Demand Forecasting
By Prof. Sujata Jhamb
Introduction :
A forecast is prediction or estimation of a future situation. Most business
houses decide in the face of risk and uncertainty. All business decisions are
based on some forecast of the level of future economic activity in general and
demand for the firm’s products in particular. The aim of economic forecasting
is to reduce the risk that the firm faces in its short term operational decision-
making and in planning for its long growth.
Forecasts can be classified into broad categories:
(1) Passive forecasts, and
(2) Active forecasts.
Passive forecast is where prediction about future is based on the assumption
that the firm does not change the course of its action and active forecast is
where forecasting is done under the condition of likely future changes in
actions by the firm
Generally, business firms are interested in both types of forecasts. Often they
predict sales in a host of policy variables, like prices of substitutes and
complements, design, quality, advertisement etc.
Forecasting Steps
The following steps are necessary for an efficient forecast :
(a) Identification of objective: It is necessary to be clear about what
one wants to get from the forecast.
(b) Determining the nature of goods under consideration: Different
categories of goods like consumer goods, durables and non-durables,
have their own characteristic and distinct demand patterns.
(c ) Selecting a proper method of forecasting: The selection of
appropriate method is based on type of data available, period for
which the forecast is to be made etc.
(d) Interpretation of results : Efficiency of forecast depends, to a
large extent, upon the efficiency in the interpretation of its results.
(P.T.O,)
(contd.,)
The Methods of Demand Forecasting
Forecasts involve the future, which is uncertain. Thus no forecasts can be
expected to be cent per cent correct. As is generally the case, there are several
methods of demand forecasting basically for three reasons :
1. No method is perfect and no method is useless
2. No method is best under all circumstances, and
3. The best method may not be available situation due to constraints form data
or resources (time and money)
There are two fundamental approaches to the problem of business forecasting:
1. To obtain information about the intentions of consumers by means of
market research survey, economic intelligence, etc (Qualitative)
2. To use past experience as a guide and by extrapolating past trends, to
estimate the level of future demand & to use statistical methods.(Quantitative
Methods)
FORECASTING TECHNIQUES
(1) QUALITATIVE METHODS
Survey methods
Consumer Survey Jury of Executive Delphi (Export Composite
Opinion Method Opinion Method) Sales - Force
Opinion survey
Interview Method
Census /Survey Sample Survey Method
Method
(2) QUANTITATIVE METHODS
Leading Exponential Moving Trend Chain End Use Econometric
Indicator Smoothing Average Analysis Ratio Analysis Method
Method Method Method Method Method Method
Survey Methods (Qualitative)
1. Expert’s Opinion Survey Method
Obtaining views from a group of specialists outside the firm has the
possible advantage of speed and cheapness. This method is best suited in
situations where intractable changes are occurring, example: forecasting
future technological states (here basic data are non-existent).It is possible
that in cases where basic data are lacking, experts may give divergent views
but even then it is possible for the manager to adapt his thinking on the
basis of these views. Delphi techniques is an example of this group of tools.
At its simplest, panel members are asked by letters to give their predictions
of the likelihood of occurrence of specified events. Postal anonymity from
other panel members minimises the impact of personal inhibitions on the
making of speculations about the future. Panel members are then informed
by letters of the outcome and particulars of the consensus. Those who
dissent are invited to give reasons or else modify their forecasts. This
process may be repeated and the final range of outcome regarded as a
probabilistic forecast.
2. Consumer’s Survey Method
This method uses the most direct approach to demand forecasting by
directly asking the consumers about their future consumption plan. It is of
three types :
1. Complete Enumeration Survey
2. Sample Survey
3. End - Use Method
I) Complete Enumeration Survey :
In the complete enumeration survey, the probable demands of all the
consumers for the forecast period 9as given by consumers themselves) are
summed up to have the sales forecast for the forecast period. For example,
if there are n consumers and their probable demands for commodity X in
the forecast are x1 x2 x3 ….xn the sales forecast would be
X = X1 + X2 + X3 …….. + Xn
(P.T.O)
(contd.,)
The advantages of this method are :
1. It give unbiased information
2. If all consumers expect accurately, the forecast will be accurate.
However, the disadvantages are:
1. Contact with a large number of consumers is tedious and
cumbersome
2. The authority of data is doubtfull. (Consumers may misjudge or
withhold data)
Nevertheless, sales forecasts for products having a few consumers
may be attempted by this method.
(P.T.O)
(ii) Sample Survey :
Under the sample survey method, the probable demand expressed by
each selected unit is summed up to get the total demand of sample units
in the forecast period. It is then blown up to find the total demand in
the market. That is, the total sample demand is multiplied by the ratio
of number of consuming units in the population to the number of
consuming units in the sample.
