Demand Forecasting

Views:
 
Category: Entertainment
     
 

Presentation Description

No description available.

Comments

Presentation Transcript

Demand Forecasting : 

Demand Forecasting By S.K Tannan

Introduction : 

Introduction Demand forecasting is the activity of estimating the quantity of a product or service that consumers will purchase. Demand forecasting involves techniques including both informal methods, such as educated guesses, and quantitative methods, such as the use of historical sales data or current data from test markets. Demand forecasting is used in making pricing decisions, assessing future capacity requirements, or in making decisions on whether to enter a new market or not.

Necessity for forecasting demand : 

Necessity for forecasting demand Often forecasting demand is confused with forecasting sales. But, failing to forecast demand ignores two important phenomena There is a lot of debate as how to measure and represent historical demand, since the historical demand forms the basis of forecasting. Should we use the history of outbound shipments or customer orders or a combination of the two to proxy for demand.

Market response effect : 

Market response effect The effect of market events that are within and beyond a retailer’s control. Demand for an item will likely rise if a competitor increases the price or if you promote the item in your weekly circular. The resulting sales increase reflects a change in demand as a result of consumers responding to stimuli that potentially drive additional sales. Regardless of the stimuli, these forces need to be factored into planning and managed within the demand forecast.

Methods of Forecasting Demand : 

Methods of Forecasting Demand No demand forecasting method is 100% accurate. Combined forecasts improve accuracy and reduce the likelihood of large errors because: No method is perfect and no method is useless. No method is best under all circumstances. The best method may not be available in a particular situation due to the constraints from data or resources (time and money)

Methods of Forecasting Demand : 

Methods of Forecasting Demand The different methods of demand forecasting are given below: (I) Survey Methods: Expert’s Opinion Survey Method Consumer’s Survey Method Complete Enumeration Survey Sample survey End-Use Method.

Methods of Forecasting Demand : 

Methods of Forecasting Demand (II) Statistical Methods: Trend Method Regression Method Barometric Techniques

Survey Methods : 

Survey Methods (1) Expert’s Opinion Survey Method: Obtaining views from a group of specialists outside the firm has the advantages of speed and less expensive. Under the Delphi technique, panel members are asked by letters to give their predictions. After getting replies from all the experts, they are being informed by letters about the outcomes and particulars of the consensus. Those who dissent are requested to give reasons or else modify their forecasts.

Survey Methods : 

Survey Methods This process may be repeated and a final range of outcomes may be regarded as a likely forecast. (2) Consumer’s Survey Methods: These methods uses the most direct approach to demand forecasting by directly asking the consumers about their future consumption plans. It is of three types:

Survey Methods : 

Survey Methods Complete Enumeration Survey: In this method, the probable demands of all the consumers for the forecast period, as given by the consumers themselves are summed up to have the sales forecasts for the forecast period. If there are n number of consumers and their probable demands for commodity X in the forecast period are x1,x2, x3….xn The sales forecast would be: X= X1+X2+X3...……..+Xn

Survey Methods : 

Survey Methods The advantages of this method are: It gives an unbiased information If all consumers indicate their demands accurately, the forecast will be accurate. More suitable for the sales forecasts for products having a few consumers However, the disadvantages of this method are: Contact with a large number of consumers Tedious and cumbersome The authenticity of the data is doubtful.

Survey Methods : 

Survey Methods (2)Sample Survey: Under the sample survey method, the probable demand expressed by each selected unit is summed up to get the total demand of the sample units for the forecast period. It is then blown up to find out the total demand in the market. Total sample demand x ratio of the number of consuming units in the population/number of consuming units in the sample.

Survey Methods : 

Survey Methods This method when carefully applied gives good results especially for new products and brands. However, the care should be taken that sample size should not be too small. The advantages of the sample survey are: It is less tedious It is less costly Less chances of data error.

Survey Methods : 

Survey Methods (3) End Use Method: The demand of the product under consideration is projected on the basis of the demand survey of the industries using this product as an intermediate product. However, an intermediate product may have several end-uses, like steel can be used for the production of agricultural machinery, construction, production of engineering goods etc.

Survey Methods : 

Survey Methods The intermediate product may have the demand in the domestic as well as international market. The demand for the intermediate products is estimated through survey of its user industries using their production plans and input-output co-efficient. The advantages of this method are: (i) Provides use-wise or sector-wise demand forecasts. (ii) Does not require any historical data. (iii) If the number of end-users are limited, it will be convenient to use this method.

Statistical Methods : 

Statistical Methods Trend Method: A trend is a long-term increase or decrease in the variable. For example, the time series of human population in India exhibits an upward trend, while the trend for endangered species, such as the tiger, is downward The seasonal component represents changes that occur at regular intervals.

