I was pleasantly surprised this morning to receive a note from IPC indicating a change to its data collection methods.
In an announcement touting its new market research subscription series, IPC said it would use aggregate company data along with “carefully vetted secondary research.”
That sounded like IPC would now use data from the publicly traded EMS companies to help flesh out numbers its participating members report. A quick inquiry to IPC director of market research Sharon Starr confirmed as much.
“The growth rates reported each month will be based on IPC’s survey sample, but we will be looking at published data from the publicly traded EMS companies as a cross-check and may also use this data in our market-size calculations. If growth rates from the publicly-traded companies differ significantly from what our survey participants are telling us, we will address that in the reports,” Sharon told me.
This represents a departure from the past 20 years, where IPC would rely only on the data directly reported by survey participants. For years, report after report showing various manufacturers’ monthly sales numbers would arrive on an unsecure fax machine in the IPC office. Eventually, that fax was moved to someone’s office, but there was little in the way of confidentiality offered, and even if the the reporting company used a unique code, its name on the top of the fax always gave it away. As companies went public and disclosures became more scrutinized, there was a real disincentive for manufacturers to submit monthly data. Over the years, this trend undermined the usefulness of the IPC data: if larger companies weren’t reporting, were the figures as published truly representative?
Parsing the publicly traded firm’s quarterly numbers will go a long way toward ensuring that the data reflect the macro industry trends.
In doing so, the site acknowledged recent announcements of sequential gains by semi gear makers. More interesting, however, was its reliance on data from three quarters ago, which show “turning points in semiconductor equipment sales.”
“Both our long and short indicators turned up in late 2008, pointing to a business recovery cycle and giving visibility that the days of the recession are numbered,” analyst Robert Castellano wrote.
Since chip sales are a leading indicator for electronics assembly, that’s certainly good to hear.
This is a much wiser, useful way to approach forecasting. Whereas some may call it a copout — by providing a range, iSuppli gives itself a greater chance of being “right” — I see it as a long overdue move into what is standard practice elsewhere. Indeed, providing a range is a method many researchers in other fields use to describe what might actually happen.
The next step would be to offer probabilities for the different scenarios; in other words, 50% of the time, sales will be up by X%; 75% of the time, sales will be up by Y%, and so on. But I’ll settle for other research firms picking up on what iSuppli has started. Kudos!