Semiconductor Components - Excerpt

Methodology

This market forecast consists of several parts:

o	forecast of assembly of vehicles by region,
o	forecast ratio of systems produced per vehicle assembled by region,
o	forecast ratio of component value per system produced for each component by region.

Component value is derived by multiplying:

	Vehicle assembly
		times
	System-per-vehicle ratio
		times
	Component value-per-system ratio.

Components totals are cross-checked in three ways:

o	supplier survey of semiconductor shipments by product and application into automotive by region
o	supplier survey of total automotive shipments by region,
o	annual end-use survey from semiconductor suppliers.

Supplier Survey

Based on TIER ONE's prior experience, the most accurate and least costly way to calibrate a market was to survey suppliers. These are well known and tend to collect common data, whereas it is easy to miss a user.

Calendar Year was chosen over Model Year or Fiscal Year. Generally, automotive components change only slightly with Model Year.

Primary research was conducted in North America, Europe, Japan, Korea, Taiwan, China and Brazil. Professional market researchers interviewed:

o	Semiconductor suppliers in each region for historical shipments
o	Automotive analysts for system installation rates

Additional installation rates were reviewed from published sources such as Automotive News, Wards and Autofacts.

Shipments from unknown suppliers in Eastern Europe and China were estimated from: intallation rates on similar vehicles built elsewhere, information gathered from interviews, and ratioing from known good data.

Research in 1993-94 concentrated on detailed breakouts for 1990-1992. These where a combination of personal and telephone interviews, with data exchanged by fax. Data from each company was analyzed for reasonabaleness, missing aftermarkets, application detail, and estimating accuracy. Initial estimates frequently changed when asked to spread shipments by geographic region. Regional managers of the larger companies were separately interviewed for applications insight. Follow-up interviews in 1995, 1996 and 1997 by telephone and fax concentrated on total shipments for years 1993-96, plus major new applications and technology.

In 1996 the database was converted to Excel 5.

Survey Analysis

Comapny data for 1990-1992 was developed by product, application and region, individual line items were analyzed for incomplete data, wrong categories, unlikely ratios, and changes made. Data for all companies was then summarized into a single total by product and application.

Application totals were compared with OEM installation rates and aftermarket sales in each geographic region. In certain product areas a number of adjustments had to be made to account for extensive importing and exporting.

Product summaries by major product lines were compared with previous semiconductor surveys. This required construction of models for each survey: who participated, what was measured, what was missing, sources of known error, sources of unknown error, keypunch error.

Products were then adjusted until the sum of products equaled the sum from suppliers, and was consistent with top-level end-use surveys fromWSTS (World Semiconductor Trade Statistics, Sunnyvale, CA).

Definitions 

Three category lists (Suppliers, Products, Applications) were subject to the following conditions:

o	Theshold for inclusion of an application, supplier or product of roughly $1 million.
o	All tables provide Value only, not ASP or Quantity
o	Supplier listing is intended to be complete outside of the China/Soviet sphere. Included are captive suppliers Delco, Bosch and Denso.
o	Subsidiaries are generally summed into the parent organization
o	Total Semiconductors by supplier is made public
o	Product and system categories must be clearly defined, easily measured and meaningful to the automotive world.

Forecast Procedure

The forecast is obtained in a sequence of steps for each region.

1.	Estimate the number of electronic systems produced by multiplying together: (lists)
2.	For each system, estimate the average value of each component over the forecast period. This parameter represents (average selling price) X (no. of components per system).
3.	For each component, multiply (system quantity) X (component value per system) to obtain component value for each system.

This parameter is stored in the database.

 

TOC | Press Releases | Inquire