European Tribune - Community, Politics & Progress.

Statistics aren't everything ...

by Colman
Tue Jan 30th, 2007 at 09:55:06 AM EST

John Kay in the FT is inspired by the news that productivity growth in Europe has surpassed that of the US by 0.1% to warn us against the "naive acceptance and popular distribution of economic statistics"

Output seems like a hard number – and would be if it were simply a matter of counting the widgets that leave a production line. But the output of a modern economy is made up of thousands of differentiated products of changing quality and composition. The US productivity miracle was in part created, not by finding new facts about the US economy, but by reclassifying software expenditure as investment and adopting aggressive assumptions about falling computer prices.
The key number used to measure economic performance is gross domestic product. But few politicians or traders could actually define it. GDP is not, exactly, a measure of either business output or consumer welfare, although it is loosely related to both. It is safest to say that GDP is the number you arrive at if you follow an internationally agreed set of statistical conventions.
So long as everyone follows these conventions, movements in GDP tell you something about national prosperity and economic progress, even if it is not entirely clear what. But no economic data, hard or soft, can ever tell the whole story. Prosperity and progress are soft concepts and official statistics are at best a supplement, not a substitute, for evidence of eyes and ears.

Would anyone else like to join me in losing my mind? How come this story wasn't written when 12% German unemployment rates were being quoted, or when US GDP growth was showing how moribund Europe was?

. Home
. About
. Contact
. New User Guide
. FAQ
. Search
. Search (Google)
. Archives (Wiki)
Art, Economics, Energy, Environment, EU Politics, Mech & Tech, By Country

Username:
Password:

. Make a new account
. Reset password

Statistics aren't everything ... | 41 comments (41 topical, 0 editorial, 0 hidden)

Re: Statistics aren't everything ... (4.00 / 2)

When the facts are in your favour, cite them. When they're not, cast aspersions upon them.

by Colman (colman at eurotrib.com) on Tue Jan 30th, 2007 at 10:20:42 AM EST

Re: Statistics aren't everything ... (none / 0)

Oh, and note the implication that Europe's productivity growth is illusional, rather than the other way around.

by Colman (colman at eurotrib.com) on Tue Jan 30th, 2007 at 10:21:40 AM EST

Re: Statistics aren't everything ... (4.00 / 2)

A shorter John Kay: economic analysis is pseudoscience.

"It's the statue, man, The Statue."

by Migeru (migeru at eurotrib dot com) on Tue Jan 30th, 2007 at 10:22:34 AM EST

Re: Statistics aren't everything ... (none / 0)

No, I believe that should be "economic analysis that doesn't support my ideological position is pseudoscience".

by Colman (colman at eurotrib.com) on Tue Jan 30th, 2007 at 10:24:37 AM EST
[ Parent ]

Re: Statistics aren't everything ... (none / 1)

That's what he'd like to be saying.

"It's the statue, man, The Statue."

by Migeru (migeru at eurotrib dot com) on Tue Jan 30th, 2007 at 10:31:27 AM EST
[ Parent ]

Re: Statistics aren't everything ... (none / 0)

The US productivity miracle was in part created, not by finding new facts about the US economy, but by reclassifying software expenditure as investment and adopting aggressive assumptions about falling computer prices.

Is he admitting that hedonic pricing, and not economic reality, is responsible for "the US productivity miracle".

"It's the statue, man, The Statue."

by Migeru (migeru at eurotrib dot com) on Tue Jan 30th, 2007 at 10:24:46 AM EST

Re: Statistics aren't everything ... (none / 0)

Yes, yes he is. But you'll find that some of the current EU growth comes from increasing use of such things here.

by Colman (colman at eurotrib.com) on Tue Jan 30th, 2007 at 10:25:43 AM EST
[ Parent ]

Re: Statistics aren't everything ... (none / 0)

Peudoscience, I'm telling you. I'll go with "the evidence of my eyes and ears" instead.

"It's the statue, man, The Statue."

by Migeru (migeru at eurotrib dot com) on Tue Jan 30th, 2007 at 10:30:54 AM EST
[ Parent ]

Re: Statistics aren't everything ... (none / 0)

But you'll find that some of the current EU growth comes from increasing use of such things here.

Are you saying that he is implying this, or that this is a fact?

