Skip to main content

Change in GDP v change in SNALT 1830 to 2000


Alas, in the end I never found comprehensive, annual labour time data, so I settled with something less rigorous but hopefully capable of making the point.

The data I was able to bring together I used to create the chart below showing change in GDP per capita versus change in hours worked per year per employed person, with the latter acting as my way of calculating Marxian SNLT. You can now have a look-see for any strong, lasting corelations between the two.

I chose per capita figures to remove any noise from the GDP growth figures, by which I mean GDP growth from population growth. In a discussion about something that is compiled in such a way as to offer a per capita dataset anyway, as labour time records are, you might as well skip aggregate GDP and go straight to the per capita stuff.

I extrapolated change data from the static data in the sources because the static data are not in comparable/commensurable units, whereas the change data are both percentages, making the comparison more meaningful.

My GDP data are from Angus Maddison. My labour time data are from EH and Our World in Data. All measures of change in these two things are by Yours Truly.

~~~

Change in GDP is in blue - Change in SNALT is in red

No doubt you can see that sometimes there appears to be a positive correlation, and sometimes a negative, but that there is no fixed ratio of change in one factor versus the other. Overall there is, as I would expect, no correlation. Since economic growth due to population growth has been abstracted away by making the GDP data per capita, why is there no neat relationship, whether linear or exponential, between change in GDP and hange in labour time?

A Marxist might answer "because the kind of changes in technology and capital allocation that increase productivity are heterogeneous" but does this response not render the theory of SNLT untestable? If SNLT is a macroeconomic abstraction then it stands or falls under the kind of macroeconomic lens applied in the graph above.

Under my lens, it falls. And that's that. Marx was wrong. Marxians are wrong.

~~~

The intervals are not uniform, but they are shared across both factors, with the latter fact makes the former redundant. All of the data points used are below. I got them onto this blog post quickly using something called Tableizer. Once I've done my preliminary analysis of the data I have produced I will stick the spreadsheet file up on this blog, including appending it to this post.

  YEARGDP CHANGE - %SNALT CHANGE - %
1830
19.37%
1830 - 2000
1840
9.24%
-1.95%
1850
9.41%
-0.74%
1860
21.20%
-4.83%
1870
9.10%
-4.03%
1880
30.22%
-2.47%
1890
6.53%
-2.02%
1900
20.60%
-2.72%
1904
7.80%
-1.58%
1909
13.76%
-1.47%
1914
-4.35%
-4.62%
1919
18.36%
-7.70%
1924
9.74%
0.98%
1929
10.69%
-5.55%
1934
-25.87%
-10.18%
1919
18.36%
-7.70%
1924
9.74%
0.98%
1929
10.69%
-5.55%
1934
-25.87%
-10.18%
1938
19.79%
-17.92%
1940
14.30%
19.16%
1944
75.93%
32.63%
1947
-27.95%
-22.24%
1950
7.60%
-20.40%
1953
11.00%
23.86%
1958
0.17%
-1.88%
1960
5.63%
-2.48%
1963
9.01%
6.63%
1968
21.41%
-2.43%
1970
1.12%
-2.26%
1972
6.08%
0.28%
1973
4.67%
-8.48%
1978
10.09%
7.21%
1980
1.11%
-5.76%
1982
-1.36%
3.87%
1990
26.61%
-9.72%
2000
23.71%
2.07%

EDIT 21st January 2017: I got into a Reddit argument about the Marxian LoV, and it was a response from the other party that motivated me to do the study above.

Popular posts from this blog

So I was reading a piece on The Outline about identity politics when the author, Sean McElwee, brought up a survey he had penned and collated to establish how positions on economic and racial issues align;
Could Democrats win over racially conservative whites with economic populism? It’s unlikely, because people who oppose racial justice also tend to oppose liberal economic policies.  To test this, I created scales of economic and racial liberalism, using two questions that have been on the American National Election Studies surveys since 1972. One question asks respondents to place themselves on a one-to-seven point scale on government aid to black Americans, the other on a one-to-seven scale on guaranteeing jobs and income for all Americans. In 1972, only 54 percent of white Americans who took the racially liberal position (supporting aid to black Americans) also took the economically liberal position (guaranteeing jobs and income).  By 2016, 74 percent did. And in 1972, 77 perce…

Will Automation Make All of the Jobs Disappear?

... No.

There is no reason to suggest that automation will dramatically increase unemployment in the short term, or at all in the long term.

Seriously, it will not.

Do read the links in the order in which they appear please. Finding the right comments in the third link might be quite interesting. They are all by a user called BestTrousers and start with "RI" meaning R1.

The main argument used by HealthcareEconomist3 is to give a survey of several works, while BestTrousers goes for comparative advantage.

Capital & Labor in the Race to Exploit the Other

The idea that labor exploits capital is equally as plausible, sans assumptions*, as the idea that capital exploits labor. This is only intended as a response to the formal concept, descriptive or normative, of exploitation in Marx's schema from Capital Volume I.

* Assumptions include the power relation whereby capital is just assumed to be above labor hierarchically.

~
~ Capital exploits labor because... ... Capital earns income from production done by labor that capital didn't perform
&
~ Labor exploits Capital because... ... Labor earns income from capital that labor didn't buy
~
Basically in good old formal logic fashion both of those cases above, being factual descriptions, are true at once or are false at once.