Dust in the Wind

"All we do
Crumbles to the ground, though we refuse to see".

 (Kansas, "Dust in the wind")

There once was a town called Gary, Indiana.

Founded by the US Steel in 1906, its population reached a peak of 178,320 (*), in 1960 and its health depended on US Steel's profits.

With greater opening to international trade, American manufacturers increasingly lost profitability and layoffs became generalized.

The unionized labour force, mostly white, migrated to areas where better employment opportunities still existed, just to find that these opportunities often disappeared.

By 2000, Gary's population had been reduced to 102,746 (*) [with 95,920 (*) estimated for 2008], predominantly black and impoverished [83.2% of the total (**)].

Paul Mason, from BBC, reports here what became of Gary, Indiana. More  images of Gary can be found at the "Gary, Indiana, Ghost Town" web page.

From an Australian perspective, Gary's fate, however tragic, may look as something remote, utterly alien, unrelated to us.

Still, this may be the shape of things to come, not only for us in Australia, but for humankind.

(*) Gary, Indiana. Wikipedia
(**) Gary City, Indiana - Fact Sheet. US Census Bureau, 2006-2008 American Community Survey.

Lies, Damn Lies and Statistics (II)

Mark Twain, it appears, is an unlimited source of quotations, even if sometimes the quotations can only be tentatively attributed to him.

Regardless, the following is a pearl of wisdom:

"If you don't read the newspaper, you're uninformed;
if you do read the newspaper, you're misinformed."

Enter Mr. Michael Pascoe, from Fairfax, resolute to prove Twain's dictum.

Commenting on the ABS report Measures of Australia's Progress, 2010 (ABS catalogue 1370.0), Mr. Pascoe declared, among other things:

"Per capita real net national disposable income grew by 2.6 per cent a year over the decade - that's the money in our collective wallets and purses;
"Our national per capita wealth (the national balance sheet, what Australia is worth divided by the number of people) increased by about 0.9 per cent a year.
"And, the real clincher against the whinging class, real average household income of low-income Australians grew by 41 per cent. The poor have become substantially better off, if the word 'richer' doesn't quite fit. Middle-income folk saw their after-inflation household income rise by a whopping 46 per cent."

Allow me to focus on these three items.

But first a note of caution: one must understand that journalists, even business journalists with long years of experience, are constrained by time: they have to abide by deadlines and are likely to have little time or inclination to fact-check the information they come across. In other words, it seems, is unreasonable to expect them to read "Explanatory notes" or any other such minutiae.

So, one may forgive Mr. Pascoe for not checking the notes referring to the third item above (the "real clincher" of 41% enrichment enjoyed by low income households).

Still, that's a shame. Had he opened just one link, he'd have seen the following note, and perhaps saved himself this rejoinder:

"The headline indicator shows a rise in the real income of low income households between 1997-98 and 2007-08, with their average real equivalised disposable household weekly income increasing by 41% over this period. However, part of this increase reflects improvements to the way income was measured from 2003-04 onwards."

Have a look at the chart below, reproduced here for your viewing pleasure.

The jump in the data, in the 2003-2004 period, reflects the "improvements" mentioned in the paragraph above. You'll also notice that the "slope" of the curve changed in that period: I suppose now Mr. Pascoe would say that the whingeing class are getting richer, faster...

That chart was in the very first page of the ABS report on Household economic wellbeing. Is that chart (together with the accompanying text) that pushed Mr. Pascoe into the display of neoliberal triumphalism, reflected in his piece.

I'm planning to write, in the not-so-distant future, about these "improvements". For now, let's just say things that were there but weren't considered income then, still are there, but are now considered as income.

Mr. Pascoe's failure to notice the caveat put there by ABS can be blamed on a lack of time, and excessive rush. That's fine: he would have had to click a couple of links to find the warnings about that data.

However, his failure to notice that wealth and income per capita (or average) measures mean diddly squat (revealed in his comments about "our collective wallets and purses" and about our "national balance sheet" increasing 0.9% per year) is simply appalling.

Here he didn't even need to click one link and read. He only had to ask his Fairfax colleague, Mr. Ross Gittins:

"Next, more attention needs to be paid to changes in the distribution of income, consumption and wealth. That's because changes in the averages of each of those things can be misleading. If, for example, much of a rise in income goes to the highest income group, it's possible that the real incomes of people on lower incomes could actually fall without this showing up in the average."

If he had checked more thoroughly the material made available by the ABS, he might even have seen the following table:

And, so as not to tire busy readers, the good folks at the ABS even provided the interpretation:

"Percentile ratios are one measure of the spread of incomes across the population. For example, the P90/P10 ratio is the ratio of income at the 90th percentile (i.e. the income level dividing the bottom 90% of the population from the top 10%) to that at the 10th percentile. In 2007-08, this ratio was 4.30, meaning that the income of households at the 90th percentile was over four times as great as the income of households at the 10th percentile. This represented an increase of 14% from 1997-98 (3.77)."

