Monday, December 8, 2014

Demand Forecasting MEFA NOTES - Material for Demand Forecasting

Posted by Vishnureddy at 11:36 PM

Share this Post and Be Awesome

By: B. Mallesham,
Assistant prof,
Dept of MBA, JITS.


It is customary in all advanced countries to estimate demand for goods and services in a realistic manner so as to carry out the production plans.
Where the supply is not in accordance with demand, it results in the development of black market or excessive prices. The demand forecasting guides the entrepreneurs to set up there business/industrial activities.
Where there is a lot of competition, the entrepreneur has to estimate the demand for this product and services so that he can plan his material inputs, such as manpower, finance advertising and others. In India, the services of the National Council of Applied Economics Research (NCAER) are very valuable in terms of demand estimates. It estimates the probable demand for both industrial and consumer goods at regional and national levels. These estimates guide the entrepreneurs to plan there production.

  • Functional nature of demand.
  • Functional nature of market demand.


Based on the period under forecast, the demand forecasting can be of two types
  • short-run forecasting
  • long-run forecasting

It covers the period of 1 year. It helps to foresee the fluctuations in the demand based on seasonal and cultural factors such as festivals and plan this schedules of production or supply accordingly.
It forecasts facilitate decisions by which the utilization of resources can be optimized. The annual requirements of inventory can be procured at competitive prices.
In the light of short run forecasts, the trader can formulate an appropriate pricing policy in tune with seasonal fluctuations.
The performance of the sales force can be monitored with necessary incentives very closely if short run forecasts are available.
A short range forecast of the total demand for a particular product helps to provide a basis for ordering raw materials, to plan and schedule production activities to seek short run finances.
It covers any period ranging from 1 to 20 years. It provides information for major strategic decision that results in extension or reduction of limiting resources, it take takes long run finances.
It provides an appropriate basis to recruit human resources with a required specialization.
A long range forecast of regional demand for its major product line helps to provide the basis for considering market expansion.


Forecasting at Firm level means estimating the demand for the products and services offered by single firm.
The aggregate demand estimated by goods and services of all the firms constitutes the Industry level forecast.
The total estimate of different trade associations can also been viewed as Industry level forecasting.
National level forecasting is for the whole economy. These are worked out based on the levels of income, savings of the consumers.
It is essential to consider the appropriate indices of industrial production to work out national forecasts. With globalization and deregulation, the entrepreneurs have started exploring the foreign markets for the global level forecasts

Established products are existing goods but new products require some advertisements.
Joel deans suggested some number of possible approaches to forecast demands for new products

1. Find out the demand for a related existing product and project the demand for the new product as its outgrowth.

2. The new product, in consideration, can be analyzed ass a substitute for some existing product.

3. Asses the pattern of growth of established products and estimate the rate of growth and the likely level of demand for the new product over the given period.

4. Asses the demand through a sample or total survey of consumers’ intentions over the new product features and price.

5. Offer the product for sale in a limited market and test the results. If the results are good offer the same in wider market. When the results are not satisfactory, find out the reasons for the poor sales same in wider market and work out on the product again to ensure wider acceptance.

6. Contact dealers to know the customers reaction to the new product.


The goods are classified into...
  • Producer goods
  • Consumer goods
  • Consumer durable and services


These may be a single trader or a few traders depending upon the nature of goods and services.


Every forecast of demand should clearly list the assumptions made about the demographic, economic, technological, political, cultural environment.
Other factors like…
  • Political developments.
  • Changing fashions.
  • Customer references.
  • Changes in technology.
  • Changes in price level or inflation.


Market demand for a particular product is the total volume that would be bought by a defined customer group in a marked geographical area in a certain marketing program.
Market demand is affected by a host of controllable factors such as the demographic, economic, technological, political &cultural environment, as elasticity with respect to industry price promotion ,product improvements & distribution efforts at given industry prices & marketing outlays.


Market demand is a function of many variables out of which marketing effort is permanent.

From the above fig...

The curve shows market forecast here refers to the market demand function that rises with higher levels of industry marketing effort.
A chance in industry marketing effort will bring forth a change in the curve above. These could be a minimum amount of sales even when there is no significant marketing expenditure by the industry.

The results increase initially for every increase in the market expenditure. But this is not eternal. This is only up to a particular level. Beyond this level, the returns decrease. Higher marketing expenditure would not necessarily stimulate further demand. This implies an upper limit to market demand called the market potential or maximum possible level of demand.

