Demand Forecasting

• 1. Demand forecasting In modern business forecasting is often made I onanticipation of demand. Anticipation of demandimplies demand forecasting Forecasting means expectations about the futurecourse of development. Future is uncertain but notentirely so. Demand forecasting is not a speculative exercise intothe unknown. It is reasonable judgment of futureprobabilities of market events based on scientificbackground. Demand forecasting is an estimate of thefuture demand. It cannot be cent percent precise.

2. Levels of forecasting Micro level:- It refers to demand forecasting byindividual business firm for estimating thedemand for its product. Industry level:- It refers to the demand estimatefor the product of the industry as whole. Itrelates to market demand as whole. Macro level:- It refers to the aggregate demandfor the industrial output by nation as whole. It isbased on the national income or aggregateexpenditure of the country. 3. Importance of forecasting Production planning. Sales forecasting. Control of business. Inventory control. Growth and long-term investmentprogrammes. Stability. Economic planning and policy making. 4. Types of forecasting Short term forecasting:- is for a short period up to one year. Itrelates to policies regarding sales, purchases, pricing and finance. Inmost of firms the information regarding the immediate future isnecessary for formulating a suitable production policy. Medium term forecasting:- it is an intermediate between short-term and long-term forecasting. This is usually followed by a firmwhich is subjected to the medium term variation in trade cycle. Long term forecasting:- refers to a period beyond one year. Thepurpose of long term forecasting are:- 1) planning of a new unit ofexpansion of the existing unit. A multi-product firm must know notonly total demand situation, but also the demand for differentitems. 2) Planning of man power needs. 3) Planning long termfinancial requirements is necessary for the firm to make necessaryarrangements to secure fresh capital investments. 5. Factors involved in demand forecasting Time period. Levels of forecasting. Purpose General or Specific. Methods of forecasting. Nature of commodity. Nature of competition. 6. Objectives of demand forecasting Helping continuous production. Regular supply of commodities. Formulation of price policy. To formulate effective sales performance. Arrangement of finance. To determine productive capacity Labour requirements. 7. Methods of forecasting.Methods of forecastingSurvey methodStatistical method1 .Survey of buyers intention. 1. Trend projection method 2.Survey of experts opinion. 2. Method of moving3. Combined experiments. averages.4. Simulated market situation. 3. Regression method. 4. Barometric method. 5. economic indicators. 8. Survey method. Forecast are done both for established productsand new products. Demand forecasting for theestablished products can be done in routinemanner with information drawn from existingmarkets and past behavouir of sales. Forecasts for new products are necessarilycustom built jobs that involve more ingenuity andexpense. Since the product has not been soldbefore it is difficult to get any clue for demandforecasting. 9. Survey of buyers intentions orconsumers survey. Least sophisticated method and most direct method of estimatingsales in the near future. In this method customers are directly contacted in order to findout their intention to buy commodities for future. This method isopinion survey method. Intentions are recorded through personal interview, mail or postsurveys and telephone interviews. There are two types of survey Complete enumeration method: It covers all potential consumersin the market and interviews conducted to find out probabledemand. Sample survey method: It covers only few customers selectedfrom total potential consumers interviewed and then the averagedemand is calculated on the basis of the consumers interviewed. 10. Survey or expert opinion. There are people who are experts in the field ofselling goods like wholesalers, and retailers. They will be in position to tell what consumerswould buy. Many companies get their basicforecast directly from their salesman who havemost intimate feel of the market. The wholesalers and retailers by their experienceare in the position to feel about the probablesales in the coming year. 11. CONTROLLED EXPERIMENTS Under this method different determinants ofdemand are varied and price and quantityrelationships are established at differentpoints of time in the same market or differentmarkets. Only one determinant is varied others are keptconstant and controlled. This method isrelatively new. 12. SIMULATED MARKET SITUATION Under this method an artificial market situation iscreated and participants are selected. These are called consumers clinics Those participants are given some money andasked to spend the same in artificialdepartmental stores. Different prices are set upfor different groups of buyers. The responses toprice changes are observed and accordinglynecessary decisions about price and promotionalefforts are undertaken. 13. STATISTICAL METHODS Demand forecasting uses statistical methods topredict future demand. This method is useful forlong run forecasting for the existing products. There are several ways of using statistical ormathematical data. They are: 1. Trend projection method or Time Series 2. Method of moving averages 3. Regression method 4. Barometric methods. 5. Other methods 14. 