Budgeting, Forecasting and Scenario Analysis

Introduction

Financial and risk analytics helps managers convert uncertain business conditions into informed operating and financing decisions. Budgeting specifies intended resource use over a defined period; forecasting estimates what is likely to occur; scenario analysis examines how results change under alternative assumptions. These activities are closely related but should not be confused. A budget is usually an approved plan and performance target, whereas a forecast is an updated estimate that may differ from the original plan.

For a business analyst, the purpose is not to produce a single impressive number. It is to build a transparent model whose assumptions can be challenged, updated and linked to operational drivers. A useful planning system connects sales volume, prices, costs, working capital, capital expenditure and financing needs.

Learning Objectives

By the end of this chapter, you should be able to:

• prepare operating, cash and integrated financial budgets;
• distinguish budgets, forecasts, rolling forecasts and scenarios;
• construct driver-based financial models;
• apply trend, regression and time-series methods appropriately;
• calculate and interpret forecast error and uncertainty;
• evaluate best-case, base-case and worst-case outcomes;
• communicate assumptions, limitations and risks responsibly.

1. Budgeting as a Management System

A budget is a quantified plan for a future period. It normally expresses expected revenues, expenses, assets, liabilities and cash flows. Budgets support coordination, authorization, resource allocation and performance evaluation. They also impose discipline: managers must state what they expect to do and what resources they require.

A traditional annual budget has several weaknesses. It can become obsolete when demand, inflation or exchange rates change. It may encourage managers to negotiate easy targets or spend remaining funds merely to protect the next budget. These weaknesses do not make budgeting unnecessary; they indicate that budgets should be supplemented by frequent forecasts and clear operational measures.

A sound budgeting process commonly follows this sequence:

1. Establish strategic objectives and planning assumptions.
2. Forecast demand, prices and operating volumes.
3. Prepare the sales budget.
4. Develop production or service-capacity, procurement and staffing budgets.
5. Prepare operating expense and capital expenditure budgets.
6. Convert the operating plans into an income budget, cash budget and projected balance sheet.
7. Review constraints, risks and scenarios.
8. Approve, communicate and monitor the plan.

The sales budget is often the starting point because many other budgets depend on the volume of goods or services sold. In a service organization, the equivalent driver may be billable hours, clients served or transactions processed.

2. Operating Budgets

An operating budget describes the revenue-generating and cost-incurring activities of an organization. Typical components include sales, production or service delivery, direct materials, direct labour, variable overhead, fixed operating costs, marketing and administration.

Sales budget

For each product or segment, planned revenue is:

Quantity is measured in units, hours, subscriptions or another relevant volume measure; price is measured in currency per unit. If several products are involved, total revenue is the sum of product-level revenues. The calculation assumes that quantity and price assumptions are defined for the same period and that the stated price represents recognized sales rather than merely quotations.

Production and purchasing budgets

For a manufacturer, planned production reconciles sales with inventory policy:

All three quantities must use the same unit of product. The formula assumes inventory is measured consistently and that losses, returns and work-in-progress are either negligible or separately modelled.

If each unit requires a standard quantity of material, the materials-use budget is:

Materials to purchase are then:

The purchasing budget should also include supplier lead times, minimum order quantities, price changes and payment terms. A mathematically correct quantity budget can still be operationally infeasible if suppliers cannot deliver on time.

Cost behaviour

Costs should be classified according to how they respond to a relevant activity driver. Variable costs change approximately in proportion to activity within a relevant range; fixed costs remain approximately constant in total within that range; mixed costs contain both components. A simple mixed-cost model is:

This model is useful for planning but assumes a stable cost structure, linear behaviour and a relevant activity range. Step costs, capacity limits, inflation and learning effects may invalidate the linear approximation.

Worked example: operating budget

An illustrative retailer expects to sell 12,000 units at £25 each. Variable cost is £14 per unit and monthly fixed operating costs are £72,000. Expected revenue is:

Revenue = 12,000 × £25 = £300,000.

Variable cost is:

12,000 × £14 = £168,000.

Therefore, the contribution margin is £132,000 and operating profit before other items is:

£132,000 − £72,000 = £60,000.

