Can SEMI find some
12/01/1998
Peter N. Dunn
Editor & Publisher, WaferNews, [email protected]
Can SEMI find some "leading indicators"?
There was a poignant pair of sentences in the statement issued by SEMI`s All Industry Forum, held on the eve of SEMICON Southwest to discuss topics for industry cooperation. The second paragraph in the "Capacity Forecasting" section states, "It is agreed that the SEMI staff will undertake a review of available industry data that might provide some `leading indicators` of what is to come and then to hold quarterly meetings of industry executives to review these indicators and agree on how to interpret them. `It is not so much a lack of information that is available, what is missing is the ability to properly interpret what that information should tell us about the future,` it was said."
This statement clearly reflects the feelings of a shell-shocked community that has recently dealt with some less-than-perfect forecasts, both from private market researchers and from chipmakers` plans reported by SEMI`s own 300mm Initiative. It`s hard enough to run a profitable equipment or materials business when you have a general sense of future market conditions. When predictions are off the mark, it turns into a real mess: layoffs, wasted R&D investments, facility planning nightmares, and much more.
I really hope the plan works - that the leading indicators can be tracked down in a timely fashion and that an assembly of experienced executives can look at the same numbers and come to the same opinion of what they mean. As I have mentioned (WaferNews, March 2, 1998), SEMI`s monthly bookings and billings report provides one of the few forward-looking indicators in the entire semiconductor universe. If the association could extend this into an actual forecast, it would be one more data point, and could help generate more attention for the equipment and materials sector, much as the Semiconductor Industry Association`s (SIA) annual forecast does for chipmakers.
However, a few caveats should be kept in mind:
A number of companies out there have full-time staffs of bright people who forecast trends in
semiconductor and equipment sales and technology. Usually their predictions are close to the actual numbers, but sometimes they`re off. In general, this is because any economic system based on the free market is by definition chaotic and subject to sudden change by unexpected factors. It is easy to be overtaken by events. As I noted recently (WaferNews, Sept. 18, 1998), Siemens` closure and selloff of its UK DRAM fab and TI`s unglamorous departure from memories are perfect examples of how even top-tier companies can misjudge markets. This is no reason not to develop a SEMI forecast, but it is worth remembering that if the task were easy, someone would have done it by now.
If an ideal set of raw numbers could be assembled in a timely fashion for those quarterly meetings, the problem of interpretation arises. Barring development of a perfect analytical algorithm, the assembled execs must reach a consensus. Not
easy - reasonable people can look at the same numbers and draw different conclusions. The process will need a coordinator with a light but firm hand, a diplomatic tongue, and the ability to maintain focus on the quality of the final product.
Finally, what would happen if another downturn came and the SEMI system failed to call it? It is important to maintain moderate expectations, to avoid possible recriminations among members of the executive committee, and against SEMI itself, which presumably would be putting its imprimatur on the data. It is probably worthwhile having some conversations with the SIA; its World Semiconductor Trade Statistics forecasts of the last few years have been over-optimistic, and the organization does not seem to have self-destructed.
Good luck to SEMI in its forecasting project; we will all be watching with interest.
For subscriptions to WaferNews, contact Jill Wood at ph 603/891-9174 or visit www.wafernews.com.