3 ms·

The article doesn't explain the quality adjustment problem well: > The problem is, the processor released in 2017 is superior to that sold in 2016 in many tang

The article doesn't explain the quality adjustment problem well: > The problem is, the processor released in 2017 is superior to that sold in 2016 in many tangible ways. But how do you account for the fact that a 2017 processor provides users with more value? In general, statisticians assume the difference in value between the two models is just the difference in their prices. If, say, the 2017 processor costs twice as much as the 2016 one does, then selling one 2017 processor counts as selling two of the 2016 versions in the statisticians’ books. This makes it sound like the problem is prices rising from year-to-year. The real issue is that the comparison is made with the price of the 2016 processor in 2017. See "methods of quality adjustment" in https://www.bls.gov/ppi/qualityadjustment.pdf https://www.bls.gov/ppi/qualityadjustment.pdf So basically - let's say in 2016 a new CPU costs $200. Then in 2017 Intel releases a new model, prices it also at $200, and sells the old model at $100. The quality adjustment algorithm will see that right now, the new one is selling for twice as much as the old one, and assume that there's been a 2x jump in value per item. [EDIT: One way of thinking about this is: the method treats clearance prices for last year's model as deflation in this particular market, and adjusts for it.]

whack 9y ago
I was thinking about this as well, and it doesn't seem like an easy problem to solve. Subjectively, we know that the reason 2017 processors cost 2x as much as a 2016 processor, is because of improvements in process technology and architectural design. There might be incremental improvements in the manufacturing process, but not enough to say that "manufacturing capabilities have doubled compared to last year". The obvious solution here would be to compare the price of a 2017 processor in 2017, to the price of a 2016 processor in 2016. But that would give hopelessly cynical answers in other circumstances. For instance, consider a factory for paperclips that produced 10M units in 2016, and due to various improvements, 20M units in 2017. The paperclips in both years are mostly identical to one another. But for various market reasons, the price of each paperclip has fallen by 50%. Possibly due to the enlarged supply from factories all over the country doubling their production. In such a situation, if you rated the factory's output using the alternative metric suggested above, you would think that the factory hasn't made any improvements at all in the last year. When in reality, it has doubled its production capabilities and efficiency. There just doesn't seem to be a purely quantitative to compare the manufacturing capabilities of factories that produce qualitatively different products year over year. This seems to make moot any single metric that rates aggregative manufacturing capabilities across multiple years.
HillaryBriss 9y ago
I'm probably misunderstanding or oversimplifying somehow, but, isn't it also a problem for these types of economic estimates that sometimes the new chips are actually cheaper than the the old chips in today's real world markets? Example: Right now Amazon is listing the overclockable Skylake Core i7 6700K 4 GHz 4 core at $392.76 https://www.amazon.com/Intel-Unlocked-Skylake-Processor-BX80662I76700K/dp/B012M8LXQW/ https://www.amazon.com/Intel-Unlocked-Skylake-Processor-BX80... Also right now, the newer, overclockable Core i7 4.7 GHz 6 core 8700K is offered at $347.06. https://www.amazon.com/Intel-i7-8700K-Processor-Unlocked-BX80684i78700K/dp/B07598VZR8/ https://www.amazon.com/Intel-i7-8700K-Processor-Unlocked-BX8... The newer chip has two additional cores and a higher peak frequency, but it's about $50 cheaper. Would an economist conclude that this means industry output went down in this particular case? How do economists work this sort of market-pricing reality into their estimates of computer industry value?

whack 9y ago

I was thinking about this as well, and it doesn't seem like an easy problem to solve. Subjectively, we know that the reason 2017 processors cost 2x as much as a 2016 processor, is because of improvements in process technology and architectural design. There might be incremental improvements in the manufacturing process, but not enough to say that "manufacturing capabilities have doubled compared to last year". The obvious solution here would be to compare the price of a 2017 processor in 2017, to the price of a 2016 processor in 2016. But that would give hopelessly cynical answers in other circumstances. For instance, consider a factory for paperclips that produced 10M units in 2016, and due to various improvements, 20M units in 2017. The paperclips in both years are mostly identical to one another. But for various market reasons, the price of each paperclip has fallen by 50%. Possibly due to the enlarged supply from factories all over the country doubling their production. In such a situation, if you rated the factory's output using the alternative metric suggested above, you would think that the factory hasn't made any improvements at all in the last year. When in reality, it has doubled its production capabilities and efficiency. There just doesn't seem to be a purely quantitative to compare the manufacturing capabilities of factories that produce qualitatively different products year over year. This seems to make moot any single metric that rates aggregative manufacturing capabilities across multiple years.

HillaryBriss 9y ago

I'm probably misunderstanding or oversimplifying somehow, but, isn't it also a problem for these types of economic estimates that sometimes the new chips are actually cheaper than the the old chips in today's real world markets? Example: Right now Amazon is listing the overclockable Skylake Core i7 6700K 4 GHz 4 core at $392.76 https://www.amazon.com/Intel-Unlocked-Skylake-Processor-BX80662I76700K/dp/B012M8LXQW/ https://www.amazon.com/Intel-Unlocked-Skylake-Processor-BX80... Also right now, the newer, overclockable Core i7 4.7 GHz 6 core 8700K is offered at $347.06. https://www.amazon.com/Intel-i7-8700K-Processor-Unlocked-BX80684i78700K/dp/B07598VZR8/ https://www.amazon.com/Intel-i7-8700K-Processor-Unlocked-BX8... The newer chip has two additional cores and a higher peak frequency, but it's about $50 cheaper. Would an economist conclude that this means industry output went down in this particular case? How do economists work this sort of market-pricing reality into their estimates of computer industry value?