The contribution margin ratio is:

Thus, £132,000 ÷ £300,000 = 44%. This means that, under the assumptions, each additional pound of revenue contributes approximately £0.44 toward fixed costs and profit.

3. Cash Budgets and Integrated Financial Statements

Profit and cash are not equivalent. A sale may be recorded before the customer pays, while depreciation reduces accounting profit without being a cash payment. A cash budget identifies the timing of receipts, payments, financing and cash balances.

The basic cash reconciliation is:

All amounts must refer to the same period and currency. Financing inflows include borrowing or equity contributions; financing outflows include principal repayments, dividends or share repurchases. Interest may be included in operating or financing cash flows depending on the reporting framework, but the model must use one consistent treatment.

A monthly cash budget generally includes:

An integrated model should link the budgeted income statement, cash budget and balance sheet. For example, credit sales increase revenue and accounts receivable; collections reduce receivables and increase cash. A capital purchase increases property, plant and equipment and reduces cash, while depreciation reduces profit but does not reduce cash in the period recorded.

4. Forecasting and Rolling Forecasts

A forecast is a current estimate of future results using available information. Forecasts may be prepared monthly, quarterly or annually. A rolling forecast always extends a fixed horizon: when one month closes, another future month is added. For example, a 12-month rolling forecast in April may cover May to the following April.

Rolling forecasts are valuable when conditions change rapidly. They should not simply revise every number mechanically. The analyst should identify which assumptions changed, why they changed and whether the change is temporary or structural. A forecast should also distinguish:

• assumption: an input, such as volume growth or exchange rate;
• driver: an operational factor that causes a financial outcome;
• output: a calculated result, such as revenue or cash balance;
• constraint: a limit, such as capacity, liquidity or staffing;
• risk: an uncertain event that may affect results.

Trend methods

A simple trend model can describe a variable over time:

Here, y_t is the observed value at time t, a is the intercept, b is the estimated change per time period, and ε_t is the unexplained error. The model assumes an approximately linear trend and is inappropriate when strong seasonality, structural breaks or saturation effects dominate.

The least-squares estimates are commonly written as:

where t̄ and ȳ are the sample means of time and the observed values. The units of b are units of y per time period.

Regression and driver-based forecasting

Regression estimates the relationship between an outcome and one or more explanatory variables. A multiple linear model is:

Here, y is the predicted financial outcome, x_i are drivers, β_i are coefficients, β₀ is the intercept, and ε is the residual. A coefficient represents the expected change in y associated with a one-unit change in x_i, holding the other included variables constant.

Regression is not automatically causal. A high historical correlation may arise from a common trend, omitted variables or reverse causality. Analysts should examine data quality, outliers, multicollinearity, residual patterns and whether the relationship remains credible under future conditions.

A driver-based financial model starts with operational quantities rather than arbitrary percentage changes. For example:

Revenue = Customers × Conversion rate × Average transaction value.

Payroll cost = Number of employees × Average compensation per employee.

Cloud cost = Usage units × Cost per usage unit.

This structure improves explainability and allows managers to test operational actions directly.

Time-series methods

Time-series data may contain level, trend, seasonality, cycles and irregular variation. A moving average smooths short-term fluctuations:

It is useful when recent observations are representative and no strong trend is being ignored. An exponentially weighted forecast gives greater weight to recent observations:

where 0 ≤ α ≤ 1, y_t is the actual value in period t, and F_t is the forecast for that period. A high α responds quickly to changes but may be noisy; a low value is smoother but slower to react. The method assumes that a stable updating relationship is useful and does not, by itself, model known seasonality.

Seasonal forecasts may use seasonal indices. If the average seasonal index is normalized to 1, an index of 1.20 indicates that the period is typically 20% above the underlying average. Seasonal indices must be estimated from suitable historical data and reviewed when customer behaviour or market structure changes.

5. Measuring Forecast Performance and Uncertainty

Forecast quality should be evaluated using out-of-sample or holdout observations where possible. Common measures include:

Here, y_t is the actual value, ŷ_t is the forecast, and n is the number of observations. MAE has the same units as the outcome. RMSE also has those units but penalizes large errors more heavily. MAPE is a percentage but becomes unstable or undefined when actual values are zero or close to zero. No single metric is universally best.