In the long run, we're all dead. John Maynard Keynes

by Jerome a Paris (jeromeguillet@yahoo.fr) on Tue Jan 30th, 2007 at 10:54:53 AM EST
[ Parent ]

Re: Statistics aren't everything ... (none / 0)

I think you'll find that the EU stats have started slipping in some of this, but I'm working from memory - I'm not sure of the precise details.

by Colman (colman at eurotrib.com) on Tue Jan 30th, 2007 at 11:06:28 AM EST
[ Parent ]

Linky goodness (none / 0)

Link to his article now updated.

In the long run, we're all dead. John Maynard Keynes

by Jerome a Paris (jeromeguillet@yahoo.fr) on Tue Jan 30th, 2007 at 10:56:31 AM EST

Re: Statistics aren't everything ... (none / 0)

The key number used to measure economic performance is gross domestic product. But few politicians or traders could actually define it. GDP is not, exactly, a measure of either business output or consumer welfare, although it is loosely related to both. It is safest to say that GDP is the number you arrive at if you follow an internationally agreed set of statistical conventions.

If this is the best the FT can publish about GDP, it would seem ET does a better job of being a popular economic publication than the FT.

"It's the statue, man, The Statue."

by Migeru (migeru at eurotrib dot com) on Tue Jan 30th, 2007 at 11:03:15 AM EST

Actually, its safest to say that GDP ... (none / 0)

... is the market value of newly produced goods and services, and the confusion is caused by trying to stretch it to mean all sorts of things it does not, like economic welfare or national economic strength.

We could also add that unemployment comparisons are what you get when you ignore international standards for accounting for economic activity ... but noticing that would not serve to apologize for a not quite elected regime that has allowed corporations to write it laws regulating corporations.

Utsukushikereba sore de ii

by BruceMcF (agila61 at netscape dot net) on Tue Jan 30th, 2007 at 11:12:49 AM EST

Re: Actually, its safest to say that GDP ... (none / 0)

The problem with such a simple explanation of GDP is that it doesn't help one decide what inflation (em, the "GDP deflator") should be when both the quantities produced and their quality change from year to year.

"It's the statue, man, The Statue."

by Migeru (migeru at eurotrib dot com) on Tue Jan 30th, 2007 at 11:15:22 AM EST
[ Parent ]

Re: Actually, its safest to say that GDP ... (none / 0)

So?

The problem is only a practical problem because of Central Banks that over-react to moderate inflation while under-reacting to real unemployed labour of 10% and more.

There is intrinsically no such thing as "the" correct inflation rate, because what matters is the price changes in the goods and services that matter, and what goods and services matter is not something to be decided by some pretence at an "objective" formula.

The arguments over "the" correct inflation rate are therefore as endless a source for the amusement of the scholastic school of Economics as any other argument over how many angels can dance on the head of a pin.

Utsukushikereba sore de ii

by BruceMcF (agila61 at netscape dot net) on Tue Jan 30th, 2007 at 11:43:45 AM EST
[ Parent ]

Re: Actually, its safest to say that GDP ... (none / 0)

So the difference between real and nominal GDP growth is scholastic?

"It's the statue, man, The Statue."

by Migeru (migeru at eurotrib dot com) on Tue Jan 30th, 2007 at 11:50:59 AM EST
[ Parent ]

Re: Actually, its safest to say that GDP ... (4.00 / 1)

The difference between increases in real and nominal GDP growth is that increases in nominal GDP are actual summaries of actual transactions, and increases in "real" GDP are changes in GDP after being modified by whichever inflation correction has been chosen.

The quest for price indices that are appropriate to different questions at hand is certainly not a sholastic endeavour, provided that we bear in mind that any "real" GDP is a rough proxy for a change in a vector that cannot be unambiguously expresses as a change in a single value.

It is the quest for "the correct" price index that is a scholastic endeavour. The appropriate price index depends upon the question we are asking. Indeed, if we are worried about a wage-price spiral, that would be the appropriate price indices, since cost of production and cost of living each have their part to play in that process.

Utsukushikereba sore de ii

by BruceMcF (agila61 at netscape dot net) on Tue Jan 30th, 2007 at 03:43:53 PM EST
[ Parent ]

Re: Actually, its safest to say that GDP ... (none / 0)

There is intrinsically no such thing as "the" correct inflation rate, because what matters is the price changes in the goods and services that matter, and what goods and services matter is not something to be decided by some pretence at an "objective" formula.

Well, actually, the sensible thing to do is to weigh price changes by each product's share of GDP.

