Friday, 29 June 2018

The Preconditions of Socialism: a Critical Review (ix)

It’s time to go back to Thomas Piketty’s data and their application to the assessment of Preconditions.

A first question about Piketty’s 1910s Europe “capital” data would be how representative of a particular, historical, observed national economy that data are?

At the World Inequality Database the only series (personal wealth distribution: analogue of “capital”) with data for the lowest 50%, the middle 40%, the upper 9% and the top 1% I could find for that general time period was the one for France, for the years 1902, 1903, 1904, 1905, 1907, 1909, and 1910.

This chart compares the data presented last time for 1910s Europe and the WID data for France 1902, using a graphical device called the Lorenz curve, which I explained a few months back:

Without going into the meaning of that chart (readers are invited to re-read that old post for that), it’s easy to see the fit appears to be quite good although according to the France 1902 data the situation of the “lower class” seems slightly more dire than in the 1910s Europe data.

Correspondingly, the situation of the French “middle class” appears slightly better than that of its European counterpart.

Compare what either of both charts say about the 90% of the European population with Bernstein's claim absolutely unencumbered by any data:
“I have previously remarked that modem wage-earners are not the homogeneous mass uniformly devoid of property, family, etc., as predicted in The Communist Manifesto, that it is precisely in the most advanced manufacturing industries that a whole hierarchy of differentiated workers is to be found, and that among these there is only a tenuous feeling of solidarity.” (p. 104, E3§b)

Perhaps some observations would help readers get a better sense of that data.

Histograms are a popular device to visualise data. Given a random variable (like “capital”), one should expect that starting with a histogram one could find a Lorenz curve. However, to proceed in the opposite direction (i.e. to find the histogram starting from a Lorenz curve), for reasons that will be clear soon, isn’t possible.

Normally, this doesn’t matter either: we’ve seen elsewhere that to represent a wealth/income distribution in histogram form is difficult, because wealth/income are usually highly positively skewed, meaning that most of the cases are located in the “lower reaches” of the distribution, but it has a really, really, really long right tail (you’ll see below what I mean by that).

Still, for those things in life, it just so happens that the 20th Century Greatest Economists data we studied back then yields a Lorenz curve quite similar to the 1910s Europe Lorenz curve:

That means that the hybrid histogram/bar chart we introduced for the 20th Century Greatest Economists data set should be very similar to the equivalent hybrid histogram/bar chart for 1910s Europe “capital”:

The legends would, evidently, change: it wouldn’t be “likes” we measure along the horizontal axis, but “capital” measured in a unit of currency (say, $). Population would be the vertical axis and the numbers along it would increase too. But the general shape should be quite similar: a big mountain near the origin, and a long, long plain with a few bumps as one walks “eastwards”. (The problem, of course, is that we don’t know the labels that go along the horizontal axis or the numbers that go along the vertical.)

This insight, however partial, is worth the trouble: as before, one should keep in mind that the right-most bar represents the top 10% of the “capital” distribution and, although in the chart it looks exactly like the bins to its left, in reality it is 3,219 times (yes, three thousand two hundred nineteen times) wider than them. If one were to search for variability, it’s there, not in the lower reaches, where one should look first. Like I said before: welcome to inequality.

One can get an additional sense of the “capital” distribution from a set of statistics presented in The Case for the Labour Party, which I mentioned last time. That data (p. 41) illustrate the specifically British situation (for the financial year ending on March 1908 -- barely nine years after Preconditions was published -- coming apparently from the Board of Trade). Out of 699,533 estates that financial year “no less than 632,000 [i.e. 90.3%] were too poor to be taxed”. Those estates were valued less than £100.

To be suitably charitable to Bernstein, as modern petty bourgeois intellectuals demand, I’ll abstain from making any comment beyond reminding readers of Bernstein’s own words: “modem wage-earners are not the homogeneous mass uniformly devoid of property”.

To put Bernstein’s words in terms of the chart above: he’s searching for variability in the lower reaches of the distribution (believe it or not).

The remaining 67,533 tax-paying estates (i.e. 9.7%) were divided thus:

Number of   Aggregate value  Average per
persons      of estates (£)     head (£)
      7          15,779,000    2,254,142
     17          16,638,000      978,705
     51          20,086,000      393,843
     90          18,748,000      208,311
    109          16,452,000      150,935
    422          33,740,000       79,952
  3,249          75,790,000       23,327
 17,356          65,737,000        3,766
 46,232          19,688,000          425
 67,533         282,658,000

At the risk of stating the bleedin’ obvious, that table, as it happens, represents the top 10% of the estates distribution: 67,533 estates, for an aggregate value of estates of 283 million pounds.

One doesn’t know what fraction of the total aggregate value that 10% commanded, as the aggregate for the 90% un-taxed estates is not given; therefore, one can’t draw its Lorenz curve. Limitations notwithstanding, that data show several things. Something important here is how enormous those 10% estates are compared to the lower 90%.

But there’s more. The sum of the first 6 rows shows something equally important: the 696 top estates (yes, that’s not a typo, six hundred ninety six) correspond to the 0.1% of the total number of estates. Modern readers will hopefully have little difficulty understanding this, but I suspect Bernstein would have been shocked: the 0.1% of a population is certainly a tiny fraction, but may still conceal a large in absolute terms (and increasing) subset of the population. (More on this in the next post).