Forecast uncertainty has several sources:

• aleatory uncertainty, arising from natural variation in demand or processing time;
• epistemic uncertainty, arising from limited knowledge or poor data;
• model uncertainty, arising from selecting an inappropriate model;
• parameter uncertainty, arising from estimating coefficients from a sample;
• scenario uncertainty, arising from alternative economic or strategic conditions.

A point forecast should therefore be accompanied by assumptions and, where possible, a range or prediction interval. A confidence interval concerns uncertainty about an estimated parameter or mean; a prediction interval concerns uncertainty about a future individual outcome and is generally wider. Analysts must not describe a narrow interval as certainty.

6. Scenario Analysis and Sensitivity Analysis

Sensitivity analysis changes one input, or a small group of inputs, to observe the effect on an output. Scenario analysis changes a coherent set of assumptions representing a plausible future state. For example, a downside scenario may combine lower volume, slower collections, higher input prices and delayed investment rather than changing only sales.

A useful scenario set contains:

• base case: the most defensible current expectation;
• best case: favourable but credible conditions;
• worst case: adverse but plausible conditions.

“Best” and “worst” should not mean impossible extremes. The scenario definitions and probabilities, if used, must be documented. A probability-weighted expected value is:

where p_i is the probability assigned to scenario i, x_i is its outcome, and the probabilities sum to 1. Expected value is useful for aggregation but may conceal severe downside exposure. A project with a positive expected profit can still threaten liquidity in its downside case.

A sensitivity table can show how profit changes with volume and price. The analyst should identify variables with the greatest effect, examine threshold values and test interactions. Scenario analysis is especially important for covenant compliance, minimum cash, debt service and capacity utilization.

7. Integrated Worked Example

Consider an illustrative subscription business. The base case forecasts 10,000 subscriptions, an annual revenue of £120 per subscription and a variable service cost of £48 per subscription. Fixed cash operating costs are £420,000. Assume all subscriptions are billed evenly and cash is collected within the year.

Base-case revenue is:

10,000 × £120 = £1,200,000.

Variable cost is £480,000, so contribution is £720,000. Operating cash contribution after fixed cash costs is £300,000 before tax, capital expenditure and financing.

The analyst develops three scenarios:

The worst-case operating cash result is:

Revenue = 8,000 × £115 = £920,000.

Variable cost = 8,000 × £53 = £424,000.

Operating cash result = £920,000 − £424,000 − £425,000 = £71,000.

The best-case result is:

£1,500,000 − £552,000 − £430,000 = £518,000.

The range from £71,000 to £518,000 demonstrates operating leverage and the importance of both volume and cost assumptions. If a £150,000 capital payment is also required, the worst-case ending cash before financing would be negative relative to a zero opening balance. Management should therefore arrange liquidity before the adverse scenario occurs, even though the scenario still shows positive operating cash.

8. Illustrative Case: Northstar Foods

Background

Northstar Foods is a fictional medium-sized producer of chilled meals. Its annual budget was approved before a major supermarket changed its ordering pattern. The finance team continued comparing actual monthly results with the original budget, but the comparison did not distinguish market deterioration from controllable inefficiency.

Problem

Sales volume fell, customer payment days increased and packaging prices rose. The static budget overstated expected sales and understated the cash needed to fund receivables and inventory. Managers needed an updated 12-month view and a credible downside assessment.

Analysis and approach

The analyst first separated volume, price, packaging cost, labour and working-capital drivers. A rolling forecast was built with monthly sales volume linked to supermarket orders, price linked to contract terms and packaging cost linked to units produced. Collection assumptions were linked to customer payment days rather than to revenue alone.

Three scenarios were prepared. The base case assumed gradual recovery, the best case assumed restoration of historical order levels, and the worst case assumed another customer reduction plus a 10% packaging-price increase. The model included a minimum cash balance and a borrowing limit. Historical data were reviewed for seasonality, and forecast accuracy was evaluated using MAE rather than relying on visual fit alone.

Outcome

The forecast showed that the principal risk was not annual accounting profit but a cash deficit in the third month of the downside scenario. Management renegotiated supplier payment timing, postponed a non-essential equipment purchase and introduced weekly order monitoring. The rolling forecast was updated monthly, while the approved annual budget remained as a benchmark for strategic accountability.