The curve also reflects alternative forecasts of, market demand associated with alternative levels of industry marketing efforts. What is shown in the curve is only one of the many possible levels of industry marketing efforts that will actually occur. The market demand for a given marketing efforts is called the estimated demand. It is the expected level of industry marketing effort in a given environment.


The following steps constitute a scientific approach to demand forecasting.

  1. Identify & state the objectives of forecasting clearly.
  2. Select appropriate method of forecasting
  3. Identify the variables affecting the demand for the given product or service.
  4. Express these variables in appropriate firms.
  5. Collect the relevant data represent the variables.
  6. Determine the most probable relationship between the dependent, independent variables, using the appropriate statistical techniques.
  7. Make appropriate assumptions to forecast & interpret results in terms of market share, turn over in terms of value & volume. Product groups, individual products, sizes & brands of each individual product.
  8. Let these be alternative forecasts to make this forecasting exercise more meaningful.


>>Survey method.
1) Survey of buyer’s intention.
A] Census method
B] Sample method.
2) Sales force opinion method.

>>Statistical methods
1) Trend projection method.
A] Trend line observation.
B] Least square method.
C] Time series analysis.
D] Moving average method.
E] Exponential smoothing.
2) Barometric techniques.
3) Simultaneous equations method.
4) Correlation & regression method.

>> Other methods
1) Expert opinion method
2) Test marketing.
3) Controlled experiments.
4) Judgmental approach.

Survey methods:-

Survey of buyer’s intention:-

To anticipate what buyers are likely to do under a given set of circumstances, a most useful source of information would be the buyers themselves. It is better to draw a list of all potential buyers, approach each buyer to ask how much does her plans to buy of the given product at a given point of time under particular conditions.
This is the most effective method because the buyer is the ultimate decision maker and we are collecting the information directly from him.
The survey of the buyers can be conducted either by covering the whole population or by selecting a sample group of buyers.
Suppose there are 10,000 buyers for a particular product and the company wishes to collect the information of all its buyers, this method is called Census method or Total Enumeration method. This method is time consuming and costly.
The firm can select a group of buyers who can represent the whole population .This method is called Sample method.
It can be done in less time and with fewer costs.
There are professional organizations specialized in market research on behalf of the firms who wish to forecast the demand for their products and services.

A Questionnaire is designed to elicit the information.
The data collected forms the information base to design the consumer profile. These consumer profiles guide the firms to identify the factors that influence demand.
The firms work to strengthen their weak areas of operation and to refocus their marketing strategy.
Specialized organizations such as the ORG-Marg and other conduct consumer surveys on large scale and offer necessary market intelligence.
Surveys of this nature focus directly on the consumer requirements and their behavior.
Advantages of the survey methods:-
  1. Where the product is new in the market for which no data exists previously.
  2. When the buyers are few and they are accessible.
  3. When the cost of reaching them is not significant.
  4. When consumers stick to their intentions.
  5. When they are willing to disclose what they are willing to do.
  1. Survey may be expensive.
  2. Sample size and timing of survey.
  3. Methods of sampling.
  4. In consisted buying behavior.

Sales force opinions:-

Another source of getting reliable information about possible level of sales or demand for a given product or services is the group of people who sell the same. Thus we can control the limitation of cost and delays in contacting the costumers.
The sales people are those who are in constant touch with the main and large buyers of particular market.
The sales force is capable of assessing the likely reaction of the costumers in their territories quickly; giving the company’s marketing strategy. It is less costly and can be conducted through telephones, fax, video conferences and many more.
Here also there is a danger that salesmen may sometimes become biased with their views.
  • The sales people are paid based on their results.
  • Targets are set for the salesmen.
  • The salary of the salesmen depends upon the targets.
  • Incentives are paid to the salesmen who achieved the targets.
  • Salespersons having more knowledge about the information of sources.
  • Salesmen are cooperative.

Statistical methods:-

For forecasting the demand for goods and services in the long-run, statistical and mathematical methods are used considering the past data.
(a)Trend projection methods:-
This is based on past sales patterns. The necessary information is already available in company files with different time periods.

There are five main techniques:
  1. Trend line by observation.
  2. Least square method.
  3. Time series analysis.
  4. Moving average method.
  5. Exponential smoothing.

(1)Trend line by observation:-
It is easy and quick as it involves plotting the actual sales data on a chart and then estimating just by observation when the trend line lies.

(2) Least square method:-
In this statistical method is used. The trend line is the basis to extrapolate the line for future demand for the given product or service on graph. Here it is assumed that there is a proportional exchange in sales over a period of time. In such a case the trend line equation is in linear form.

