The Basics of Scenario-based Forecasting and Planning
Forecasting is becoming an increasingly important issue for private and public sector decision makers. As economies become more productive and efficient, there is increasing recognition of the very real costs associated with being unprepared. Intense competition makes even small missteps costly in most industries.
We all want to know what the world will look like next year, in five years, or in 25 years. Good examples of rich forecasting dilemmas can be found in any daily newspaper:
- What will the U.S. federal deficit be next year?
- How much oil will we have in 2020?
- What will be the impact of global warming?
- Will Social Security be around when I retire?
- Will Company X hit their earnings number next quarter?
Some of these questions seem almost unanswerable due to their complexity and the many unknowns they pose. Each of the questions involves dozens of factors that can change the ultimate outcome. Yet there are good analytical approaches to addressing the unknowns and breaking down the complexity posed by such tough forecasting questions.
The most basic forecasting device is the assumption. Everyone uses them, right? Assumptions help us break down complexity and uncertainty.
A fancier term for assumption is "scenario". Scenario implies a series of assumptions that are being considered to create a forecast. In the messy world of people and behavior, there can be no forecast without a scenario.The only question is whether to make your assumptions explicit (viewable, known) or implicit (buried, hidden, etc.). All things being equal, (big assumption!) the more your assumptions are known the better your ability to forecast with accuracy.
Use care when creating scenarios. In the 1980s, the U.S. Administration was famously accused of unrealistic budget forecasts due to the use of favorable scenarios, dubbed the Rosy Scenario. These scenarios used optimistic assumptions about economic growth, consumer income, productivity, etc. to generate forecasts that were politically palatable. A "blackbox" forecast might have hidden the scope of this optimism, but people are now sophisticated about forecasting. They know it's all about the scenario.
Once a scenario for the future has been prepared,we then must decide what impact those future conditions will have, e.g., the forecast. There are many different ways to express the impact of a scenario into a forecast, ranging from purely intuitive to more rigorous quantitative methods.
Most modeling approaches found in business forecasting are cumbersome to use with many different scenarios. They tend to be geared toward one set of assumptions or require a lot of work to recalibrate to a new scenario. Moreover, most models tend to be trend-following: e.g., they slavishly extrapolate the trend most recently experienced (an implicit assumption that the future will look like the past). This is a problematic feature when experience tells you that the future environment is likely to be highly dynamic. To handle scenarios well, a modeling technique should make all assumptions explicit and thereby under user control and inspection.
Every business forecast contains assumptions. Despite increasing sophistication about forecasting, it is human nature to focus more on the result instead of the assumptions that drive the forecast. For example, it's a well-known secret that most organizations' planning & budgeting processes start with the number they want, and then build backwards to a plan that justifies that outcome.
Why? First, it's more work to do it the other way. Ideally, core drivers should be identified and understood, data collected, and benchmarks set. The effort becomes a built-in, year-round process instead of an ad hoc checklist item. A modeling technique must be adopted that can tie together events and their impacts with statistical rigor. Organizations that make this effort get a far more accurate and valuable planning process—better visibility on results and the factors contributing to them—and a competitive advantage over those reacting to change as it hits them.
The firm that has internalized scenario-based planning can easily test its operational plan monthly or weekly instead of once or twice a year. A sensitivity analysis of the business (e.g., a stress-test) can be stated in terms such as "what is the impact of a 1-point decline in GDP?" or "impact on profit from increased capital requirements stemming from a two-year decline in consumer incomes."
Scenario-based planning reinforces the concept of risk management over risk avoidance. Without the ability to plan quickly across all contingencies, firms don't know when they are taking too little risk, or conversely, too much. Being able to forecast the full range of outcomes is the only way to know if you've reached an optimal point.
The next time someone hands you a forecast, optimistic or gloomy, remember to ask: "Under what scenario did you make this forecast?"