Friday, March 27, 2009

How we forecast 2009 even with low visibility

Now that our laser market report is out, it's helpful to explain how we can do forecasts for 2009 when the visibility is only about 3 months. I won't go into all the detail about forecasting. I want to focus on how we used quarterly data to narrow the range of forecasts.

We have a lot of information about suppliers and customers for quarters up to December, 2008. What's striking is how 2008 was generally ahead of 2007, even though the last quarter was bad for most everyone. This varies from sector to sector, but for the most part it's true.

The figure shows actual laser revenues for several selected companies through Q4 2008. It then shows what happens if all the quarters in 2009 are reset to a level near or below Q4 2008. As good as 2008 was overall, 2009 comes out well below that. Continuing the decline into Q1 2009, but no further, yields a decline of 32% for 2009 over 2008. It turns out that this supplier-side approach also matches our projections for 2009 in general from the demand side, aggregating all the different segments together.

What's useful about this approach is not that it predicts each quarter (it doesn't), but that it narrows the range of plausible projections. Of course, anything is possible in 2009, but running simple scenarios like this helps to test a hypothesis.

It also allows us to update the range of possible outcomes as the year proceeds. By Q4 2009, the error bars for 2009 will be much narrower, while the focus shifts to 2010.

It also illustrates the range of outcomes from, say, a recovery, or of a price war. Either one could move the overall 2009 revenues up or down, especially if it happens early in the year. A recovery late in the year won't affect the overall 2009 revenues as much.

This is an example how we use an analytical tool, and this one is especially useful when the market is in a period of rapid change, as it is today.

No comments:

Post a Comment