1. Trend projection Method This method is based on analysis of past sales. Afirm which has existence for quite long time willhave accumulated considerable data regardingsales for a number of years. Such data isarranged chronologically with intervals of time.This is called Time series. It has 4 types of components namely: 1. Secular trends 2. Seasonal variation 3. Cyclical variation 4. Random variations. 15. The real problem in forecasting is to separate andmeasure each of this 4 factors. When a forecastis made the seasonal, cyclical, random factors areeliminated from the data and only the seculartrend is used. The trend in Time series can be estimated byusing any one of the following of methods 1. Least square method 2. Free Hand method 3. Moving averages method 4. Method of semi averages. 16. TREND PROJECTION A Time series analysis of sales data over a period oftime is considered to serve as a good guide for sales ordemand forecasting. For long term demand forecasting trend is computedfrom the time base demand function data. Trends refer the long term persistent movement ofdata in one direction upward or downward. There are2 important methods for trend projection. 1. Method of moving averages. 2. Least square method. 17. LEAST SQUARE METHOD The trend line if fitted by developing an equation giving the natureand magnitude of the trend. The common technique used inconstructing the line of best fits is by the method of least squares. The trend is assumed to be linear. The equation for straight linetrend is y=a+bx Where a is the intersect and b shows the impact ofindependent variable. Sales are dependant on variable y sincesales vary with time periods which will be the independent variablex Thus y intercept and the slope of line are formed by makingappropriate substitutions in the following normal equations Y = na+bx --------------(1) XY = ax + bx2----------------- (2) 18. YEARSALES XX2 XY 1996 4511 45 1997 5224104 1998 4839144 1999 554 16220 2000 60525 300 N=5Y=260 X=15 X2=55 XY=813 19. SUBSITITUTING THE ABOVE VALUES IN THE TWONORMAL EQUATIONS WE GET THE FOLLOWING:- 260=5a+15b---------------- 813=15a+55b----------------- Solving both equation we get b=3.3 260=5a +15 260=5a+49.5 A=42.1 Therefore the equation for the line of best fit is equalto: Y=42.1+3.3X. 20. Using this equation trend values for previous years andestimates of sales for 2001. The trend values andestimates are as follows:- Y 1996 = 42.1 +3.3(1)= 45.4 Y 1997 = 42.1+3.3(2)= 48.7 Y 1998 = 42.1+3.3(3)= 52.2 Y 1999 = 42.1+3.3(4)=55.3 Y 2000 = 42.1+3.3(5)=58.6 Y 2001 = 42.1+3.3(6)=61.9. Based on the trendprojection equation illustrated above, the forecastsales for the year 2001 is Rs 61.9 Lakhs. 21. Method of moving averages. The trend Projection method is very popular inbusiness circles on account of simplicity andlesser cost. The basic idea in this method is thatpast data serves a guide for future sales. This method is inadequate for predictionwhenever there are turning points in the trenditself. While irregular factors such as storms andstrikes can be averaged out and contained intothe equation it is desirable to know how valuablesuch an exercise could be. 22. The calculation depends upon whether the period shouldbe odd or even. In the case of odd periods like (5, 7, 9) the averageobservations is calculated for a given period and the valuecalculated value is written in front of central valuable of theperiod, say 5 years. The average of values of five years iscalculated and recorded against the third year. In the caseof five yearly moving averages the first two years and lasttwo years of data will not have any average value. If the period is even say four years then average of fouryearly observations is written between second year andthird year values. After this centering is done by findingaverage of paired values. Let us take up the followingillustration:- 23. The following are the annual sales of dressesduring the period of 1993-2003. We have tofind out trend of the sales using a) 3 yearlymoving averages, b)4 yearly moving averages. 3 yearly moving averages will be a+b+c/3, b+c+d/3,c+d+e/3 ,d+e+f/3------- The value of 1993+1994+1995/3 12+15+14/3 = 41/3=13.7. 24. YEAR SALES3 YEARLY3 YEARLY MOVINGMOVING TOTALAVERAGE trend values 1993 12(-)- 1994 1541 41/3=13.7 1995 1445 45/3=15 1996 1648 48/3=16 1997 1851 51/3=17 1998 1754 54/3=18 1999 1956 56/3=18.7 2000 2061 61/3=20.2 2001 2267 67/3=22.3 2002 2571 71/3=23.7 2003 24 - - 25. Advantages and disadvantages This method is simple and can be appliedeasily. It is based on mathematical calculations andfinally this is more accurate. The disadvantage of this method of movingaverage is that it gives equal weight age to thedata related to different periods in the past. Itcannot be applied it if some observations aremissing. 26. Regression method The sales of any commodity depends on time. It may be associated with competitors, advertising ones ownadvertising change in population, income and size of familyandenvironmental factors. The nature of relationship can be used and future sales can beforecast. Regression analysis denotes methods by which the relationshipsbetween quantity demanded and one or more independentvariable can be estimated. It includes measurement of errors thatare inherent in the estimation process. Simple regression is usedwhen the quantity demanded is estimated as a function of singleindependent variable. Multiple regression analysis can be used toestimate demand as function of two or more independentvariables. 27. Trend projection by regression method This is a mathematical tool, with this adaptingMethod of least squares a trend line can befixed to know the relationship between time anddemand/sales. Based on this trend line sales/demand can be projected for future years. This is an inexpensive method of forecasting. Thedata will be available with the organization andbased on this data demand or sales, can beprojected for future years. 