If GDP is the sum of quantity and price for each product, GDP = sum pq, then

d(GDP) = sum (p dq) + sum (q dp)

where the first term is real GDP growth and the second term is GDP inflation

So the GDP deflator is sum(pq dp/p), that is, the rate of change of each product's price, weighted by that product's share of GDP.

"It's the statue, man, The Statue."

by Migeru (migeru at eurotrib dot com) on Tue Jan 30th, 2007 at 12:01:10 PM EST
[ Parent ]

Re: Actually, its safest to say that GDP ... (none / 0)

The sensible thing to do depends on what it is that we are trying to accomplish.

For example, like many differential approximations being applied to values defined across periods of time, the chain index GDP deflator will give slightly different values between quarterly and annual values, and substantially different values between annual and decennial values.

And to assess the impact of price inflation on the real wage, the inflation from the perspective of the wage recipient is in terms of the goods and services in the household budget, which changes the weights, and the inflation from the perspective of the producer is in terms of ongoing financial obligations and the goods and services in the production budget.

Utsukushikereba sore de ii

by BruceMcF (agila61 at netscape dot net) on Tue Jan 30th, 2007 at 03:50:54 PM EST
[ Parent ]

Re: Actually, its safest to say that GDP ... (none / 0)

For example, like many differential approximations being applied to values defined across periods of time, the chain index GDP deflator will give slightly different values between quarterly and annual values, and substantially different values between annual and decennial values.

Actually, if what I wrote is correct the discrepancies in the "chain index GDP deflator" are fundamental and reflect the fact that the GDP deflator is not an exact/integrable differential. Only the sum of the GDP deflator and the GDP growth rate is an exact differential. I am not aware of many places outside thermodynamics where that distinction is operative.

"It's the statue, man, The Statue."

by Migeru (migeru at eurotrib dot com) on Tue Jan 30th, 2007 at 05:07:36 PM EST
[ Parent ]

Re: Actually, its safest to say that GDP ... (none / 0)

The fundamental problem is the national accounts version of the Heisenberg principle ... the shorter the time period, the closer a series of discrete period differences approxmate a differential, but the measurement of the Q vector is only at its most precise in the year of an industrial census, which is somewhere in the range of every 5 to 10 years, for those of you lucky enough to be focusing on high income nations. Among low income countries, its not uncommon for the Q vector to never defined with any great precision.

And when we are talking about a five year span, then working with the differential approximation without the squared term is only a rough approximation, accurate to the first decimal (10%'s), but not the second (1%'s).

Utsukushikereba sore de ii

by BruceMcF (agila61 at netscape dot net) on Tue Jan 30th, 2007 at 05:22:25 PM EST
[ Parent ]

Re: Actually, its safest to say that GDP ... (none / 0)

I don't understand how discrete time steps are a problem. It's not like you can't define the rate of change of x over a period as log(1 + dx/x)/dt, which is accurate and additive no matter what the time period is.

"It's the statue, man, The Statue."

by Migeru (migeru at eurotrib dot com) on Tue Jan 30th, 2007 at 06:01:56 PM EST
[ Parent ]

Re: Actually, its safest to say that GDP ... (none / 1)

The longer the periods, the bigger the dP's and dQ's get in percentage terms, and the less accurate it is to discard the dQdP term in the differential. The dP and dQ are both on the order of 10%, dPdQ is on the order of 1%, and if they are both on the order of 20%, dPdQ is on the order of 4%.

That is on top of all the intrinsic chain index problems in a system with innovations.

Utsukushikereba sore de ii

by BruceMcF (agila61 at netscape dot net) on Tue Jan 30th, 2007 at 08:16:57 PM EST
[ Parent ]

Re: Actually, its safest to say that GDP ... (none / 0)

Duh, thanks.

"It's the statue, man, The Statue."

by Migeru (migeru at eurotrib dot com) on Wed Jan 31st, 2007 at 01:59:43 AM EST
[ Parent ]

Re: Actually, its safest to say that GDP ... (none / 0)

The reference to Heisenberg ties into what is now called Quantum Calculus. It's actually kind-of fun, and it has lots of interesting connections to combinatorics.

"It's the statue, man, The Statue."

by Migeru (migeru at eurotrib dot com) on Wed Jan 31st, 2007 at 03:51:32 AM EST
[ Parent ]

Shhhh, be vewy vewy qwiet ... (4.00 / 1)

... if the math-before-reality economists hear you, they'll charge off to apply Quantum Calculus with even less thought about its relevance to the economy than they did with human behavior and the classical calculus.

