Wednesday, March 30, 2011


Oh my this post is very rude. I will delete some of the more extreme parts (with indication). In general, I wish I hadn't written it (plus comments at blogs which I can't delete). I will type new text using italics, because I don't want to type update dozens of times.

Karl Smith seems not to know the referent of the acronym GDP and to feel free to redefine "productivity" at his whim.

Economics can not survive economists who redefine terms at will.
Wow what was I on yesterday ? (answer nicotine withdrawal). It's not like I think economics is in such great health to begin with.

He wrote

"The only way to get GDP wrong is either to miscount the number of goods and services sold in the US or to misestimate the price index of final goods – not intermediate goods."


I don't agree with one of your claims at all "The only way to get GDP wrong is either to miscount the number of goods and services sold in the US or to misestimate the price index of final goods – not intermediate goods." This can't be true, since not all intermediate goods are domestically produced.

The BEA starts with nominal quantities. To calculate real value added by sector, in manufacturing it double deflates. I am going to assume it hasn't changed since I last checked, my information is out of date, but I am sure that the range of sectors for which it uses double deflation has increased or remained the same, since double deflation is the method it considers best.

I tried to look this up yesterday. I think they double deflate everything now, except for flahs estimates. I might be wrong. I am not going back to BEA methodological notes today (the thought makes me gag)

it calculates nominal final production and nominal materials inputs. It divides nominal final production by a price index and it divides materials inputs by a price index. then it subtracts the second quotient from the first. Then it adds the resulting numbers up over sectors.

An error in a materials price index implies incorrectly measured GPO and incorrectly measured GDP. If one person at BEA were responsible for a sector and typed in the wrong number for materials prices for that sector, then reported GDP would be incorrect.

I can't understand how you can imagine that only final production deflators affect calculated GDP. Calculated GDP is, in principle, a funcition of nominal quantities by sector, final product deflators, and materials deflators. In sectors other than manufacturing the BEA has in the past made absurd assumptions in order to come up with value added numbers without double deflating. For some sectors they have made assumptions about labor productivity so, in fact, and directly contrary to what you said, GDP is calculated from assumptions about labor productivity. The preceding two statements are true, but irrelevant, if I understand correctly that this is all in the past. They do help explain my mistrust of the BEA, but I think they've cleaned up their act.
Please check my note in the Journal of Polit
ical Economy vol 99 no 6 pp 1315-1321 in which I explaining how BEA measured GDP is systematically wrong for references to explanations of how they got the numbers (I did admit that my information was out of date ).

Note that the parenthetical comment isn't in italics. I wrote that yesterday when in a huff but not freaked out enough to assume that 20 year old information is current. I now don't think there is any point looking up that old note (which is hardly ever cited) because, as far as I know, the BEA has improved it's approach.
But their best numbers are created by double deflation and are to incorrect if they use the wrong materials price index. This is clearly explained by the BEA.

This may be a shock, but it isn't unusual at all for an economist to have no idea how the numbers we use (such as GDP) are calculated. It is shocking that an economist dismisses a detailed argument about the numbers which demonstrates unusual knowledge on the basis of the assumption that the method is obvious and there is no need to look it up.

Oh my I left all of the substance of the argument out of my post yesterday. My comment at modelled behavior explained that a key problem with materials prices is imported and exported materials. Imported materials are not domestic products and overestimating the increase in their prices implies underestimating the change in the quantity and overestimating production and productivity growth. Also net inventory investment is part of GDP including inventories of materials. On average inventory investment is a tiny share of GDP (and most inventories are of retailers of what we generally considered finished goods even though they are really some of the materials used by retailers). However net inventory investment is huge and negative in recessions and huge and positive early in recoveries. This means that getting the price of inventories wrong implies large errors in recent calculations.

Finally, and my main problem with the claim, the correct price index for intermediate goods sold by sector A to sector B is typically different from the correct index for goods sold by sector A to say consumers or to the government or to firms which are investing. As far as I know, the BEA just assumes that all of these indices are the same. This is an approximation which makes the numbers they generate different from the theoretical quantity true GDP. I try an example, say sector A makes equal amounts A1 sold to consumers and A2 sold to sector B which makes B sold to consuers. I think the BEA calculates (price A1 + price A2)/2. If the price of A1 goes up and the price of A2 goes down, then the BEA will report a spurious increase in GDP. Here the point (if any) is that "the price of final goods"
isn't determined from producer price indices unless the price of materials is known, so this point (if worth anything) is an argument that Smith's claim doesn't refute Mandel and not that it is false as written.

But that is far from the worse thing that Smith did. He decided that he can redefine productivity as he pleases. To him, evidently, it means "something good." He wrote

To the extent Apple can squeeze it suppliers while producing really high valued products this seems like the essence of productivity growth. We have both lower input costs and higher value for the consumer.

This seems to me to be a rejection of any thought that economics might be a semi serious intellectual effort in which technical terms are used with agreed meanings. I think that economists should not tolerate such an abuse of our language. Now I am not discussing violence or revocation of tenure or anything, but I don't think that any economist shouls refer to Karl Smith as an economist until he has abjured his offence against economics. Wow I was totally off my nut yesterday.

I can explain my extreme distress. The word which bothered me was "squeeze." To me, to squeeze suppliers is to drive a hard bargain and pay a low price. In increase in such squeezing is a change in a price. A change in a price with no change in quantities is not a change in production or productivity according to the standard definition. I wouldn't object to "if Apple invents a new product which is valuable and made of cheap materials then productivity has increased." My objection is entirely based on my interpretation of "squeeze". Now this is not an economists term of art. If Smith uses "productivity" the way I use it and "sqeeze" with a different meaning, then we can communicate just by avoiding slang.

If the BEA over under estimates the increase in the price of imported intermediate goods, then it underestimates the growth of GDP.

Of course it is very likely that a fall in price of goods exported from China to the USA is made possible by an increase in productivity *in China* (it could also be reduced profits for the Chinese manufacturer or reduced wages for the workers or something). But increased productivity in China has nothing to do with US productivity or GDP.

He who knows and knows that he knows is a teache. Learn from him.
He who knows not and knows that he knows not is a student. Teach him.
He who knows not and knows not that he knows not is a menace. Avoid him.

The above three lines still seem relevant, but they don't seem to me to refer to Smith.

No comments: