Unit 3: Demand Estimation - Subjective Questions

Demand estimation is the process of quantifying the relationship between the demand for a product and the factors that influence it, using statistical and analytical techniques based on past and present data.

Relevance for a firm:

• Production planning: Helps decide how much to produce and schedule inventory levels.
• Pricing decisions: Estimating price elasticity guides optimal pricing strategy.
• Sales forecasting: Provides a basis for projecting future revenue.
• Resource allocation: Enables efficient deployment of labour, capital and raw materials.
• Investment decisions: Supports capacity expansion and long-term planning.
• Marketing strategy: Identifies which factors (advertising, income, price) drive demand.

In short, demand estimation reduces uncertainty and forms the foundation for rational managerial decision-making.

Key point: Estimation gives the structural relationship; forecasting uses that relationship (or trends) to look ahead. They are complementary steps in demand analysis.

Qualitative methods rely on judgement, opinions and experience rather than numerical data. They are useful when historical data is scarce (e.g., new products).

Major qualitative methods:

• Survey of buyers' intentions: Potential consumers are directly asked about their future buying plans.
• Sales force opinion (collective opinion) method: Salespersons estimate expected sales in their territories; these are aggregated.
• Expert opinion method: Views of trade experts, dealers and consultants are gathered.
• Delphi method: A structured, iterative survey of a panel of experts who respond anonymously over several rounds until consensus emerges.
• Market experiment / test marketing: The product is launched in a limited market to observe actual consumer response.

Advantages: Simple, quick, useful when data is unavailable.
Limitations: Subjective, prone to bias, less reliable for long-term precision.

The Delphi method is a qualitative forecasting technique that gathers and refines the opinions of a panel of experts through successive rounds of anonymous questionnaires.

Procedure:

• A coordinator sends questionnaires to a panel of experts.
• Experts give their forecasts independently and anonymously.
• The coordinator summarises responses and shares them (without revealing identities).
• Experts reconsider and revise their estimates in the next round.
• The process repeats until a consensus or acceptable convergence is reached.

Advantages:

• Avoids dominance by influential individuals (anonymity).
• Pools diverse expert knowledge.
• Useful for long-term and technological forecasting.

Limitations:

• Time-consuming and expensive.
• Depends heavily on the quality of the experts.
• Consensus may be forced or superficial.

Time series analysis is a quantitative forecasting technique that analyses a set of observations recorded over successive time periods to identify patterns and project future values.

Components of a time series:

• Secular Trend (): The long-term smooth upward or downward movement over a long period.
• Seasonal Variation (): Regular, periodic fluctuations occurring within a year (e.g., festive-season sales).
• Cyclical Variation (): Wave-like fluctuations over periods longer than a year, linked to business cycles (boom, recession, recovery).
• Irregular / Random Variation (): Unpredictable, erratic movements caused by unforeseen events (floods, strikes, wars).

Models linking components:

• Additive model:
• Multiplicative model:

The survey of buyers' intentions method involves directly asking potential customers about their purchase plans for a future period, either through a complete census or a sample survey.

Process:

• Identify the target population of buyers.
• Prepare a questionnaire on quantity, timing and price expectations.
• Collect responses through interviews, mail or online surveys.
• Aggregate the intentions to estimate total demand.

Merits:

• Provides first-hand information directly from consumers.
• Useful for new products and industrial goods with few large buyers.
• Reveals consumer preferences and reasons behind buying.

Demerits:

• Buyers' stated intentions may differ from actual purchases.
• Expensive and time-consuming for a large population.
• Reliability depends on respondents' honesty and ability to predict their own behaviour.

The method of least squares fits a straight-line trend to time series data such that the sum of squared deviations between actual () and estimated () values is minimised.

Objective: Minimise .

Derivation of normal equations:
Differentiate with respect to and and set to zero.

With respect to :

With respect to :

Where = number of periods.

Simplification: If the origin is taken at the middle so that , the equations reduce to:

These values of (intercept) and (slope) define the trend line used to forecast future demand.

Let the middle year (Year 3) be the origin so that .