This method when carefully applied gives good results especially for
new products and brands. Care should be taken in choosing a sample
size which should not be too small (it will have high sampling error) or
too big (it may have little error but will be costly and tedious)
The advantages of sample survey over complete enumeration method
are:
1. It is less tedious,
2. It is less costly, and
3. It has less data error.
(iii) End-use Method :
The sale of the product under consideration is projected on the basis
of demand survey of the industries using this product as an
intermediate product. Demand for the final product is the end-use
demand of the intermediate product used in the production of this
final product. However, an intermediate product may have many
end-uses (like steel can be used in agricultural machinery,
construction, etc0. It may have demand in both domestic and
international markets. The demands for final a consumption and
exports net of imports are estimated through some other forecasting
method and its demand for intermediate use is estimated through a
survey of its user industries regarding their production plans and
input-output coefficient. Then the sum of final consumption demand
and exports demand net of imports of any commodity can be
obtained with the help of an input - output model.
Statistical Methods
(1) Trend Method
These are generally based on analysis of past sales patterns. These
methods dispense with the need for costly market research
because the necessary information is often already available in
company files. This method is used in case the sales data of the
firm under consideration relates to different time periods, i.e., it is
a time-series data. There are five main techniques of the trend or
mechanical extrapolation method.
(ii) Trend Through Least Squares Method :
The least squares method is a mathematical procedure for fittings a
line to a set of observed data points (S,T such that the sum of the
squared deviations between the calculated and observed values of S
is minimised. This techniques uses statistical formulas, rather than
visual estimates, to find the trend line which best fits the available
data. The trend line is the estimating equation which can be used
for forecasting demand by extrapolating the line for future and
reading the corresponding values of sales on the graph. The
equation of sale is :
Sales = a + b (year number)
or
S = a + bT
(P.T.O.)
Where, a and b have been calculated from past data and T is the year
number for which forecast is to be done.
Example : The sales record of company X reveals the following :
Year 1999 2000 2001 2002 2003 2004
Sales 30 40 45 50 48 57
([Link])
Estimate sales for the year 2005 and 2006
Solution : To find the values of a and b in S = a + bT we will have to
solve the normal equations
S = Na + bT
ST = aT + bT2
(P.T.O)
In the table below we find S, ST, ΞT and T2
Year Year No. Sales (S) Rs. SxT T2
(T) crores
1991 1 30 30 1
2000 2 40 80 4
2001 3 45 135 9
2002 4 50 200 16
2003 5 48 240 25
2004 6 57 342 36
T= 21 S=270 ST= 1027 T2= 91
Substituting, 270 = 6a + 21b
1027 = 21a + 91b
(P.T.O.)
(Contd.,)
Solving these equations for a and b, we get
a = 28.58
b = 4.69
Thus the trend equation becomes
S = 28.58 + 4.69T
Years 2005 and 2006 take on the year numbers 7 and 8. By
substituting these values for T, we get the forecasted sales for these
years as Rs.61.41 crore and Rs.66.10 crore respectively. The Annual
Trend is 4.69 crore rupees of sales.
(P.T.O.)
(Contd.,)
The trend method is popular because,
1. It is simple
2. Often yields good forecasts
3. Does not require the knowledge of economic theory and market.
The major drawbacks are :
1. It assumes that the past rate of change of the variable under
forecast will continue in future.
2. Not appropriate for short-run forecasts.
3. Cannot usually explain the turning points of a business cycle.
(iii) Barometric Techniques :
It is based on the idea that the future can be predicted form certain
events occurring in the present. The barometer techniques involve
statistical, usually time series, which when combined in certain ways
provide indications of the direction of change in the economy or
specific industries. These are termed as the barometers of a market
change.
1. Leading Indicators : These tend to reflect future market changes.
For example, blazer sales can be forecasted by examining the birth
rate five years earlier.
2. Coincident Indicators and Lagging Indicators: These are the
indicators which coincide with or fall behind general economic
activity or market trends like gross national product. Lagging
indicators are manufacturer’s stock levels and consumer’s credit
outstanding.
Regression Method
Regression analysis denotes methods by which the relationship
between quantity demanded and one or more independent variables
(like income, price of commodity, advertisement cost) is estimated.
It includes measurements of errors that are inherent in the
estimation process. Simple regression analysis is used when the
quantity demanded is estimated as a function of a single
independent variable, such as price. Multiple regression analysis is
use to estimate demand as a function of two or more independent
variables that vary simultaneously
In case the trend of the dependent variable is approximately linear,
we fit a linear regression equation, while if the trend is other than
linear, we fit a non-linear regression equation to the data.