Utsukushikereba sore de ii

by BruceMcF (agila61 at netscape dot net) on Wed Jan 31st, 2007 at 11:00:35 AM EST
[ Parent ]

Re: Shhhh, be vewy vewy qwiet ... (none / 0)

I don't have a problem with that, as "classical calculus" has epistemological problems stemming from continuity and uncountability that the "quantum calculus" (nothing to do with quantum mechanics, by the way, it's more like a return to 17-th century mathematics) doesn't have.

"It's the statue, man, The Statue."

by Migeru (migeru at eurotrib dot com) on Wed Jan 31st, 2007 at 11:05:48 AM EST
[ Parent ]

Re: Shhhh, be vewy vewy qwiet ... (none / 0)

The epistomological problems from the mathematical utility theory are deeper than technical flaws in the mathematical technique.

Utsukushikereba sore de ii

by BruceMcF (agila61 at netscape dot net) on Wed Jan 31st, 2007 at 03:27:14 PM EST
[ Parent ]

Re: Shhhh, be vewy vewy qwiet ... (none / 1)

Oh, well, I have been bashing utility on this blog at the slightest opportunity.

"It's the statue, man, The Statue."

by Migeru (migeru at eurotrib dot com) on Wed Jan 31st, 2007 at 03:53:18 PM EST
[ Parent ]

Re: Shhhh, be vewy vewy qwiet ... (none / 1)

Oh, and Mathematics is never before reality, by the way.

"It's the statue, man, The Statue."

by Migeru (migeru at eurotrib dot com) on Wed Jan 31st, 2007 at 11:08:55 AM EST
[ Parent ]

Re: Shhhh, be vewy vewy qwiet ... (none / 1)

Hmph.

by Colman (colman at eurotrib.com) on Wed Jan 31st, 2007 at 11:46:45 AM EST
[ Parent ]

Re: Shhhh, be vewy vewy qwiet ... (none / 1)

I hadn't run across WMCF. Thanks for the link.

A couple of comments:

Lakoff and Núñez's avowed purpose is to begin laying the foundations for a truly scientific understanding of mathematics ...

Oh, yeah. Most of the (interesting) work in 20th Century Logic was devoted to that particular rat-hole.

WMCF emphatically reject the Platonistic philosophy of mathematics.

Hurray! The belief in Platonic Archetypes in Mathematics is widely shared. On can almost state it is the default position amongst mathematicians. I have always thought it was a silly position and prefer to ally with Laplace, "I do not need that hypothesis."

Whether a transcendent mathematics, independent of human thought, can be said to exist is an unanswerable and perhaps meaningless question.

"meaningless questions" was a major pre-occupation of the Logical Positivists: Carnap, Russell, Ayer, et.al. No question skepticism regarding the questions, investigatory fundamentals, one bases ones work on is healthy. And valuable. But one has to know when to stop and get on with it.

Och nu den svenska kocken bakar en Alaskan älg jägare. Bonk! Bonk! Bonk!

by ATinNM on Wed Jan 31st, 2007 at 12:12:10 PM EST
[ Parent ]

Re: Actually, its safest to say that GDP ... (none / 0)

Your equation is wrong.

GDP'= sum ((p + q dp/dq)q') + sum ((q + pdq/dp)p')

Another reason why inflation numbers can differ is because producers and consumers also have a different opinion on what the same product is (This is not totally the same as substitution)

by charlyv on Wed Jan 31st, 2007 at 10:56:29 PM EST
[ Parent ]

Not even (none / 0)

A lot of economic activity (like government spending) is measured purely at cost.

Another big chunk of economic activity is ressource extraction, which is value in so far only as the existing stock is not accounted for.

There's been a push to bring governments to state balance sheets and proper accounts, instead of pure cash-in, cash-out budgets, but the same should probably be done for GDP. Bring in depreciation, amortization and asset write offs (car accidents, here we come), and we might be surprised...

In the long run, we're all dead. John Maynard Keynes

by Jerome a Paris (jeromeguillet@yahoo.fr) on Tue Jan 30th, 2007 at 11:18:50 AM EST
[ Parent ]

Re: Not even (none / 1)

Then its not GDP anymore, its an improved NDP.

It certainly is arguable that we should switch to an improved NDP as we enter an era where net yields are more important than gross yields.

But even then, NDP will only measure what NDP measures ... it will not be a measure of economic welfare, which is intrinsically multi-dimensional and composed of incommensurate elements.

