15 Jun 15 Forecasting market growth and the recency effect
For businesses of all kinds, yesterday’s success is history. What matters is the future.
As a result, competitive and market intelligence analysts are routinely tasked with forecasting market (or product) growth. Very often, business plans are built on such forecasts. But if you ask the users of these estimates, there is more than a healthy skepticism. After all, who can predict the future?
Despite this skepticism, all kinds of companies, divisions, businesses and many, many of their executives spend hours trying to forecast markets. It is something that few believe we can do (or do well), but we need to do it regardless.
Regardless of models or methods used for market sizing and forecasting, there is one common failing with most of the forecasts I’ve seen – whether from CI analysts, equity analysts or industry associations.
And that arises from the simple straight-line extrapolation of the very recent past.
There are two issues here: the “recency” effect, and the straight line.
If the last year was bad, we tend to be pessimistic. And after a couple of years of high growth, we’re even more optimistic. This is often due to a “recency effect”, where the most recent events tend to overshadow those of greater vintage. It’s also easier for the lazy analyst, and usually acceptable to their bosses or clients. Few are ready to take the risk of forecasting a change in trend.
Extrapolating from the past three or even five years does not make sense. All mature industries tend to be cyclical – even in high growth markets. Of course, the intensity (amplitude) and length of cycles will vary substantially. For newer businesses, we can use product life cycle models, and other tools that might help depict the curve, if not its velocity and acceleration.
Predicting cycle behavior may be as hard as growth forecasts, but the fact that a cycle turn will happen in the medium term is in itself a big input for a decision maker. For instance, in the auto sector in country x, if we’re able to say that we’re likely to see at least one big boom in the next decade followed by a slump – this might be more valuable than a prediction of a 10% average CAGR through the period. Markets never grow in a straight line. The shape of the curve is arguably as important as the average growth rate over a period.
In a recent research assignment, we played with the data in different ways. First, borrowing techniques from scenario analysis, we constructed curves that paralleled boom-bust cycles seen in the past. These had varying lengths (time) and amplitude (highs and lows). We fitted these curves onto long term trend-line growth for the auto sector – and arrived at different scenarios. An analysis of lead indicators and macro-economic factors that drive spending on automobiles, allowed us to derive rough probabilities for each scenario.
These helped our client much more than a mere growth rate prediction would have. With forecasts so difficult to be right about – decision making was benefited by a structured approach to analyze different possibilities.
Moral of the story: Don’t think in straight lines