Chapter Two

Forecasting

1.10 MATERIALS DEMAND FORECASTING

The success of an organization depends on how well the organization see the future
environment which is full of risks and uncertainties. In order to make prediction about the
future, we must use the past and present data. These data helps in minimizing risk and/or
uncertainties about the future.

Forecasting is one of the techniques which helps to see the future. It is also a basic tool to
help managerial decisions making. Managerial decisions are seldom made in the absence of
some form of forecasting. Every day mangers have to take decision in the face of uncertainty
without knowing what would happen in the future. For example, to keep inventory of items
without knowing future demand or sales, making investment in shares without knowing their
future return. In this case it is possible to reduce the level of uncertainty by making better
estimate of what is likely to happen in the future through forecasting. Forecasting can be
made for anything but the focus here is demand forecasting.

Type of Forecasts

Forecasts can be obtained by variety of techniques. The two type of forecasting models are:

1) Qualitative 2) Quantitative

1) Qualitative Forecasting
The qualitative model uses personal judgment and involves qualities like intuition and
experience as the bases of forecasts and are subjective by their very nature. On the other
hand quantitative models are objective in their very nature and they employ numerical
information.

Qualitative models of forecasting

Some of the common qualitative methods of forecasting are:

1. Delphi method
This method involves judgment. It is an interactive group process and employs a group of
experts, not an oracle to obtain forecasts. The experts are usually not known to each other
and their interaction takes place through a coordinator. The other participants in the delphi
process are the staffs who are involved in collecting and analyzing data. The respondents are
the subjects whose judgment is being sought.

In this model, the coordinator obtains the forecast from the experts and then assimilates each
of their forecast. Each of the experts whose estimate is not included in the middle of 50% of
the forecast are asked to explain the analysis underlying their forecast. The coordinator
disseminates to each expert a summary of all the forecasts and of the explanation of forecasts
that lies outside the middle 50%. Based on this additional information, each of the experts
may be reliable his/her original forecast. The process will be repeated appropriately, then the
experts tend to reach a consensus forecasts.

2. Sales force composite-

In this method sales force members will be asked to estimate the likely sales in their
respective areas. The estimates are then received to ensure that they are realistic. Finally, the
estimates are combined at the district, regional and national level to obtain the overall
forecast.

3. Consumer Panel Survey

Under this method consumers are questioned about their purchase plan in a consumer panel.
The aim of this method is to forecast product and service demand on the basis of subjective
judgment of consumer purchase.

There is one basic assumption to this model. i.e. the consumer in the panel are the
representatives of the ultimate/final purchasers.

2) Quantitative Forecasting
This model includes time series model and causal models.

The time series models attempt to predict the future values using the historical data. Here the
demand forecast is done on the basis of the past demand value. This prediction is based on
the premise that the future is a function of what has happen in the past.
Causal models are used when one variable is related to and therefore, depend on the values of
some other variable(s).
For example, the demand of sugar may depend on the consumption of tea.

1. Time series models.

They use time based data. A time series is a collection of readings belonging to different
time periods, usually equally spaced (which can be a month, Week or year) of some
economic variable.

Decomposition of time series:- A time series is the result of number of movements which are
caused by economic, political, natural, and other factors.

This analysis involves decomposing the past data in to components and then projecting them
forward.

A time series has four components.

1. Secular trend
Over a long period, time series will have an overall tendency either to move upwards or
downwards, though the actual movement will not be regular.

2. Seasonal variation
The fluctuation occurs periodically, the movements recurring within a defnite period may be
every month or every Week.

3. Cyclical movement
They are caused by business cycles. For example, the sales of company may be high because
the level of economic performance may be high.

4. Random movement
They are residual or erratic movements that have any set pattern and are usually caused by
some unpredictable reason.

The time series model deal first with projections of values without decomposing and then
relating to trend also making adjustment for the seasonal variation.

The following part time series models. Discuss the most commonly used

A. Moving Averages

It attempts to forecast values on the basis of the average of the values of past few periods.
Successive values are calculated by considering the new values and dropping the old one.
Example:
The demand for A is observed for 10 months & it is given below.

Month Demand ( in unit)

1 420
2 380
3 456
4 412
5 429
6 366
7 392
8 440
9 452
10 396

Question:
What is the forecast for month 11 using a 3 month forecast & a 4 month Forecast?