Skeptical readers could oppose that one cannot assume those estates are a random unbiased sample of the population. That would be a reasonable objection: death rates at lower wealth levels were higher than death rates at higher wealth levels. The Case shows mortality figures from the Local Government Report on Public Health and Social Conditions (p. 37): for every 515 deaths among clergymen (an occupation for the wealthy), the number of deaths among workers ranged from a minimum of 935 (printers) and a maximum of 2,169 (tin miners).[*]


Although things may have changed since 2014, when Piketty’s Capital was published, wealth inequality had reached its high-water mark in 1910s Europe, at least in what concerns the main capitalist economies of western Europe, the US, and Scandinavia. The following chart illustrates that:

For ease of comparison, the chart represents graphically the Piketty “capital” data from Table 7.2 (“Inequality of capital ownership across time and space”): it shows the five “capital” distributions presented in Capital in order of increasing inequality. Four of the five Piketty series are historical distributions (series 3 to 6 in the chart legend); the fifth series, which I included here for completeness, is Piketty’s normative “ideal” (the solid darker blue curve).

To those series I added two: the absolute equality -- first in the list -- and the absolute inequality curves -- last -- (dashed lines, light blue for absolute equality, grey for absolute inequality). I’ve explained those series elsewhere; suffice it here to say they are not meant to be actual, feasible, or even desirable: they (like Piketty’s “ideal”) are benchmarks.

The closer a historical curve is to one of the benchmark curves, the more or less “capital” is equally or unequally distributed.

Of the curves represented, the one corresponding to 1910s Europe (dark purple) is closest to the absolute wealth inequality curve. One could say that 1910s Europe holds the record for wealth inequality, at least for advanced capitalist economies (meaning western Europe, Scandinavia, the US and maybe more peripheral economies, like Australia, Canada and New Zealand). Further, compared to their 1910s European counterparts, the 2010 US and Europe “lower class” (green and yellow, respectively), one hundred years later, are in the same situation: up to the 50% mark the three curves overlap. In other words, in 100 years many things have undoubtedly changed, the fraction of wealth owned by the “lower class” isn’t one of them.


The chart above, taken directly from Piketty’s Capital, reinforces that conclusion: although previous levels of wealth inequality were already high over the whole 19th century, they seemed to have increased to reach a maximum around the first decade of the 20th century.

There’s plenty more one could say about this (particularly about “middle class” wealth and about why Europe 1910 evolved into Europe 2010), but we’ll leave that for another opportunity.


Like I said last time, the bottom line is that by the late 1890s Marxists would have had good reason to believe that “society as a whole is more and more splitting up into two great hostile camps, into two great classes directly facing each other — Bourgeoisie and Proletariat”.

Against that data Bernstein offers either nothing at all (as in the case above), or produces undated, unsourced statements like this:
“Altogether the number of shareholders in England is estimated at considerably more than a million, and that does not appear extravagant if one considers that in the year 1896 alone die number of joint-stock companies in the United Kingdom ran to over 21,223 with a paid-up capital of 22,290 million marks, which moreover does not include foreign enterprises not negotiated in England itself, government stocks, etc.” (p. 59, E2§b)
Just to have that particular dubious claim contradicted by a Liberal MP barely seven years later:
“The total number of shareholders in the United Kingdom does not, according to Mr. L.G. Chiozza Money, exceed 500,000. The accuracy of this estimate (made in the Morning Leader, October 25th, 1906) has never been seriously disputed” (The Case, p. 54)
UPDATE: I had forgotten something. Preconditions Chapter 3 § b contains a footnote (#14, p. 60) by Henry Tudor. It refers to a claim Bernstein attributes to Giffen and it reads: “The statistic Bernstein quotes does not occur in Giffen's Recent Change in Prices and Incomes Compared (London, 1888) or in his The Growth of Capital (London, 1889).”


Let’s retake the Preconditions as criminal trial metaphor. As my own case against Bernstein and his book is nearing an end, it seems opportune to sum up. So far, this is what Bernstein’s case against Marxism amounts to:
  1. Bernstein intentionally chose the most indefensible understanding of “wissenschaftliche Sozialismus”.
  2. In spite of operating under the rules he himself pulled out of his own ass, a wholesale rejection of Marxism was still unwarranted.
  3. He moves the goal post; he re-draws the line in the sand (points 1 to 3 argued here)
  4. Bernstein uses selective quoting to deliberately misinterpret historical materialism and he does that in the clumsiest, most inept way (argued here).
  5. He is utterly ignorant of all the previous investigations in the field of economic inequality (argued here).
  6. He either offers no data to back up his claims or the “data” he produces look suspiciously like bullshit. But that’s not the worst. The worst is that he has no idea how to deal with data.
I can see how Matt, Noah Smith, and Sidney Hook can recommend that book.

[*] There were, of course, many causes for that higher rate of mortality affecting workers. One may suppose disease, given the Hamburg cholera outbreak mentioned last, quite likely was one. The Case for the Labour Party, however, suggests two other likely additional causes.

UK: Industrial Poisoning Rate 1904-1908
(per thousand)

Year   Cases    Rate
1904      833     66
1905      769     52
1906      822     69
1907      769     58
1908      939     78
Total   4,132
Source: Home Office (p.38)

England and Wales: Suicide Rate 1856-1906
(per 100 thousand)

Year  Number   Rate
1856   1,314   6.90
1866   1,360   6.35
1876   1,713   7.03
1886   2,222   8.07
1891   2,459   8.45
1896   2,639   8.57
1901   3,106   9.52
1906   3,434   9.94
Source: the Official Returns for England and Wales (p. 39)

(When it comes to assess the relative merits of Western Capitalism versus the so-called “Socialist” countries behind the Iron Curtain, apologists for the former seem utterly oblivious to that excess mortality in their own backyards. Funny that, uh?)

No comments:

Post a Comment