Compute coefficients:

Trend equation:

Forecast for Year 6: Here .

Estimated sales for Year 6 ≈ 19.1 lakh units.

The moving averages method smooths out short-term fluctuations in a time series by replacing each value with the average of a fixed number of surrounding periods, revealing the underlying trend.

Procedure:

• Choose a period (), e.g., 3-year or 5-year moving average.
• Compute the average of the first values.
• Move forward one period, drop the first value and add the next, and recompute.
• Continue for the whole series.

For a 3-year moving average:

Advantages:

• Simple to compute and understand.
• Effectively smooths random fluctuations.
• Flexible—period can be adjusted.

Limitations:

• Cannot compute trend values for the beginning and end periods.
• No single formula, so cannot forecast far into the future.
• Choice of period is subjective and affects results.

Conclusion: Qualitative methods suit situations lacking data or facing high uncertainty, while quantitative methods are preferred when adequate historical data with identifiable patterns exists. Firms often combine both.

The sales force opinion method, also called the collective opinion or grass-roots method, builds a forecast from the estimates of the company's own salespeople.

Procedure:

• Each salesperson estimates expected sales in their assigned territory for the future period.
• Estimates are reviewed and adjusted for the salesperson's known optimism or pessimism.
• Territory forecasts are aggregated to obtain a regional and then a total company forecast.
• Management refines the total using its judgement about market conditions.

Advantages:

• Uses the direct knowledge of those close to customers.
• Simple and inexpensive.
• Breaks forecasts down by territory, product and customer.

Limitations:

• Salespersons may lack broad economic perspective.
• Estimates can be biased (over- or under-estimation, especially if linked to targets).
• Suitable mainly for short-term forecasts.

Both models express a time series value () as a combination of its components—Trend (), Seasonal (), Cyclical () and Irregular ().

Additive Model:

• Assumes components are independent of each other.
• Magnitude of seasonal/cyclical variation stays roughly constant regardless of the trend level.
• Suitable when fluctuations do not grow with the level of the series.

Multiplicative Model:

• Assumes components are interdependent.
• Magnitude of variations increases (or decreases) proportionally with the trend level.
• Components (other than trend) are expressed as ratios/percentages.
• More common in economic and business data where variation is proportional.

Key difference: Additive assumes constant absolute variation; multiplicative assumes constant relative (percentage) variation.

Seasonal variation refers to the regular and repetitive fluctuations in a time series that occur within a period of one year (or shorter) due to seasonal factors.

Causes:

• Natural factors: Climate and weather (e.g., higher demand for umbrellas in monsoon, coolers in summer).
• Man-made factors: Customs, festivals and holidays (e.g., higher sales during Diwali or Christmas).

Characteristics:

• Occurs regularly and can be predicted.
• Completes within a year.

Importance of measuring seasonal variation:

• Inventory management: Stock the right quantities before peak seasons.
• Production scheduling: Plan output to match seasonal demand peaks and troughs.
• Pricing and promotion: Time discounts and campaigns effectively.
• Workforce planning: Arrange temporary staff during peak periods.
• Deseasonalising data: Removing seasonal effects reveals the true underlying trend.

Accurate measurement helps a firm avoid stock-outs, overstocking and inefficient resource use.

Expert Opinion Method:
Here, forecasts are based on the views of specialists such as market consultants, dealers, distributors and trade associations who possess deep knowledge of the market.

• Process: Firm gathers opinions from experts and combines them into a forecast.
• Merits: Uses specialised knowledge; quick; useful when internal data is limited.
• Demerits: Subjective; experts may disagree; accountability is diffuse.

Market Experiment (Test Marketing) Method:
The product is introduced in a small, representative market segment to observe actual consumer response before a full-scale launch.

• Process: Select a test area, vary factors like price or advertising, and record actual sales.
• Merits: Based on actual behaviour rather than stated intentions; helps test marketing variables.
• Demerits: Costly and time-consuming; test market may not represent the whole market; competitors may interfere or gain advance information.

Comparison: Expert opinion relies on informed judgement, while market experiments rely on observed real-world data.