Utsukushikereba sore de ii

by BruceMcF (agila61 at netscape dot net) on Tue Jan 30th, 2007 at 11:47:15 AM EST
[ Parent ]

Re: Not even (none / 1)

All depreciation will be blown away by appreciating house
prices these days :)

http://www.vie-publique.fr/actualite/alaune/patrimoine-forte-progression-entre-1995-2003.html

Selon une étude de l'Insee parue le 6 janvier 2005, le patrimoine national des ménages, entreprises et administrations a augmenté de 73% avec une forte progression du patrimoine des ménages de 82% entre 1995 et 2003.
L'étude révèle que le patrimoine national, défini comme celui des résidents sur le territoire, s'élevait à 8 244 Mds € fin 2003 et que les ménages en détenaient 77%. La forte progression du patrimoine des ménages est due, selon l'Insee, à la hausse des prix de l'immobilier et coïncide avec l'augmentation du nombre de ménages payant l'impôt sur la fortune. L'étude souligne l'importance des plus-values boursières et des prises de participations dans la part des actifs financiers des entreprises, notamment fin 2000. Le patrimoine des administrations, soit 3,7% du patrimoine national fin 2003, se répartit en actifs non financiers (bâtiments, ouvrages publics), détenus essentiellement par les administrations publiques locales, et en actifs financiers, détenus majoritairement par les administrations centrales.
La part des actifs immobiliers dans le patrimoine des ménages français atteint, selon la Banque de France, près de 58% à la fin 2003. En comparaison, la part de l'immobilier dans le patrimoine des ménages néerlandais et britanniques atteint, en 2003, des niveaux de l'ordre de 45% et 55% respectivement.

Of course the difference between house price increase and improvement spending minus real average house aging depreciation might return to its historical average one of these days :).

by Laurent GUERBY on Tue Jan 30th, 2007 at 03:40:46 PM EST
[ Parent ]

Quick LTE (4.00 / 2)

It was highly revealing that you would choose the time when a statistic favors Europe over the USA to start telling us about how statistics are unreliable and should be taken with a grain of salt ("Why data, soft or hard, cannot replace eyes and ears", January 29).
I do not remember such precautionary words being published as supposedly superior US statistics were used relentlessly in your columns to push for "reforms" in Europe. It unfortunately creates the impression that statistics that favor the neo-liberal agenda are somehow of higher quality than those that do not, and creates a very real doubt as to whether such reforms are really what is needed. Was that your intention?

In the long run, we're all dead. John Maynard Keynes

by Jerome a Paris (jeromeguillet@yahoo.fr) on Tue Jan 30th, 2007 at 11:35:52 AM EST

Re: Quick LTE (none / 0)

How about "Setting aside the casual admission that 'the US productivity miracle' has been the result of changes in statistical practice, " at the beginning of the second paragraph.

"It's the statue, man, The Statue."

by Migeru (migeru at eurotrib dot com) on Tue Jan 30th, 2007 at 11:43:39 AM EST
[ Parent ]

Yes (none / 0)

Good idea. I'll be offline for the next two hours, and will incorporate whatever comments come in the meantime.

In the long run, we're all dead. John Maynard Keynes

by Jerome a Paris (jeromeguillet@yahoo.fr) on Tue Jan 30th, 2007 at 11:47:35 AM EST
[ Parent ]

Re: Quick LTE (4.00 / 1)

I do not remember such precautionary words being published as supposedly superior US statistics were used relentlessly in your columns to push for "reforms" in Europe. It unfortunately creates the impression that statistics that favor the neo-liberal agenda are somehow considered by the FT to be of higher quality than those that do not, and creates a very real doubt as to whether such reforms, advocated by the FT, are really what is needed : biased protagonists have lower credibility. Was that your intention?

Just trying to be a bit more sarcastic...

by Sargon on Tue Jan 30th, 2007 at 11:51:43 AM EST
[ Parent ]

Re: Statistics aren't everything ... (none / 0)

Some ads for this diary:

http://economistsview.typepad.com/economistsview/2007/01/coordinated_cap.html

http://www.prospect.org/deanbaker/2007/01/alan_sloan_uncovers_bush_steal.html

by Laurent GUERBY on Tue Jan 30th, 2007 at 03:30:50 PM EST

Statistics aren't everything ... | 41 comments (41 topical, 0 editorial, 0 hidden)

Go to: [ European Tribune Homepage : Top of page : Top of comments ]