The estimated linear trend equation of sales is written as:
S = x + y (T)
x & y have been calculated from past data.
S = sales;
T = year no. for which the forecast is made.

To find x & y values,
ΣΣ S = N x + yΣΣ T
ΣΣΣ ST = x Σ T + yΣ (T * T )
S = sales;
T = year number
N = no. of years.

Example 1:

Sales (lakhs)

Estimate the sales for the years 2004 & 2006.

Σ S = N x + yΣ T
Σ ST = x Σ T + yΣ (T * T )

Year no. (T)
Sales (s)
T * T

ΣT = 25

Substituting the values in the formula,
437 = 5x + 25 y
2265 = 25x + 165 y

By solving these equations
x=77.4 & y=2;
Years 2004 & 2006 take on the year numbers 11 and 13 respectively.
By substituting the values in the trend equations x + y (T)
S 2002 = 77.4 + 2 (11)
= 99.4 lakh units
S 20o4 = 77.4 + 2 (13)
= 103.4 lakh units.
Thus the forecast sales for year 2004 & 2006 are 99.4 and 103.4 lakh units.

3) Time series analysis:-

Where the surveys or market tests are costly and time consuming, statistical and mathematical analysis of past sales data offers another method to prepare the forecasts that is time series analysis.
The product should have actively been traded in the market for quite sometime in the past.
Considerable data on the performance of the product or service over significantly large period should be available for better results under this method.
Time series emerge from a data when arranged chronologically, given significantly large data.

The following 4 major components analyzed from time series while forecasting the demand.

Trend (T):
It also called as long term trend, is the result f basic developments in the population, capital formation & technology. These developments relate to over a period of long time say 5 t0 10 years, not definitely over night. The trend is considered statistically significant when it has reasonable degree of consistency. A significant trend is central and decisive factor considered while preparing a long range forecast.

Cycle Trend (C):
It is wave like movement of sales inflation, during the period of inflation prices go up and down.

Seasonal Trend (S):
More goods are sold in festivals seasons, weather factors, holidays.

Eratic Trend (E):
Results from the sporadic occurrence of strikes, riots etc.


This method considers that the average of past events determine the future events.
This method provides consistent results when the past events are consistent and unaffected by wide changes.
The average keeps on moving depending upon the no. of years selected. Selection of no. of years is the decisive factor in this method. Moving averages get updated as new information flow in.
This method is easy to compute. One major advantage with this method is that the old data can be dispensed with once the averages are calculated. These averages, not original data, are further used as the forecast for next period. It gives equal weightage to data both in the recent past and the earlier one.

Example: - Compute 3-day moving average from the following daily sales data.

Date and month
Daily sales (lakhs)
3-day moving average
Jan 1
Jan 2
Jan 3
Jan 4
Jan 5


To calculate 3-days moving avg…
S4 = (40 + 44 + 48)/ 3 == 44
S5 = (44 + 48 + 45)/3 == 45.7


This is a more popular technique used for short-run forecasts. This method is an improvement over moving averages method.

All time periods ( ranging from the immediate part to distant part ) here are given varying weights , that is the value of the given variable in the recent times are given higher weights and the values of the given variable in the distant past are given relatively lower weights for further processing.
The formula used for exponential smoothing,
S t + 1 == c S T + (1 -- C) S MT
S t + 1 == exponentially smoothed average for New Year.
S t == actual data in the most recent part.
S Mt == most recent smoothed forecast.
C = smoothing constant.
If the smoothing constant ` c ` is higher, higher weight is given to the most recent information. The value of `c` varies between `0` and inclusive and the exact values of `c` is determined by the magnitude of random variation. If the magnitude of random variations is large, lower values of c are assigned and vice versa. However, it is considered that a value between 0.1 & 0.2 is more appropriate in most of cases.

Time period
Actual sales(st)
(units in lakhs)
Predicted sales
(units in lakhs)

Let us take four period averages as the initial forecast years while smoothing constant of c == 0.1

S5 = (s 1 + s2 + s3 + s4)/4;
= (5.0+5.6+6.7+5.8)/4;
= 23.1/4 == 5.775.
Sales for s5 = 6.9
S6 is calculated as given below...
S6 = c * s5 + (1 - c) S m t
= 0.1 * 6.9 + (1 - . 9) * 5.775
= 0.69 + (0.9) * 5.775
= 0.69 + 5.1975
= 5.887.
Predicted sales for year 7 can be worked out.
S7 = c * s6 + (1 - c) * S m t
= 5.808