Statistical Methods : 

Statistical Methods A large increase in sales of rain coats and umbrellas in the monsoons will be an example of seasonality. There are cyclical patterns, defined as sustained periods of high values followed by low values. Finally, the remaining variations in the data that does not follow any pattern is due to random fluctuations.

Statistical Methods : 

Statistical Methods Trend Projection: One of the most commonly used method of forecasting is trend projection. This approach is based on the assumption that there is identifiable trend in the time-series data. (a) Fitting a trend line by observation: This method is elementary, easy and quick.

Statistical Methods : 

Statistical Methods It involves merely plotting the actual sales data on the chart and estimating by observation where the trend lies. This line can be extended towards a future period and the corresponding sales forecasts

Statistical Methods : 

Statistical Methods (b) Trend through Least Squares Method: The least squares method is mathematical procedure for fitting a line to a set of observed data points (S, T) so that the sum of the squared deviations between the calculated and observed values of S are minimized. 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.

Statistical Methods : 

Statistical Methods Advantages of Trend: The trend method is popular because, It is simple, Often yields good forecasts Does not require the knowledge of economic theory and market.

Statistical Methods : 

Statistical Methods The major disadvantages of the Trend Method are: It assumes that the past rate of change of the variable under forecast will continue in future. Not appropriate for the short-term forecast. Cannot usually explain the turning points of the business cycle.

Statistical Methods : 

Statistical Methods (2) Barometric Method: Trend projection method use the time-series data to predict the future based on the past relationship. But if there is no clear pattern in the time series, the data are of little value for forecasting. The barometric technique is based on the principle that future can be predicted based on the certain events occurring in the present.

Statistical Methods : 

Statistical Methods A time series that is correlated with another time series is sometimes called an indicator of the second series. Business Conditions Digest, a monthly publication of the U.S Department of Commerce reports information on over 300 time series. These data are closely followed by economists, managers and financial analysts.

Statistical Methods : 

Statistical Methods Leading Indicators: If changes in one series consistently occur prior to changes in another series, a leading indicator has been identified. For the purpose of forecasting, leading indicators are of primary interest. The value of leading indicators depend on several factors:

Statistical Methods : 

Statistical Methods First, the indicator must be an accurate i.e. its fluctuations must closely co-relate closely with fluctuations in the series that it is intended to predict. Second, the indicator must provide adequate lead time. Even if two series are highly co-related, an indicator may be of little use. Third, lead time must be relatively constant. If a series lead another by six months on one occasion and by two years next time, the indicator will be little use because it cannot provide useful forecasts.

Statistical Methods : 

Statistical Methods Fourth, there should be logical explanation as to why one series predicts another. Finally, an indicator’s value is affected by the cost and time necessary for data collection. A time series that can be maintained only at a very high cost may not be worth the expense. Similarly, if there is a long delay before the data are available, the effective lead time of the indicator may be too short to be useful.

Statistical Methods : 

Statistical Methods Leading Indicator Economic Variable predicted by the Indicator (i) Average workweek (i) Manufacturing Output (ii)New Orders for durable goods (ii) Sale of Durable Goods (iii) New Orders for capital goods (iii) Sale of capital goods (iv) New Building permits (iv) Private housing starts (v) Changes in mfg output (v) General economic conditions (vi) Industrial material prices (vi) Consumer Prices (vii) Common stock prices (vii) General economic conditions.

Statistical Methods : 

Statistical Methods (ii) Coincident Indicators: If two series of data frequently increase or decrease at the same time, one series may be regarded as a coincident indicator of another. These indicators are those which change at approximately the same time as the whole economy, thereby providing information about the current state of the economy. Personal income, GDP, industrial production and retail sales are coincident indicators. A coincident index may be used to identify, the dates of peaks and troughs in the business cycle.

Statistical Methods : 

Statistical Methods (iii) Lagging Indicators: Lagging Indicators are indicators that usually change after the economy as a whole does. Typically the lag is a few quarters of a year. The unemployment rate is a lagging indicator: employment tends to increase two or three quarters after an upturn in the general economy. Lagging indicators are manufacture’s stock levels and consumer credit outstanding

Statistical Methods : 

Statistical Methods (iii) Regression Analysis: YEAR (X) Time X2 SALES (Y) XY deviations (Rs. In crores) 1998 -2 4 240 -480 1999 -1 1 280 -280 2000 0 0 240 0 2001 +1 +1 300 +300 2002 +2 +4 340 +680 Σ=0 Σ=10 Σy=1400 Σxy=220

Statistical Methods : 