Learning points

The case illustrates four principles. First, forecasts should be driven by operating realities. Second, cash timing can be more urgent than reported profit. Third, scenarios should combine linked assumptions. Fourth, a budget and a forecast serve different purposes and should not be treated as interchangeable.

9. Governance, Ethics and Technology

Budget and forecast models influence staffing, investment, lending and executive evaluation. Analysts therefore have a duty to present assumptions honestly. Deliberately optimistic assumptions, hidden exclusions or selective scenario reporting can mislead decision-makers. A responsible model includes version control, an audit trail, documented data sources, approval of material assumptions and clear separation between actuals, estimates and management targets.

Confidential payroll, customer and supplier data should be protected through access controls and appropriate retention policies. Forecast models using personal or customer data must comply with applicable privacy and financial reporting requirements. Automated or AI-assisted forecasts require particular care: historical data may encode discrimination, structural breaks may be missed, and generated explanations may sound plausible without being valid. Human review, independent validation and override documentation remain necessary.

Technology can improve planning through spreadsheet controls, SQL data pipelines, business-intelligence dashboards, cloud planning systems and machine-learning models. However, automation does not remove the need for accounting reconciliation or managerial judgement. A complex model that users cannot understand may be less valuable than a simpler, transparent model with known limitations.

10. Best Practices and Limitations

Effective budgeting and forecasting practices include:

• define the decision the model is intended to support;
• use a small number of economically meaningful drivers;
• reconcile operational volumes to accounting totals;
• separate fixed, variable and step-cost behaviour;
• model cash timing explicitly;
• retain the original budget while updating forecasts;
• validate forecasts against holdout periods;
• test downside liquidity and covenant conditions;
• document assumptions, ownership and review dates;
• communicate ranges and uncertainty rather than false precision.

Limitations remain. Historical data may not represent the future. Regression relationships can fail after strategic or economic changes. Scenario labels can create false confidence if managers treat only three cases as the complete range of possibilities. Forecasts are also vulnerable to data latency, inconsistent definitions and behavioural responses: once a target is published, employees may change their actions. These limitations should be acknowledged, monitored and incorporated into decision processes.

Chapter Summary

Budgeting translates strategic intentions into quantified operating, investment and financing plans. Operating budgets commonly begin with sales and connect volumes, prices, costs and capacity. Cash budgets differ from profit budgets because they recognize the timing of receipts, payments, capital expenditure and financing. Integrated models link the income statement, cash position and balance sheet.

Forecasts are updated estimates, while rolling forecasts maintain a moving planning horizon. Trend, regression, moving-average and exponential-smoothing methods can support forecasting, but each depends on assumptions about stability, seasonality and data quality. Driver-based models are particularly useful because they connect financial outcomes to operational activity.

Forecast uncertainty arises from variation, limited knowledge, model choice and changing conditions. MAE, RMSE and MAPE provide different views of forecast error. Scenario analysis changes coherent sets of assumptions across best, base and worst cases; sensitivity analysis isolates the effect of individual inputs. Responsible analytics requires transparent assumptions, validation, governance, privacy protection and careful communication of downside risk.

Key Terms

Budget: An approved quantified plan for a future period.

Forecast: A current estimate of future outcomes.

Rolling forecast: A forecast that continually extends over a fixed future horizon.

Operating budget: A plan for revenues and operating costs.

Cash budget: A schedule of expected cash receipts, payments and financing.

Driver: An operational factor that influences a financial result.

Fixed cost: A cost approximately constant in total within a relevant range.

Variable cost: A cost that changes with activity volume.

Contribution margin: Revenue minus variable cost.

Trend: The general direction of a time series over time.

Regression: A statistical method for estimating relationships between variables.

Residual: The difference between an observed and model-predicted value.

Seasonality: A repeating pattern associated with calendar or operating cycles.

Moving average: An average of observations over a specified recent window.

Exponential smoothing: A forecasting method that weights recent observations more heavily.

MAE: Mean absolute error, measured in the outcome’s units.

RMSE: Root mean squared error, which gives greater influence to large errors.