28. YEAR:- 1998 1999 2000 2001 2002SALES:- 240 280 240 300 340YEAR SALES TIMETDPRODUCT DEVIATION SQUARED TIME DEVIATION1998 24O-24 -4801999 280 -1 1 -280 -7602000 2400 0 02001 300+113002002 340 +24 +680 +980X=5 y=1400 x=0x2=10 xy=220 29. The equation is y=a+bx. In this equation a and b. a=y/n=1400/5=280. b=xy/x2=220/10=22. Now applying values to regression equation theequation will be y=280+22x From this we can ascertain sales projection from 2003,2004, 2005. For the year 2003=280+22(3)=Rs. 346 crores. For the year 2004=280+22(4)=Rs. 368 crores. For the year 2005=280+22(5)=Rs. 390 crores. 30. Simple linear equation In case of linear trend in the dependentvariable a straight line to data can be fit inwhose general form would be sales=a+b price. The straight line equation can be fit in eithergraphically or least square method. In thegraphical method the sets of data of twovariables on a graph are plotted and a scatterdiagram can be obtained. 31. YPRICESXOUNITS OF DRESSES SOLD 32. The regression in line can be approximated bysketching it free hand in such a manner than theline passes through the middle of the scatter. In the least square method of estimating theregression line, S=a+bp, the value of theconstants, a and b can be with the help of afollowing formula:- b=nSiPi (Si )(Pi )/nPi2 -(Pi)2 and a=S-bPi /n. 33. Barometric method Barometric method is an improvement over trendprojection method. In the trend projection method, the future is somepast extension of past while in the barometricevents, of the present are used to predict the future. This is done by using certain economic and staticallyindicators. The barometric techniques use time seriesto predict variables. The barometric techniques using time series, whichwhen combined certain ways provide direction ofchange in the economy or in indicators. These arecalled barometers of market change. 34. Simulation Method Every day life experience can not bemathematically explained the model may becomecomplicated and its solution will become difficultin such a situation simulation method will behelpful .this method is associated with the nameof monte carla This method is used to solve the problem by trialand error approach it is a device for studying anartificial model of a physical or mathematicalprocess ,this method combines probability andsampling method to solve complicated problem. 35. Forecasting demand for new products Evolutionary approach:- project the demand for new product as anoutgrowth and evolution of existing old product. It may be assumedcolor T.V. picks up where black and white T.V. sets are off. Thisapproach is useful only when the new product is very close to theold product. Substitute approach:- According to this approach the new productis to be considered as substitute for the old product. For examplethe new Foto setter substitutes photographic composition forestablished type setting equipment as a linotype, polythene bags assubstitute for cloth bags or ball pens, or for fountain pens. Growth curve approach:- The rate of growth and ultimate level ofdemand for new products can be estimated on the basis of patternof growth for old products. For example, analyze the growth curveof the all household appliances and establish an empirical law ofmarket development applicable to new appliances. 36. Opinion polling approach:- Estimate the demand by direct inquiryof the ultimate purchasers then blow up the sample to full scale. Sending an engineer with drawing and specifications for newindustrial products to a sample company is an example of opinionpolling which is widely used to explore the demand for newproducts. Sales experience approach:- The new product is offered for sale in asample market and then the demand for new product is estimatedin fully developed market. The sample of market has to beidentified. Vicarious approach:- the consumers reactions are indirectly studiedin this approach. Specialized dealers are contacted because theyhave intimate feel of the customers. Dealers opinion are very muchsolicited regarding the demand for new products. This approach iseasy but difficult to quantify. 37. Difficulties in forecasting Changes in size and characteristics ofpopulation Saturation limit of the market Existing stock of goods Constraints of the firm 38. Importance of demand forecasting Useful for planning of production Sales forecasting depends upon demandforecasting Useful for controlling inventories Helps in achieving targets of firm To stabilize production and employment Useful for policy making regarding long terminvestment programmes 39. Criteria for good forecasting. Joel Dean lays down the following criteria of goodforecasting method:- Accuracy :- forecast must be accurate as far as possible.Its accuracy must be judged by examining the pastforecast with present situation. Plausibility :- it implies managements understanding ofmethod used for forecasting. It is essential for a correctinterpretation of the results. Simplicity :- a simpler method is always morecomprehensive than a complicated one. 40. Economy :- it should yield quick results. A timeconsuming method may delay the decisionmaking process. Quickness :- it should yield quick results. A timeconsuming method may delay the decisionmaking process. Flexibility : - not only the forecast is to bemaintained up to date there should be possibilityof changes to be incorporated in the relationshipsentailed in forecast procedure, time to time.