Solution:
A three month moving average can be obtained by adding the demand during the past
three months & dividing the sum by three, with each passing month the recent month
data is added by dropping the old to get new forecast.

 By using three months moving average, the demand in month 11 will be 492 units.

396 + 452 +440 = 429 units

3

 when a four months moving average is used, the forecast of month 11 will be
420 units.

Which can be obtained as follows:

392 + 440 + 452 + 396 = 420 units
4

B. Weighted moving average

A careful analysis of the moving average method reveals the moving average with a base of n
periods is in fact an equal weighted average of 1/n to each of the preceding n values and a
zero weighted to all the previous values. Thus in a three month forecast, the immediately last
three months’ values are given a weight of 1/3 each & the remaining values a weight of zero.
Here it is possible to assign a differential weight to the values entering in to a moving average
calculation.

Assume that the weight of the three months forecast given above is 3:2:1 then,

Weighted moving average = 3x(440)+2x(452)+1x(396)

6

436.6 units

C. Trend Projections

In this method, a trend line is fitted to the given time series data and then projections are
made in to the future by using this line. The trend line may be linear or curvilinear in nature.
In-order to obtain the trend line, the historical data are plotted on the graph, representing time
scales on X- axis. Then a line is drown through these points in such a way that
i. the sum of deviations above & below the line are equal.
ii. The sum of the squares of these vertical deviations is minimum.
In essence the trend line is drown based on the principle of least square. Such a line is
represented by
y = bx +a
where
y = is the trend line to be predicted
b = the slope of the trend line
a – the y intercept
x – independent variable (in this case time)
So far the trend line the value of b & a can be obtained as follows

b=
 xiyi  n( x)( y)
 xi  n( x)
2 2

a = y  b(x )
Example:
Consider the following demand pattern of ABC Company for iron ore.

Year Demand for iron sheet in ‘000 tons

2011 13
2012 17
2013 16
2014 16
2015 21
2016 20
2017 20
2018 23
2019 25
2020 24
2021 25
Required:
A. Forecast the demand of the iron sheet for the year 2022 for ABC Company.
Using 2016 as base year.
B. Forecast the demand of the iron sheet for the year 2022 for ABC Company.
Using 2019 as base year.

Solution:
Here by using the square method we can develop the estimating (regression) equation.
By using the last square method we can develop the line of the best fit.
y = bx +a
The line of the best fit always pass through the two points X & Y i.e. ( X , Y ).
Now in order to develop the equation of the line we need to have a base year to see the
deviation of others from it. Most of the time the base year is the middle observation. In the
case of even observations we use the mean of the two middle observations as base year by
multiplying the mean by any number. for simplicity in calculation.
 A. When we use 2016 as a base year.
Year Xi yi xiyi xi2
2011 -5 13 -65 25
2012 -4 17 -68 16
2013 -3 16 -48 19
2014 -2 16 -32 4
2015 -1 21 -21 1
2016 0 20 0 0
2017 1 20 20 1
2018 2 23 46 4
2019 3 25 75 9
2020 4 24 96 16
2021 5 25 125 25
xi = 0 yi = 220 xiyi = 128  xi2 = 110

The equation of the estimating line,

When the base year is 2016  x = 0. The value of b & a for trend line can be
obtained as follows:

b = 128 –11(0)(20) = 128 =1.1636

110 – 11(0)2 110
a = y – b( x )
20 – 1.1636(0) = 20
(y = bx + a)
The trend linear equation will be;
y= 1.1636x + 20
Therefore, the forecast for the year 2022 will be as follows;
x= 6 (because it is six years after the base year 2016
y= 1.1636(6) + 20
y= 26.9816 tons of iron sheet
B) if we take 2019 as a base year

Year Xi yi xiyi xi2

2011 -8 13 -104 64
2012 -7 17 -119 49
2013 -6 16 -96 36
2014 -5 16 -80 25
2015 -4 21 -84 16
2016 -3 20 -60 9
2017 -2 20 -40 4
2018 -1 23 -23 2
2019 0 25 0 0
2020 1 24 24 1
2021 2 25 50 4
xi = -33 yi = 220 xiyi = -532  xi2 = 209

x = xi = -33 = -3
n 11

y = yi = 220 = 20
n 11

b= -532 –(11)(-3)(20) = 128 = 1.1636

209 – 11(-3)2 110

a = y – b( x )
= 20-1.1636(-3)
= 23.4908

Then the equation will be

y = 1.1636x+23.4908
the forecast for the year 2022 will be ;
x =3 (because it is 3 yrs. after the base year -2019)
y =1.1636(3)+23.4908
26.9816 tons of iron sheet.
N.B. The forecast for the year 2022 will be the same even if the base year is changed i.e.
irrespective of the base year the forecast will be the same.