Scenario analysis: Evaluation of outcomes under coherent alternative assumptions.

Sensitivity analysis: Testing the effect of changing selected inputs.

Forecast uncertainty: The range of possible outcomes arising from incomplete knowledge and variation.

Review Questions

Short Answer Questions

1. Distinguish an approved budget from a rolling forecast.
2. Why does a profitable business need a cash budget?
3. Define a driver-based financial model.
4. State one limitation of MAPE.
5. What is the difference between sensitivity analysis and scenario analysis?

Long Answer Questions

1. Explain how a sales budget, production budget and cash budget can be integrated.
2. Compare trend, regression and exponential-smoothing approaches to forecasting.
3. Discuss the sources of forecast uncertainty and how they should be communicated.
4. Explain why operating drivers are preferable to arbitrary percentage adjustments in many financial models.
5. Evaluate the governance and ethical requirements for an AI-assisted forecasting system.

Analytical / Application Questions

1. A business forecasts 6,000 units at £40 each, variable cost of £22 per unit and fixed costs of £ seventy thousand. Calculate revenue, contribution and operating profit, and state two assumptions behind the calculation.
2. Construct best-case, base-case and worst-case assumptions for a retailer facing uncertain demand, supplier prices and customer collection times. Identify which scenario threatens liquidity most and explain why.
3. A forecast model has low average error but repeatedly underestimates demand during December. Diagnose the likely issue and recommend a suitable modelling or process improvement.

Multiple Choice Questions with Answers

Question 1

Which statement best distinguishes a budget from a forecast?

A. A budget is always more accurate than a forecast.

B. A budget is an approved plan, while a forecast is an updated estimate.

C. A forecast contains cash information, while a budget never does.

D. A budget is historical, while a forecast is always operational.

Correct Answer: B. A budget is an approved plan, while a forecast is an updated estimate.

Question 2

A company plans to sell 5,000 units at £30 each. What is budgeted revenue?

A. £150,000

B. £166,667

C. £30,000

D. £5,030

Correct Answer: A. £150,000

Question 3

Which formula calculates required production?

A. Sales − desired ending inventory + beginning inventory

B. Sales + desired ending inventory − beginning inventory

C. Beginning inventory − sales − desired ending inventory

D. Sales × desired ending inventory

Correct Answer: B. Sales + desired ending inventory − beginning inventory

Question 4

Which forecast-error measure is expressed in the same units as the outcome and does not square errors?

A. MAPE

B. RMSE

C. MAE

D. R-squared

Correct Answer: C. MAE

Question 5

What is a key limitation of MAPE?

A. It cannot be expressed as a percentage.

B. It is unstable when actual values are zero or close to zero.

C. It always penalizes large errors more than RMSE.

D. It can only be used for annual data.

Correct Answer: B. It is unstable when actual values are zero or close to zero.

Question 6

In the equation Total cost = Fixed cost + variable cost per unit × activity, what does the variable-cost coefficient represent?

A. Total annual profit

B. Cost per unit of activity

C. Beginning cash

D. The forecast error

Correct Answer: B. Cost per unit of activity

Question 7

Which statement about scenario analysis is correct?

A. It changes only one input at a time.

B. It evaluates coherent sets of alternative assumptions.

C. It eliminates forecast uncertainty.

D. It is limited to historical data.

Correct Answer: B. It evaluates coherent sets of alternative assumptions.

Question 8

A business has beginning cash of £20,000, receipts of £80,000 and payments of £65,000, with no financing flows. What is ending cash?

A. £15,000

B. £35,000

C. £95,000

D. £165,000

Correct Answer: B. £35,000

Question 9

In exponential smoothing, a higher α generally makes the forecast:

A. Less responsive to recent observations

B. More responsive to recent observations

C. Independent of actual observations

D. Equal to the annual budget

Correct Answer: B. More responsive to recent observations

Question 10

Why should a downside scenario be assessed even when expected value is positive?

A. Expected value always ignores revenue.

B. A downside outcome may create a liquidity or covenant problem.

C. Positive expected value proves the model is wrong.

D. Downside scenarios are always more probable.

Correct Answer: B. A downside outcome may create a liquidity or covenant problem.