Some forecasting and planning techniques

1 . Introduction

Strategic planning needs estimates of the future: interest rates, exchange rates, demand, costs, technology. This chapter collects the forecasting and planning techniques the syllabus names for generating and developing strategic options: the Delphi technique, game theory, trend (regression) analysis, time series analysis, system modelling, value-driver analysis, scenario planning and real options. Some are quantitative, some structured judgement – all exist because ‘forewarned is forearmed’.

2 . The Delphi technique

Managers need views about future events – both positive and negative – whose outcomes are genuinely uncertain and on which opinions differ radically (the effect of a proposed regulation, say, on exchange rates, employment or an industry). The Delphi technique is an iterative way of moving a panel of experts towards a consensus estimate without the distortions of a round-table meeting.

The process works as follows:

1. Define the problem – for example: what will happen to the exchange rate and to employment if the regulation passes?

2. Choose panel members with suitable expertise – deliberately including a diversity of views, because the value of the process lies in views being challenged – and appoint a facilitator to run the process.

3. Panel members are kept separate and answer questionnaires anonymously. Questions can ask members to rank items, assign probabilities or make estimates, with room for open-ended comments and justifications. Separation and anonymity matter: no member is inhibited from expressing an honest opinion, opinions stay independent, and group-think is avoided.

4. The facilitator collects the questionnaires and summarises the responses.

5. The summaries – including the arguments given – go back to the (still separate) members, who reconsider and complete another questionnaire. Anonymity again helps: members can change their minds without embarrassment, and changing your mind when the evidence points the other way is exactly what the process wants to allow.

6. The cycle repeats until the responses have substantially converged on a consensus.

2.1 Strengths and weaknesses

3 . Game theory

Game theory is used for strategic decisions in competitive markets, where the outcome of your choice depends on what competitors choose – and they know that, and are reasoning about you in the same way. It models well-defined scenarios involving a number of players (individuals or firms), each making specific choices, with a payoff calculated for every combination of choices. It is suitable wherever choices and outcomes can be modelled – sealed-bid auctions, pricing moves, capacity decisions.

It assumes that players act rationally and in the long-term interest of their organisation, and that the payoffs can be estimated. Games are typically run many times to explore different strategies. Note that not every game has a winner and a loser: in non-zero-sum games both parties can gain – a component supplier and its customer both profit if quality improvements increase end sales, and knowing this shapes how they negotiate the supply price. Where co-operation is legal, such games reward it.

3.1 Example: a pricing game

The value of the analysis is that the best strategy has been identified by explicitly working through the moves available to competitors, not just to ourselves.

4 . Regression (trend) analysis

Linear regression fits the best straight line through a set of points. In business the points typically relate:

• Cost and volume

• Selling price and sales volume

• Hours worked and units produced

Regression produces constants for a line of the form:

y = ax + b

where y is the dependent variable (cost, hours, volume sold) and x is the independent variable (units made, selling price). The constant a could be the additional cost of each additional unit made (variable cost per unit); b would be the cost even when nothing is made (fixed cost).

4.1 Caution: how good is the line?

Linear regression will draw the ‘best’ straight line through any set of points. Number the days of the year 1–365, note the day each person was born and their bank balance, and regression will still dutifully fit a line – but the relationship is meaningless. To test a relationship, calculate the coefficient of correlation (r) or the coefficient of determination (r²).

• r = +1: perfect positive correlation – all points on the line; as one variable increases so does the other.

• r = −1: perfect negative correlation – all points on the line; as one variable increases the other decreases.

• r = 0: no correlation at all.

r² states how much of the variation in one variable is explained by the other: if r = 0.7, r² = 0.49, so about 49% of the change in y is explained by changes in x – and the remaining 51% is due to other factors.

Before relying on any regression-based prediction, remember:

• If r (or r²) is low, the variables are not well associated and predictions are liable to be poor.

• The more points the better – two points prove nothing; a near-straight line through twenty points is real evidence.

• Extrapolation – predicting outside the range of the observed data – is dangerous: there is no evidence of how the relationship behaves out there (costs might suddenly step up).

• Other known influences, such as inflation, should be stripped out first (restate all costs in current terms).

• Even good correlation does not prove cause and effect: both variables might move together under a third influence. Ice-cream sales and sunburn correlate – because of sunshine, not because ice cream burns.

5 . Time series analysis

A time series is any variable that moves with time: sales each day, rainfall each month, machine breakdowns each week. A time series has four components:

• The trend – an underlying increase or decrease.

• Seasonal variations – regular variations with a cycle of less than a year (days of the week in a shop; quarters of the year).

• Cyclical variations – regular variations with a cycle longer than a year (economic cycles) – too slow to be captured usefully for most planning.

• Random variations – irregular and unpredictable; nothing can be done about them anyway.

Time series analysis tries to capture the first two: the trend and the seasonal variations. Forecasting then means extrapolating the trend and applying the seasonal variation appropriate to the period being predicted. The figure shows a rising trend with regular seasonal variations around it:

As with regression: historically reliable behaviour is no guarantee that the same patterns will continue into the future.