Perhaps, had Mr. Pascoe read this information, he'd have realized it's exactly what Mr. Gittins was warning against in the paragraph above!

In a future opportunity, he might want to try the website Statistics Every Writer Should Know, or even my own humble writings.

In any case, I feel perfectly justified to whinge, Mr. Pascoe, about neoliberal policies and about journalists who don't do their homework.

Observation vs deduction

What are the roles of observation and deduction in science? The following piece aims to shed some light on this issue.


A little history (I)

After thousands of years of observation, the list of known planets seemed to end with Saturn. How did this list come to be expanded?

Although dim, Uranus is visible to the naked eye and, in fact, it was observed before being identified as a planet. In 1690 John Flamsteed observed it several times. Between 1750 and 1769 Pierre Lemonnier observed it twelve times.

These earlier observers may have misidentified it as a star.

Initially William Herschel observed it on 13/03/1781 and reported it as a comet. The Astronomer Royal, whom Herschel had notified, identified it as a planet.

By 1783 Herschel himself acknowledged that Uranus was a planet.

One could say that Herschel's achievement was more related to the correct identification of Uranus' true nature and that it largely was a serendipitous event.



A little history (II)

Unlike Uranus, Neptune is essentially invisible to the naked eye.

It may have been observed through telescope by Galileo (1613), Jerome Lalande (1795) and John Herschel (1830). However, like Uranus before, it was probably mistaken for a star, and, in any case, its true nature was never communicated by its would-be discoverers.

But the histories of Neptune and Uranus are related in a more fundamental way.

After Uranus' discovery, studies were conducted to predict its orbit. But these studies suffered from a curious problem: after a while, they would become increasingly imprecise. The observations, in other words, falsified the predictions.

Discrepancies between Uranus' actual position and Alexis Bouvard's 1821 predictions could be explained in at least three ways:

(1) Perhaps gravity, at such great distances, operated in a manner slightly different to what Newton described;
(2) Maybe Bouvard's predictions were based on observations containing systematic measurement errors;
(3) A yet undiscovered planet could be perturbing Uranus' orbit.

Two researchers, John Couch Adams and Urbain Le Verrier, independently decided to pursue the third explanation.

By early 1846 both researchers had produced at least some calculations, based on Newton's gravitation; and, being finally aware of each other's work, commissioned the Greenwich and Berlin observatories, respectively, to conduct the astronomical search.

The exact content of the calculations made by Adams and Le Verrier, as well as the details of the search conducted at Greenwich and Berlin, are subject to controversy.

The fact is that on 24/09/1846, after less than an hour search, the Berlin Observatory reported the planet had been found, very near where Le Verrier had predicted.

Adams recognized Le Verrier's priority. Apparently, the most recent opinion prevailing among historians of astronomy favors Le Verrier.


Discussion

Unlike the discovery of Uranus, where serendipity played such a clear role, the discovery of Neptune was hailed as a major triumph for Newtonian physics.

By correct deduction on the basis of current knowledge, Le Verrier and Adams had produced new knowledge. The apparent inexactitude in the orbit of Uranus had been explained: it had very little to do with systematic measurement errors and nothing to do with an erroneous appreciation of gravity by Newton.

However, I would like to point to an asymmetry between knowledge acquired through observation and knowledge acquired through deduction, at one hand; at the other hand, I would like to call the attention to a failure of some views on the philosophy of science.

The existence and true nature of Uranus was established by observation. That of Neptune was also established by observation.

Whoever rightfully deserves be credited with the correct prediction of the orbit, mass and angular speed of Neptune, was only formulating a hypothesis. That's why they required the collaboration of the observatories of Greenwich and Berlin.

In this sense, there is an asymmetry between knowledge acquired through observation and that acquired through deduction: deduction is neither sufficient nor necessary to determine the reality of a phenomenon.

It's not necessary, because discovery of a phenomenon can happen at least through serendipitous events, as in Uranus' case.

It's not sufficient, because predictions and hypothesis could be wrong, even if they are validly derived from acknowledged principles.

This last point was illustrated by the searches by Greenwich and Berlin. An even better illustration, however, comes also from the history of astronomy: the Ptolemaic system. A treatment of this episode falls outside the scope of this piece.

The second point I would like to make is that an extreme understanding of Popper's falsificationism would probably have discarded Newton's gravity, on the grounds of the discrepancy between the predicted orbit of Uranus and its actual orbit.

Contrary to Popper's view, Newtonian gravity gained credibility, and rightly so, when its apparent failure was explained through a discovery of a new planetary body.

NOTE: Images and information from Wikipedia.

