RE: Re: Re: Those questionable productivity numbers
RE: Has Dean Baker or anyone else done a detailed analysis of the impact of "quality" adjustments to the last few years' productivity and output data? It would be nice to have some sensitivity analysis... Is the idea that price indexes understate the growth of quality (as claimed by Boskin et al)? And, if so, that productivity measures are understated when using these (too rapidly growing) price indexes. But little compelling evidence exists that quality improvements are understated when price indexes are generated. Eric Eric Nilsson Economics California State University, San Bernardino San Bernardino, CA 91711 [EMAIL PROTECTED]
RE: Re: RE: Re: Re: Those questionable productivity numbers
Doug wrote: The price indexes for computers are truly stunning, turning nominal increases of 5-10% into real increases of 50%. U.S. GDP growth without computers over the last year is 5.2%; with, 5.7%. In the GDP accounts, final sales of computers grew $24 billion in nominal terms (99Q2-00Q2), which was inflated into $131 billion in real terms. ^^ I understand the first sentence - it claims that computer related stuff is more-or-less 50% better now than in the recent past. (I question this, though, for most of my uses 1989 WordPerfect worked better than 2000 Word and I _regularly_ have to reinstall Windows 98 on my home computer because the operating system starts doing strange stuff after just a few months - my God my lost productivity during that period of time! I could have, say, mowed the lawn. I never had to reinstall DOS. Are not most productivity measures for computer are based on raw computing power? Actual improvements in what the product does for the consumer are much less than this. But I don't understand how nominal $24 b becomes real $131 b. Eric
RE: Re: Re: Those questionable productivity numbers
Doug wrote, first, In the GDP accounts, final sales of computers grew $24 billion in nominal terms (99Q2-00Q2), which was inflated into $131 billion in real terms. Then, See the spreadsheet at http://www.bea.doc.gov/bea/dn/comp-gdp.exe. I looked at this spreadsheet. It is very hard to follow as BEA did not clearly explain what the numbers were but I have some vague ideas having done something generally similar for different types of data in the past. Be that as it may, I see where you got your $24 nominal number and your $131 billion real number. I think, however, that the author of the spreadsheet does _not_ intend for you to compare the 24 with the 131. Unlike some NIPA tables, the real numbers for a given year in this spreadsheet are _not_ intended to be generated from the nominal data in any straightforward fashion. More concretely, note the nominal values over 1999II to 2000II grew from 91 to 115. The real values over the same period grew from 232 to 363. Since the starting values (91 and 232) are not the same, you can't simply compare the absolute growth over the two periods in any reasonable way (24 versis 131). But if you 'deflate' the 232 to equal 91 for the real series (and do the same deflation for the 363), then you find the new 'real absolute' growth in the real numbers is now about 50. This 50 might now be compared to the 24. (That is, you multiply both 232 and 363 by the ratio 91/232 to get about 91 and 141 and the difference between these is 50). The upshot of this is that implicit in the BEA data is not . . .it implicitly claims that stuff gets 5 to 10 times better over the course of a year. but given the 50 real growth versus 24 real growth results above this is more like 2 (given your approach). But this is, I think, not a right conclusion (that is, the number 2). If nominal grows 24 -- from 91 to 115 (as in the spreadsheet) -- while real grows 50 -- from 91 to 141 (data in the spreadsheet translated to the same base as with the nominal data) -- then this does not imply a halfing of prices. Rather an approximately 20 percent fall in prices will do the job. Such a decline leads real grow to go from 91 to 141. That is, 115 times 0.8 equals about 141. This implies that -- according to the BEA data -- prices fell by 20 percent in the computer sector between 99II and 20II. Alternatively, things are better but maybe only about 20 percent better. Computer stuff got better about 20 percent over the past year according to the BEA. Using this assumption, they conclude that _without_ the computer sector growth between 99II and 00II was 5.6 percent while growth _including_ computers was 6.0 percent. This is certainly a big difference, particularly over the long run. If anything, some might argue the improvement should be higher than 20 percent. If this is indeed true, then the US economy might have grown even faster than the 6 percent noted above. All the above is tentative. Eric .
RE: RE: Re: Re: Those questionable productivity numbers
Oops. RE That is, 115 times 0.8 equals about 141. Obviously "115 divided by 0.8" makes a bit more sense. Eric
RE: Re: RE: Re: Re: Those questionable productivity numbers
Doug wrote But why can't I compare the 24 with the 131? They're both aggregates, and the comparison shows that some massive inflation of nominal values is going on to produce the real values. These things get truly preposterous over the long term - nominal spending on computers grew 143% from 1987 to 2000, which translates into 3457% in real terms. This strikes me as meaninglessly silly. I'm not sure exactly what the number in the spreadsheet mean. Consider the data on nominal spending on computers in 1987 versus 1999: In the first year nominal spending is indicated as $48.5b while in the latter year it is $92.5b. These numbers seem so off that nominal spending in this spreadsheet must mean something strange. Nominal computer spending only doubled in last 10 years while CPI was up 30% over the same period of time? I'll try to explain more clearly the 24/131 issue in a day or so -- I'm off to be daddy at home for a few days now. Eric .
Re: Re: RE: Re: Re: Those questionable productivity numbers
Not sure what the BLS does with quality declines in air travel. The airfare component has risen much faster that the total CPI over the last twenty years. They are adjusting something. Gene Peter Dorman wrote: Dean's argument is that BLS quality adjustments in the computer industry are much too high, with ramifications for price indices, productivity measures, estimates of the industrial share of production, etc. He has also pointed out that quality *declines* (such as in air travel) have not been taken into account. Peter Eric Nilsson wrote: RE: Has Dean Baker or anyone else done a detailed analysis of the impact of "quality" adjustments to the last few years' productivity and output data? It would be nice to have some sensitivity analysis... Is the idea that price indexes understate the growth of quality (as claimed by Boskin et al)? And, if so, that productivity measures are understated when using these (too rapidly growing) price indexes. But little compelling evidence exists that quality improvements are understated when price indexes are generated. Eric Eric Nilsson Economics California State University, San Bernardino San Bernardino, CA 91711 [EMAIL PROTECTED]