6 . System modelling

The syllabus also names system modelling: building a working model of the business (or part of it) so that assumptions can be changed and consequences explored before committing real resources. In practice this ranges from a linked spreadsheet financial model of the organisation – volumes, prices, costs, capacity, cash – through simulation models that add probability distributions to key inputs and run thousands of trials, up to sophisticated ‘digital twin’ models of operations. The strategic uses are:

• What-if analysis – change one assumption (a price cut, a wage settlement, an exchange-rate move) and read off the effect on profit and cash;

• Sensitivity analysis – find which assumptions the outcome is most sensitive to; those are the ones to research, monitor and hedge;

• Testing option combinations – strategic options interact (Chapter 11’s integration requirement), and a model shows whether a proposed combination is financially coherent;

• Supporting scenario planning – each scenario (section 8) becomes a set of model inputs, so a quantified plan exists for every plausible future.

A model is only as good as its assumptions and its logic – models give estimates of consequences, not guarantees – but a tested model turns strategic debate from assertion into analysis.

7 . Value drivers and the data to measure them

When options are generated and developed, the organisation needs to understand its value drivers – the factors that actually generate value – and what data would measure each. Chapter 11 drew the tangible/intangible distinction; the table below adds the measurement data, which links directly to the KPIs of Chapter 12:

Intangible drivers now create much – often most – of corporate value, yet they are precisely the ones conventional accounting data ignores. Naming each driver and its measurement data is what makes options comparable: an option that builds an intangible driver can then be argued for with evidence rather than faith.

8 . Scenario planning

Scenario planning means building plausible, internally consistent views of how the business environment might develop, based on the key drivers of change about which uncertainty is highest. The emphasis on plausible and internally consistent is what makes the technique manageable: it takes everything that could happen and distils it into a small number of believable alternative futures worth planning for.

Why is distillation needed? With N uncertain variables, each with just two outcomes, there are 2^N permutations – four variables give 16 futures, and no planner can sensibly prepare 16 plans. Suppose the key drivers are which party wins an upcoming election, and whether interest rates will be 3% or 7%:

Four permutations – but not all are plausible. Party A favours high government spending, must borrow heavily, and heavy borrowing pushes interest rates up; Party B favours austerity and low borrowing, keeping rates low. So two permutations are internally inconsistent:

Planning collapses to two coherent scenarios – ‘A wins, rates 7%’ and ‘B wins, rates 3%’ – and business plans and forecasts can be devised for each (using the system models of section 6).

8.1 Steps in scenario planning

1. Define the scope: what time horizon, which part of the business, which countries?

2. Identify the factors that will most affect the organisation.

3. Establish the plausible outcomes for each factor.

4. Combine them into plausible, internally consistent scenarios.

5. Work out the effects of each scenario on the organisation.

6. Decide how the organisation would respond under each scenario – move up-market or down-market, retreat to an internet-only presence, withdraw from a market, and so on.

7. Go back and re-check assumptions and scenarios in the light of the decisions made, as some responses will themselves change the environment.

8.2 Advantages and disadvantages

Advantages:

• Challenges managers and other stakeholders to be more forward-looking in business planning.

• Challenges assumptions about the future and about the drivers and forces influencing the industry.

• Forces consideration of previously unimagined possibilities, and tests the rationale for current strategies.

• Does not attempt to forecast the future – almost certainly a lost cause. Instead it produces several outcomes, each with a planned response.

• Encourages communication and planning across the company, since scenario building needs inputs from many areas.

• Encourages the identification and evaluation of weak signals – small signs of change with little current effect that might be game-changing later. Early users of social media saw personal chatter; a few foresaw its transformation of marketing.

Disadvantages:

• Very time-consuming: identifying relevant variables, collecting data and distilling plausible scenarios.

• Experts are needed to assess possible outcomes, and experts’ time is expensive.

• There is a temptation to keep only the most likely or most attractive scenario. That defeats the point: the purpose is a range of futures – favourable and unfavourable – each with a plan.

9 . Real options

Real options give managers flexibility when committing to projects and contracts. Because the future is uncertain, flexibility has value – and it is usually worth paying for. The main types:

• Option to delay (defer). The right to postpone a commitment until better information exists. Obtain planning permission for a factory now but build only if the economy justifies it in three years; pay a landowner for first refusal on a plot rather than buying it today; delay signing an exclusive five-year distribution agreement while the product proves itself in other markets.

• Option to expand (scale up). Designing a project so it can be enlarged cheaply if it succeeds – building the factory so the back wall can be knocked out and the building extended, or acquiring capacity or land beyond immediate needs. A small extra cost now buys the right to scale later.

• Option to follow on. The right to a successor project on known terms – renewing a five-year lease for a further five years, or renewing that distribution agreement when it expires. Follow-on options matter where ending an activity would be disruptive or where success creates a further opportunity.

• Option to switch. The ability to change inputs, outputs or methods – equipment that can run on two fuels, a plant that can be switched between products, contracts that allow a change of supplier. Valuable where relative prices or demand patterns are volatile.

• Option to abandon. The right to escape early from a project or contract – the classic example being a break clause allowing a five-year office lease to be ended after two years. Worth paying extra for at the negotiation stage, in case the business does not develop as hoped.

10 Test your knowledge

Two short exercises close the chapter in the online notes: ten flashcards on the terms and frameworks above, and ten practice questions with worked feedback on every option. Work through the cards first, then the questions.