Despairing… progressively (II)

“Do not imagine, gentlemen, that in criticizing freedom of trade we have the least intention of defending the system of protection.

“One may declare oneself an enemy of the constitutional regime without declaring oneself a friend of the ancient regime.”

Karl Marx, “On the question of free trade”, speech to the Democratic Association of Brussels at its public meeting of 09/01/1848.

According to Linden, Dedrick and Kraemer (LDK henceforth), authors of the report “Innovation and Job Creation in a Global Economy: The Case of Apple’s iPod” higher profits and higher wages are the likely result of two factors acting simultaneously, namely innovation and free trade:

“To summarize, the iPod supports nearly twice as many jobs offshore as in the US, yet wages paid in the US are over twice as much as those paid overseas. (…) So it appears that innovation by a US company can benefit both the company and US workers, even if production is offshore and foreign suppliers provide most of the inputs.”

NOTE: The reader is advised to download that report.

That conclusion follows from these two tables, from the Executive Summary:

For instance, in LDK Table ES2, total compensation for US staff (US$ 783,803,828 million) more than doubles compensation for non-US staff (US$ 318,486,050 million).

Where do the data come from?
 
The question arises: where do that data, expressed with such precision, come from? After all, the authors themselves express in the methodological appendix: “The firms directly involved will not provide data, and we have found no systematic third-party source of data on employment by firm or by industry”.

This limitation is even more pressing for LDK who are dealing with data at an even greater level of disaggregation: a single product line.

Furthermore, as the figure below shows (taken from Wikipedia), Apple produces an extremely diversified line of goods, ranging from personal computers, computer peripherals, expansion cards, software, accessories and consumer electronics, directly and through contractors, all which  requiring a hard to precise workforce.

 

What’s more, LDK list 8 countries where iPod is produced (US, China, Philippines, Japan, Singapore, Korea, Thailand, and Taiwan), not necessarily in a single plant, performing different functions and being paid their own local wages.

Considering that Apple Inc. also needs to perform a series of corporate functions (R&D, marketing, software, and management) besides production, the problem of quantifying workforce and workforce compensation is formidable.

LDK approached the problem in the only manner that seems possible, given the circumstances: the data were generated by estimation, based on the number of iPod produced/revenue generated during 2006, regardless of their origin, on interviews and international production worker wages data compiled by the US Bureau of Labor Statistics.

Estimation, is easy to understand, involves a considerable amount of guess work. LDK are upfront about this: they report their methodology with considerable detail in the appendix.

Some estimates

It does not make sense to repeat their text here, and for the details, the interested reader is referred to the LDK report itself.

However, to give a flavor of the problems involved in producing their estimates, a brief outline follows, including comments on some of the assumptions used.

LDK estimated separately the number of jobs generated and wages paid.

As an example of the job generation procedure: 240 million flash memory chips were required for 30 million Nano iPods actually sold during 2006. A medium-sized factory, employing 1,200 staff, could produce that many flash memory chips (see the flash memory item, in LDK Table 1). Presumably in the US, around half a chip factory’s workforce is higher-paid staff, in management and engineering; the remaining are lower-paid production workers.

The same proportion is assumed for factories overseas: see the Korea item in LDK Table 2. This assumption seems somewhat questionable, it must be remarked, as R&D is performed entirely in the US. To put a concrete example, why would Foxconn, an iPod assembler located at Shenzhen require such a high proportion of engineering and managerial staff, if they only assemble components imported?

Regardless, the half/half distribution observed in US factories occurs under American workplace regulation for working conditions such as hours of work. Overseas, more “flexible”, labour legislation could result in different distributions.

The method described above (called “Factory Fraction Method”) is complemented by a second job generation method (“Revenue Fraction Method”), to estimate non-factory staff: in 2006, Amazon (one among other iPod retailers) sold US$ 1 billion worth of iPod, out of US$ 10,7 billion total revenue, employing 13,900 staff. Thus, LDK suggested, 1,300 Amazon equivalent-jobs were due to iPod sales.

This seems dubious and probably greatly overestimates the number of jobs created by iPod sales: to sell an iPod priced US$ 299, arguably, uses the same manpower as a much cheaper item, like a book costing US$ 10, but generates 30 times more revenue.

From LDK Table 2, one can see that up to 31% of the iPod generated jobs are classified as “retail and other non-professionals” and, thus, are quite sensitive to this assumption. Further: over 38% of those jobs are based overseas and do not benefit the American workers potentially displaced by outsourcing.

A number of assumptions are also made when estimating wages. For brevity, only one remark will be made here: “the 2006 hourly rates [used to estimate annual production worker income overseas] were annualized by assuming 2,000 paid hours per year” could be missing the mark by an order of magnitude.

