' opinion survey
The consumer opinion survey can be either
census or sample survey.
Where the numbers of buyers are limited,
census survey methods hold. In this case, the
opinion of the entire universe is obtained.
On the other hand, where the number of
buyers is large, universal survey is not feasible;
hence, sample survey methods are used.
09/27/2024 9
 Cont…

The sample survey can be either purposive

sampling or random sampling based on the
nature of the product or the objectives of the
survey.
The results obtained through the sample
surveys are blown up to the entire universe to
obtain the forecasted demand.

09/27/2024 10
 2. Expert opinion method

It is used under conditions of nonexistence of data or when

a new product is being launched.
 The fairest step in this method is the identification of
experts and eliciting their opinions about the likely demand
for the product.
The experts may differ in their views in which case the firm
has to pass on the opinions of one expert to the other, of
course under strict anonymity and seek their reactions.
This exercise should go on until a common line of thinking
emerges. This method will be a useful tool of demand
forecasting provided the experts did not have biased
opinions.
09/27/2024 11
 3. Collective sales force opinion survey method
In this method, the firm will extract the opinions
of the sales team, which is on the payrolls of the
company about the future demand for the product.
They express their opinions about the future
demand for the product.
The opinions so gathered are tabulated and the
demand forecasts will be arrived at.
The opinions should not be taken on the face value
as an ambitious sales man gives an over estimate
of the demand
09/27/2024 12
 4. Test marketing or controlled experimentation
Firms resort to test marketing while launching a
new product or likely to change the design or
model of the existing products.
This is also known as controlled experimentation
method as the product is likely to be launched in a
segmented market to identity its demand potential.
Test marketing will no doubt be useful method as
it ventilates the consumer preferences and
facilitates the model or design changes if necessary
but at the same time, it is an expensive proposition.
09/27/2024 13
 5. End-use or input-output methods
The end-use method applies for forecasting the
demand for intermediate products.
These are products used in the manufacture of some
other final goods.
The demand for the final product is an indicator of
the demand for intermediate product, subject to the
availability of the input output coefficients.
Once the demand for the final goods estimated, the
demand for the intermediate product can be easily
arrived at using the input-output coefficients.
09/27/2024 14
 Statistical methods of Demand Forecasting
The Regression technique
It is popular as the method of best fit. In this
case forecasts are made assuming:
(i) that there is a single determining variable
that is sales or function of time and
(ii) it is also assumed that there is a linear
relationship between the variables.

09/27/2024 15
 Cont…

Y = a + bx where
Y = sales
a= Y axis intercept i.e, constant
b = Coefficient of the determining variable x
x= time.
The normal equations for obtaining the values of a and b
are:
∑Y = Na + b∑X
∑XY = a∑x + b∑x2

09/27/2024 16
 Sales Data
Year Sales X X2 Xy
1995 25 1 1 25
1996 23 2 4 46
1997 30 3 9 90
1998 35 4 16 140
1999 43 5 25 215
2000 54 6 36 324
2001 49 7 49 343
2002 58 8 64 464
N=8 ∑y=317 ∑x=36 ∑x2=204 ∑xy=164
7
09/27/2024 17
 Cont….

Substituting these values in the normal equations

317 = 8a + 36b ……………………(1)
1647 = 36a + 204b ………………..(2)
Multiplying the equation (1) by 4.5 and subtracting it from
equation (2)
1426.5 = 36a + 162b
1647 = 36a + 204b
220.5 = 0 + 42b
b =220.5/42
b= 5.25

09/27/2024 18
 Cont…

Substituting the values of b in equation (I)

1426.5 = 36a + 162 x 5.26
1426.5 = 36a + 850.5
576 = 36a a = 16
Thus the trend equation is
Y =a + bx Y=16 + 5.25x
Applying these values the sales for the years 2003 and 2004
forecasted
Sales in 2003 Y=16 + 5.25 x 9 = 16 + 47.25 = 63.25
Sales in 2004 Y=16 + 5.25 x 10 =16 + 52.5 = 68.5
Note that 9 and 10 are the 9th and 10th years, respectively.

09/27/2024 19
 Cont…

The basic idea of the regression model is to

estimate the population parameters, β 1 and β2
from a given sample.
The sample regression function (SRF) is the
sample counterpart of the population
regression function (PRF).
Since the SRF is obtained for a given
sample, a new sample will generate different
estimates.
09/27/2024 20
 Operating with the normal equations, we have

Substituting β1in the

second normal equation

09/27/2024 21
 Regression output using stata for above table

09/27/2024 22
 ARIMA Models
The publication by Box and Jenkins of Time Series
Analysis:
Popularly known as the Box–Jenkins (BJ) methodology,
but technically known as the ARIMA methodology, the
emphasis of these methods is properties of economic
time series on their own under the philosophy let the
data speak for themselves.
Unlike the regression models, in which Yt is explained by
k regressor X1, X2, X3, . . . , Xk, the BJ-type time series
models allow Yt to be explained by past, or lagged, values
of Y itself and stochastic error terms.

09/27/2024 23
 An Autoregressive (AR) Process
Let Yt represent Demand at time t.
If we model Yt as (Yt − δ) = α1(Yt−1 − δ) + ut where
δ is the mean of Y and where ut is an random
error term then we say that Yt follows a first-
order autoregressive, or AR(1).
Here the value of Y at time t depends on its
value in the previous time period and a random
term; the Y values are expressed as deviations
from their mean value.
09/27/2024 24
 Cont…

In other words, this model says that the forecast

value of Y at time t is simply some proportion (=
α1) of its value at time (t − 1) plus a random
shock or disturbance at time t, again the Y values
are expressed around their mean values.
But if we consider this model,
(Yt − δ) = α1(Yt−1 − δ) + α2(Yt−2 − δ) + ut then we say
that Yt follows a second-order autoregressive, or
AR(2), process

09/27/2024 25
 Cont…

That is, the value of Y at time t depends on its value in

the previous two time periods, the Y values being
expressed around their mean value δ.
(Yt - δ) = α1(Yt−1 - δ) + α2(Yt−2 - δ) + ·+αp(Yt−p - δ) +
ut
In which case Yt is a pth-order autoregressive, or AR(p),
process.
Notice that in all the preceding models only the current
and previous Y values are involved; there are no other
regressors. In this sense, we say that the “data speak for
themselves.”
09/27/2024 26
 Moving Average (MA) Process
The AR process just discussed is not the only
mechanism that may have generated Y. Suppose
we model Y as Yt = μ + β0ut + β1ut−1
where μ is a constant and u, as before, is the
stochastic error term.
Here Y at time t is equal to a constant plus a
moving average of the current and past error
terms. Thus, in the present case, we say that Y
follows a first-order moving average, or an
MA(1), process.
09/27/2024 27
 Cont..

But if Y follows the expression

Yt = μ + β0ut + β1ut−1 + β2ut−2
then it is an MA(2) process. More generally,
Yt = μ + β0ut + β1ut−1 + β2ut−2 + ……..+ βqut−q is
an MA(q) process.
In short, a moving average process is simply
a linear combination of error terms.

09/27/2024 28
An Autoregressive and Moving Average (ARMA) Process
Of course, it is quite likely that Y has characteristics of
both AR and MA and is therefore ARMA.
Thus, Yt follows an ARMA(1, 1) process if it can be
written as
Yt = θ + α1Yt−1 + β0ut + β1ut−1 where θ represents
a constant term.
because there is one autoregressive and one moving
average term. In general, in an ARMA ( p, q) process,
there will be p autoregressive and q moving average
terms.

09/27/2024 29