ughhhh - you mis-understood my comment!!!!
Check out the WSTS website to see what they are all about - namely the role that they play in gathering and reporting worldwide semiconductor sales. It is this data that my model employs in order to generate my monthly forecast estimate updates. Obviously I believe it follows (logically) that the higher the accuracy of the source data (global semi sales as published by the WSTS each month) the potentially higher precision that can accrue in the forecast numbers. If three significant digits PAST the decimal point disturbs you than only use the first three significant digits of my forecast estimate numbers that I provide. If you would like more info re. the Cowan LRA forecast model than please send an e-mail to firstname.lastname@example.org and I will send you a presentation that gives more insight into the forecast methodology that I employ.
btarng - you are obviously unaware that the WSTS, in its monthly reporting of global semi sales, employs six (that is, 6) significant digits beyond the decimal point! This can be seen if you were to look at the HBR (Historical Billings Report), namely its Blue Book, that the WSTS posts on its website each month and ia available for all to see!
Oh by the way, if you were to carefully look at many semiconductor companies quarterly financial statements you would also see that they report three significant digits beyond the decimal point. Consequently, I likewise report my forecast estimates in the same format, that is, three digits past the decimal point.
Therefore, I have to infer that you are being very devious,for whatever purpose, in your comment.
I love how Mr. Cowan's estimate includes six significant digits -- as if one could accurately predict the annual sales of any single company down to a million dollars, much less the revenues of the entire semiconductor industry. Why don't you include the dollars and cents also? I'm sure the mathematical model spits those out as well.
Peter, remember - the model is strictly a pure mathematically based assessment relying on linear regression analysis operating on the past 27 plus years of historical global semi sales as gathered, tracked and reported by the WSTS. Therefore, the model is devoid of any economic assumptions or biases and thus abstracts its quarterly and yearly sales forecast predictions from the industry's historical experience as embedded in the monthly actual global sales numbers covering the past 27 plus years, that is, from 1984 through 2011 YTD.
You are on the bullish side of most with that 3 percent growth forecast but you may yet be proved right if some restocking goes on in 4Q11. On the other hand, here in Europe the macro-economic outlook is not good. :/
Peter - the just updated Cowan LRA Model run which is based upon September's (actual) global semiconductor sales of $29.442 billion yielded a sales forecast estimate of $307.258 billion. This latest result corresponds to a 2011 sales growth expectation of 3.0 percent which is down from last month's forecast prediction of 3.9 percent even though September's actual sales number was a 5-week month all-time record high value. Interested readers can send an e-mail to email@example.com requesting the details of the latest model run numbers.
Mike C. (independent semi industry analyst and developer of the Cowan LRA forecast model)
Blog Doing Math in FPGAs Tom Burke 15 comments For a recent project, I explored doing "real" (that is, non-integer) math on a Spartan 3 FPGA. FPGAs, by their nature, do integer math. That is, there's no floating-point ...