Where forecasting based on time series analyses or extrapolation may not yield significant results, barometric techniques can be made use of . Under the barometric technique, one set of data is use to predict another set.
To forecast demand for a particular product or service, use some other relevant indicator which is known as a barometer of future demand.
To assess the demand for services in India and abroad. We can see the percentage of population in each occupation. In the US 78%of the labour force is employed in services 15% of them in manufacturing. In India, according to 1991 census, 21%of work force is engaged in services, 13%in manufacturing, and 67% in agriculture. The world over, an increase in prosperity has been accomplished by an increase in demand for services.
We can know how many jobs in a particular sector can be created, based on the estimated revenues from it over a period of time. A study conducted by the National Association of Software and Service Companies (NASCOM) indicates that about, 23000 Indians are already generating $ 225 million in revenue from such services; the NASCOM estimates that these services could generate 1.1 million jobs and $ 19 billion in revenues by 2008. These estimates may guide the policy makers to initiate necessary measures to offer necessary courses in the universities to change the course syllabus.
Mc Kinsey & co, one of the globally leading many .consulting comp. identified 11 white collar services that can be performed off - shore with relative ease. The firm projects by 2010 , a global market of $180 billion of such services including finance and accounting ( $ 20 billion ) , engg & design($ 6 billion) , data search integration and management( $ 20 billion),distant education($19 billion),website services ( $ 7 billion ) ,customer integration services($ 42 billion ).
These figures guide many entrepreneurs to plan there ventures accordingly.

Simultaneous equation method

In this method al variables are simultaneously considered, with the conviction that every variable influences the other variable in an economic environment. Hence the set of eqns equal the no. of dependent variable which is also called endogenous variables.
This method is more practical in the sense that it requires to estimate the future values of only predetermined variables. it is difficult to compute where the no. of eqns is larger.


Correlation and regression methods are statistical techniques. Correlation describes the degree of association between 2 variables such as sales and advertisement expenditure, when the 2 variables tend to change together then they are said to be correlated. The extent to which they are correlated can be measured by correlation coefficient.
In regression analysis an equation is estimated which best fits in the sets of observations of dependent variables and independent variables. The main advantage of this method is that it provides the values of independent variables from with in the model itself. Thus it frees the forecaster from the difficulty of estimating them exogenously.


Well informed persons are called experts. Experts constitute yet another source of information. These persons are generally the generally the outside experts and they do not have any vested interests in the results of a particular survey.
Main advantages are:
  1. Results of this method would be more reliable as the expert is unbiased, has no direct commercial involvement in its primary activities.
  2. Independent demand forecast can be made relatively quick and cheap.
  3. This method constitutes a valid strategy particularly in the case of new products.
The main disadvantage is that an expert can’t be held accountable if his estimates are found incorrect.
It is likely that opinions give in by buyers, sales man or other experts may be at times, misleading. This is the reason why most of the manufacturers favour to test there product or service in a limited matter as test-run before they launch their products nation wide.
  1. Acceptability of the product can be judged in a limited market.
  2. Before its too late, the corrections can be made to product design if necessary, thus major catestrophy, in terms of failure, can be avoided.
  3. The customer psychology is more focused in this method and the product and services are aligned or redesigned accordingly to gain more customer acceptance.
  1. It reveals the quality of product to the competitors before it is launched in his wider market. The competitors may bring about a similar product or often misuse the results of the test marketing against the given company.
  2. It is not always easy to select a representative audience or market.
  3. It may also be difficult to extrapolate the feedback received from such a test market, particularly where the chosen market is not fully representative.

Controlled experiments refer to such exercises where some of the major determinants of demand are manipulated to suit to the customers with different tastes and preferences, income groups and such others. This method can not provide better results, unless these markets are homogenous in terms of, tastes and preferences of customers, their income and soon.
This method is in infancy state and not much tried because of following reasons:
It is costly and time consuming. It involves elaborate process of studying different markets and different permutations and combinations that push the product aggressively. If it fails in one market, it may affect other markets also.

When none of the above methods are directly related to the given product or service, the management has no other alternative than using its own judgment. Even when the above methods are used, the forecasting process is supplemented with the factor of judgment for the following reasons:

  1. Historical data for significantly long period is not available.
  2. Turning points in terms of policies or procedures or casual factors cannot be precisely demanded.
  3. Sales fluctuations are wide and significant.
  4. The reasons of statistical methods are more reliable at the national level rather than firm or industry level, in such cases, the management has to rely more on its judgment to access the validity of such results.


Love to hear what you think! Thanks Would make us Smile :)

Receive all updates via Facebook. Just Click the Like Button Below else Hit close icon


Search Entire Site

Custom Search
back to top