Whenever there is a change in the base year, there is no need of calculating the slope of the
equation. We need to calculate the y – intercept only & we can use the slope calculated in the
previous base year as it is.

Example:
y = 20 + 1.1636x when the middle year is used as a base year, when the base year is changed
by adding /subtracting from the value of x i.e. from 2016 to 2019 there is a 3 yrs. time gap.
So the value of x will be x +3. The trend equation by using 2019 as a base year is
y = 20 +1.1636 (x+3)
y = 20 +3.4908+1.1636x
y = 23.4908 +1.1636x
2. Causal Models of Forecasting

The causal model considers two types of variables the dependent and independent variables.
E.g., the sales of a company is depend on & is related to the price changed.

* By using regression analysis it is possible to develop a statistical relationship b/n the

dependent and independent variables.

Simple regression analysis;

It is used when there is one independent variable/ explanatory variable. Here an estimate of
the dependent variable is made corresponding to a given value of the independent variable by
putting their relationship in the form of regression line. Like the trend projection, the
regression line is obtained by plotting paired observations of x and y variables. The
regression line is then represented by the following equation.
y = a +bx where,
a- y intersect of the regression line
b- slope of the regression line
x- the independent variable
y- the predicted value of the dependent variable
b= ∑xy – n x y
∑x 2 - n x2

a= y–bx

Here parameter b is also called the regression coefficient.

Parameter a and b of the trend line drown on the principle of vast square are obtained using
the following pairs of normal equations.

∑y = na + b∑x
∑xy = a∑x + bx2
b= ∑xy – n x y
∑x 2 - n x2

a= y–bx Here
∑y – Summation of the value of the dependent variable
∑x – Summation of the value of the independent variable
∑xy – Summation of moderate of x and corresponding y value
∑x2 – Sum of the square of the value of the independent
variable
n- Number of date points

When the values of x are stated in the above equation we are going to consider the time sale
associated with the year of projections.

Illustration on Regression analysis

Consider a large firm organization engaged in producing barely. The organization feels that
the demand for its product (barely) is dependent or related to the number. of cans of beer of 1
liter consumed every year in a certain locality. To establish the demand forecast for its
product, the organization has collected the following historical data of hectares of land that
had been sawn & the number. of quintals of barely harvested.
 Year Consumption of beer in ‘000 of liters Demand for barely in ‘000 of quintals
2012 36 54
2013 26 30
2014 12 28
2015 40 48
2016 24 36
2017 18 30
2018 30 38
2019 30 46
2020 14 16
2021 34 42

Required:

If the numbers of liters of beer to be consumed in the coming period is 48000 litters (orders
already received from clients). Forecast the number of quintals of barely that could be
demanded in the year 2022.

Solution:

The first step is regression analysis is to identity the dependent and independent variables
usually denoted by x and y respectively. In this illustration the dependent variable is barely
and the independent variable is beer.

Beer (x) Barely (y) x2 xy

36 54 1296 1944
26 30 676 780
12 28 144 336
40 48 1600 1120
24 36 576 864
18 30 324 540
30 38 900 1140
30 46 900 1380
14 16 196 224
34 42 1156 1424
∑xi = 264 ∑yi = 368 ∑xi =7768 ∑yi = 10556

x = ∑x = 264 =26.4
n 10

y = ∑y = 368 =36.8
n 10
b= 10556 – 10(26.4)(36.8) = 840.8 = 1.05
7768 – 10(26.4)2 798.4

a= 36.8 – (1.05)(26.4) = 9.08

∴ y = 1.05x + 9.08

The demand for barely is in the year 2022 = 1.05 (48000) + 9.08 = 50.4 tons of
quintals.