The trend projection method uses historical time series data to establish a trend line and extends (projects) it into the future to forecast demand.

Techniques used to determine the trend:

• Graphical (free-hand) method: Plot data and draw a trend line by inspection.
• Method of semi-averages: Divide data into two halves, average each, and join.
• Moving averages method: Smooth data to reveal trend.
• Least squares method: Fit statistically.

Forecasting: Once the trend equation is estimated, substitute the future value of to obtain the forecast.

Assumptions:

• Past patterns and relationships will continue into the future.
• The trend is the dominant component and remains stable.
• No sudden structural changes (technology, policy, competition).
• Other factors influencing demand remain broadly constant.

Limitation: Reliability falls if the market environment changes significantly, since it assumes the future mirrors the past.

A 3-year moving average is computed by averaging every three consecutive values.

Interpretation: The moving averages (22.00, 25.33, 26.67, 30.00) smooth out year-to-year fluctuations and reveal an upward trend in production. Note that trend values cannot be computed for the first (2018) and last (2023) years.

Demand forecasting serves different managerial purposes depending on the time horizon.

Short-run forecasting (up to 1 year):

• Production scheduling: Aligning output with expected demand.
• Inventory control: Maintaining optimum stock levels to avoid shortages or excess.
• Pricing policy: Setting seasonal or promotional prices.
• Cash and working capital management: Planning short-term finances.
• Sales targets: Setting realistic targets for the sales force.

Long-run forecasting (more than a year):

• Capacity planning: Deciding on plant size and expansion.
• Capital investment: Justifying large investments in fixed assets.
• Manpower planning: Recruiting and training for future needs.
• Financial planning: Arranging long-term funds.
• Business strategy: Entering new markets or launching new products.

Overall importance: Accurate forecasts reduce uncertainty, minimise wastage of resources, improve profitability and provide a rational basis for both operational and strategic decisions.

Cyclical Variation ():

• Wave-like, recurring fluctuations that occur over periods longer than one year.
• Associated with the business cycle: prosperity (boom), recession, depression and recovery.
• Not strictly regular in length or intensity.

Irregular / Random Variation ():

• Erratic, unpredictable fluctuations caused by unforeseen and non-recurring events.
• Examples: strikes, floods, earthquakes, wars, sudden policy changes.
• Cannot be predicted or modelled.

Difference from Seasonal Variation:

Summary: Seasonal variation is short and predictable, cyclical variation is longer and semi-regular, and irregular variation is completely unpredictable.

A firm should evaluate forecasting techniques against several criteria before choosing one.

Criteria of a good forecasting method:

• Accuracy: Forecast should be as close as possible to actual outcomes; measured by comparing past forecasts with realised values.
• Simplicity: Should be easy to understand and apply by managers.
• Economy: Benefits from the forecast should outweigh the cost of preparing it.
• Availability of data: Method must suit the type and quantity of data available.
• Durability / stability: Should give reliable results over a reasonable period.
• Flexibility: Able to accommodate changing conditions.
• Timeliness: Forecast must be available in time for decisions.

Why no single method is universally best:

• Suitability depends on the time horizon, data availability, product type and market conditions.
• A new product with no history needs qualitative methods, while an established product suits time series methods.
• Cost, accuracy needs and urgency differ across situations.

Conclusion: Firms often use a combination of methods and cross-check results to improve reliability.

The method of semi-averages is a simple technique for determining the trend of a time series.

Procedure:

• Divide the time series data into two equal halves.
• Compute the arithmetic mean of each half.
• Plot each average against the mid-point (mid-period) of its respective half.
• Join the two points with a straight line—this is the trend line, which can be extended for forecasting.

Illustration: Sales over 6 years: 10, 12, 14, 16, 18, 22.

• First half (Years 1–3): , plotted at Year 2.
• Second half (Years 4–6): , plotted at Year 5.
• Slope per year.
• The line joining points (Year 2, 12) and (Year 5, 18.67) represents the trend.

Advantages: Simple and objective (unlike free-hand method).
Limitations: Assumes a linear trend; influenced by extreme values since it uses arithmetic means.