Statistical Methods The equation is Y = a + bx ‘ a’ – independent variable ‘ b’ – exhibits rate of growth a & b can be found out as follows: a = ∑y / n = 1400 / 5 = 280 b = ∑xy / ∑ x 2 = 220/10 = 22 Now, applying values to the regression equation, Y = 280 + 22x

Statistical Methods : 

Statistical Methods Hence, sales projection from 2003-2005 can be ascertained. 2003 = 280 + 22(3) = Rs.346 crores 2004 = 280 + 22(4) = Rs.368 crores 2005 = 280 + 22 (5) = Rs.390 crores

Statistical Methods : 

Statistical Methods Econometric Models: (I) Single Equation Models: Many managerial forecasting problems can be adequately solved by single equation econometric models. The first step in developing an econometric model is to express relevant economic relations in the form of an equation.

Statistical Methods : 

Statistical Methods When constructing a model for forecasting the regional demand for portable personal computers, one might hypothesize that computer demand (C) is determined by Price (P), Disposable Income (I), Population (Pop), Interest rate (i) and advertising expenditure (A).

Statistical Methods : 

Statistical Methods A Linear Model expressing this relationship is: C= ao+a1P+a2Pop+a3i +a4i+a5A The next step in the econometric model is to estimate the parameters of the system or the values of the co-efficients. The most frequently used technique used is the application of least square regression analysis.

Statistical Methods : 

Statistical Methods .once the model co-efficients have been estimated, forecasting with a single equation model consist of evaluating the equation with specific values of the independent variables. An econometric model used for forecasting must contain independent or explanatory variables whose values for the forecasting period can be readily obtained.

Statistical Methods : 

Statistical Methods Multiple Equations System: Multiple Equation Models are composed of two basic kinds of expressions: Identities Behavioral Equations (1)Identities: Identities express relations that are true by definition. The statement that profits are equal total revenue minus total costs is an example of identity.

Statistical Methods : 

Statistical Methods (2) Behavioral Equations: Behavioral equations reflects hypothesis how variables in a system interact with each other. These equations may indicate how individuals and institutions react to various stimuli. For example, multiple equations system can be used to examine a single three equations forecast model for equipment and related software sales for a personal computer retailer.

Statistical Methods : 

Statistical Methods (iii) The total revenues for a typical retailer usually includes not only sales of personal computers, but also sales of software programmes (including computer games) and sales of peripheral equipments (monitors and printers etc.)

Statistical Methods : 

Statistical Methods The three equations are: St=b0+b1TRt+u1--------------------(1) Pt= c0+c1Ct-1+u2-------------------(2) TRt= St+Pt+Ct-----------------------(3) Where: S = Software Sales, TR= Total Revenue P= Peripheral Sales, C= Personal Computers, t= Current time period, t-1= Previous time period. u1 and u2 are errors or residual terms.

Statistical Methods : 

Statistical Methods Equation (1) indicates that the current period software sales are a function of the current level of the total revenue. Equation (2) indicates that the current period peripheral sales depend on previous period personal computer sales. Equation (3) is an identity. It defines the total revenue as being the sum of the software, peripheral and personal computer sales.

Statistical Methods : 

Statistical Methods Substituting equation (1) into 3: TRt= b0+b1 TRt +Pt+Ct A similar substitution of equation for Pt : TRt= b0+b1 TRt +C0+c1Ct-1+Ct or (1-b1) TRt=b0+C0+c1Ct-1+Ct or TRt= b0+C0Ct-1+Ct = b0+C0 Ct-1 + Ct (4) (1-b1) (1-b1) (1-b1)

Statistical Methods : 

Statistical Methods The equation (4) now relates current total revenues to previous period and current period personal computer sales. Assuming that the data on previous period personal computer sales can be obtained and that current period personal computer sales can be estimated by using single equation model, the equation (4) provides a forecasting model that accounts for simultaneous relations expressed in this simplified multiple equation system.

Statistical Methods : 

Statistical Methods Moving Average Method: A moving average forecast is based on the average of a certain number of most recent periods. Under this method, either 3 yearly, four yearly or 5 yearly etc. moving average is calculated. First, moving total of the values in group of years (3,4 or 5) is calculated, each time giving up the first preceding year from the group.

Statistical Methods : 

Statistical Methods Year Demand 3 yearly 3 yearly moving (,000) Moving Total Average 10 - - 12 36 12 14 42 14 16 45 15 15 51 17 20 54 18 19 60 20 21 - -

Statistical Methods : 

Statistical Methods The moving average (trend) so obtained is plotted on a graph. From the graphical presentation, forecasting is projected through the extension of the curve for the years ahead, measured on the x-axis. The method of least squares is more scientific than the method of moving averages. Statistical Methods

authorStream Live Help