At least in the case of Foxconn, The New York Times reported on 18/06/2010 that production line workers earn US$ 0.75 an hour, working up to 13-hour-long shifts, six days a week, for US$ 235 a month or US$ 2,820 a year. The yearly wage estimated by LDK is US$ 1,540.

In fact, the employee category where wage and employment numbers appear more reliable is engineering and other professionals.

Closing remarks

Researchers performing studies of this nature will inevitably make all sorts of assumptions. This is natural, unavoidable and understandable, although a great deal of prudence is required.

It is also natural, unavoidable and understandable that some of their assumptions will be questioned, as readers with different insights go through their work.

In this sense, Linden, Dedrick and Kraemer are to be praised for their effort in making their assumptions explicit.

However, if prudence is required when making assumptions, at least every bit as much is required when drawing conclusions from studies of this kind. Sadly, the conclusions drawn by Linden, Dedrick and Kraemer appear to make little allowance for these assumptions.

An example from LDK Conclusions:

“The relationship between innovation by US companies and employment in the US is more complex than phrases such as the ‘vanishing middle class’ suggest. When innovative products are designed and marketed by U.S. companies, they can create valuable jobs for American workers even if the products are manufactured offshore.”

From LDK Table ES1 is evident that little if any replacement jobs were created by iPod for production line workers left redundant by outsourcing: their very category all but disappeared and the next category where they could be readily employed (retail and other non professional), not only employs a minuscule number, at a much lower wage, but that number probably overestimates the actual number of jobs created.

It is certainly possible for younger ex production line workers to obtain qualifications that would allow them to climb up the corporate ladder. However, this requires an investment of time, effort and money, imposed upon them without their consent, which, even if they were able to undertake, would not guarantee them a better job.

But what happens to older workers, or those with personal attachments, who will not be able to start further training and qualification? And what about those workers who due to the colour of their skins, or their religion, or their place of birth cannot realistically aspire to anything better than that job and find now that it’s gone?

In sum: to the extent that those workers were once middle class, LDK own figures (however estimated) refute their conclusion.

Because of this, readers of that report should exercise caution and common sense, virtues which often seem beyond neoclassical economists or the general media.

Thus, it would not surprise that this particular Linden, Dedrick and Kraemer report (as opposed to other research carried out by these researchers) were uncritically endorsed by the openly pro-business media: after all, they cater for their readers and the corporations that advertise through them.

Against expectations, however, mainstream and business media made very little reference to this report, the only exceptions I could find were The New York Times (and its international edition, The International Herald Tribune), BusinessWeek and a few other minor publications.

In what follows, the reader will forgive me for using a more candid, personal, non-academic language; a language more suited to me, anyway, as I am not an academic.

I would have also expected a keen support, for ideological reasons, from well-known neoclassical economists. Surprisingly, only Hal Varian (who authored the The New York Times article) has mentioned this particular report, perhaps because it was not published in any academic journal, either.

A query on Google Scholar with the string “Innovation and Job Creation in a Global Economy” returned two results: a working paper housed at SSM and a French language paper, both of which seem critical to LDK conclusion.

Well known mainstream economists like Paul Krugman, who once proclaimed that globalization had no negative effect on lower income Americans, have recanted since and now not only admit they were wrong, but also explain why.

What’s really surprising and disappointing is that a self described progressive like Prof. Mitchell has embraced this myth, not taking even the time to evaluate the report. And let’s be crystal clear on this: it doesn’t take a rocket scientist to find those warts.

It’s this readiness to accept reforms affecting other people’s lives, without question, that troubles me.

A few days ago, I saw a quote attributed to Keynes that seems appropriate here. Talking about the reasons of the triumph of economic orthodoxy, Keynes is reported to have said:

“That it [economic orthodoxy] could explain much social injustice and apparent cruelty as an inevitable incident in the scheme of progress, [with] the attempt to change such things as likely on the whole to do more harm than good, commended it to authority. That it afforded a measure of justification to the free activities of the individual capitalist,  attracted to it the support of the dominant social force behind authority.”

As a working class Australian, the reader will forgive me if I’m less than enthusiastic about Apple shareholders making big bucks on the misery of American or Chinese workers, because I can see myself, and the likes of me, following in their steps.

Neither am I excited about Apple’s management team doing well: for all I care, they could all go to hell. And they might yet go there, if you believe anything the report says.

But something positive came out of this: as a working class Australian, I’ve learned a lesson. We, the working class, the only we that includes me, are really alone. The Australian progressive intelligentsia is a myth, not too different from the Yowie and the Tassie tiger.

PS: The text appears to contain some typos. An example is the US Total compensation in LDK Table ES2, where the value shown does not correspond to the sum of the staff categories (see yellow cell in the Table presented at the beginning of this post and compare it to their version). Likewise, the figures cited in the text (under LDK Table 4) do not correspond to those shown in LDK Table ES2.