Muh empirical correlation between labor time and price

Andrew kliman debunked this Labor theory of price long ago. Factor the size of firms into the math and the numbers a much less clear and the correlation is much less strong. Check out chapter eleven of Reclaiming Marx's capital. Labor theory of VALUE not labor theory of PRICE ok. Praise Marx.

Have a bump. I don't know how anybody could walk into a store and think there is something like a labor theory of price.



Kliman has written this hilarious response illustrating how the same logic of Cockshott et. al. can be used to support a "Christianity Theory of Income."

Does anybody here actually understands the steps of Kliman's argument or do you just go with it because you like the conclusion or have heard it recommended by friends? I don't understand it. Would you mind saying it in your own words? How do you measure the size of a firm? I do understand the idea of what spurious correlation is, and of course agree with Kliman's example with the scene of a fire (the size of the firemen group correlates with size of damage in $, the number of firemen doesn't cause the damage to go up, both have the same cause in the size of the fire), but how does it fit for the issue of work-time and prices?

Kliman seems to believe tautological statements are logical fallacies, but they aren't, they just tend to be boring. If your intuition is that a conclusion must be wrong because it sounds boring, you have read too many detective novels.

Given that labour-time is such a common input, to expect on average a decent correlation between the ratios of a couple thousand random pairs of commodities when it comes to input ratios measured in that and their price ratios strikes me as something bleedingly obvious. Kliman is right to say that the question of how you group commodities and industries affects how much of a correlation you will see. Given data with high detail, an economist with an agenda could deliberately group things together so that the differences tend to cancel out or do just the opposite.

It seems to me that Shaikh etc are of course not claiming that a small random sample will show great correlation, but a big one will; and that Kliman first misrepresents them as having the view that labour-value ratios also equal prices in a very non-aggregated way (and then their position would deserve the name labour theory of price), and then he "counters" that this correlation shrinks drastically when you "control for firm size", which is just another way of stating that it works better with aggregates, which is just their original position. And their position would be a bit too banal if working on the input-output tables with some other input than labour yielded very similar correlations with price, but this doesn't happen.

I think the idea is as follows:
Suppose Firm A employs 100 hours of labor to produce a profit of $100, while Firm B employs 10 hours of labor with a profit of $20, and Firm C employs 2000 hours of labor with a profit of $1500. Then there appears to be a positive correlation between labor and value added. However, this is not valid, because it may just be that firms have higher revenue because it's a larger firm - more machines, more workers, more commodities sold, etc. In order to see if it is labor itself that is correlating with value, one would have to look at the value per commodity and track that against labor input for that commodity. In the previous example, if Firm A produced 50 commodities, whereas Firm B produced 10, and Firm C produced 1500, then we have:
Firm A: 2 labor hours per commodity, $2 per commodity.
Firm B: 1 labor hour per commodity, $2 per commodity.
Firm C: $1 per commodity, 4/3 labor hours per commodity.
A much weaker correlation. (In fact, it's a negative correlation.) In both cases, the actual ratios of labor-hours to profit received are the same, but the scale has changed so that the correlation looks much different.

Obviously, I just cooked up these numbers to exaggerate the effect, but the point is that just by expanding scale, one can make anything look like a positive correlation, but by dividing out by the size, Kliman seems to get no real correlation from the data.

I'm not sure what you mean by this. Of course, in the aggregate, total price = total value, but Kliman wouldn't disagree with that.

Well of course it's not a direct correlation, there is much more to price than just labour time. Dead labour, supply and demand, etc. I'm not sure I understand the point here.

The point is that price really correlates not with value but with price of production like Marx said.

He is factoring in dead labor.

first and last are directly adressed by kliman.

Good to know, I have not read the PDF, I was basing myself on the example user did.

Obviously my example was an oversimplification, but dead labor was taken into account (somewhat) by considering profit instead of revenue, but anyway, that's kind of beside the point. The main point is that size messes with the data. Maybe take a look at the pdf in to see how it can lead to all kinds of wonky conclusions.

But isn't that what Shaikh et al are aiming for (getting as close as given the quality of the data they have to work with allows).
Yes, for the total aggregate that's a vacuous statement set by definition, and couldn't be disproven for any other statement of that type (e. g. total price = total number of garden gnomes, money amount per garden gnome changing would just be called inflation or deflation measured in terms of garden-gnome value). This is something everybody in the dispute is aware of, Kliman, Cottrell & Cockshott, Dave Zachariah etc. The point was that the somewhat aggregated data doesn't show the same correlation it does with labour when using other inputs (Cockshott mentioned this being done for electricity showing lower correlation).

I don't have the impression that anybody here or elsewhere vouching for Kliman actually fully understands his arguments, they get the firemen thing (but not whether it's a fitting comparison) and like his conclusions and want them to be true, but everything between apparently looks as foggy to them as it is to me, perhaps more so.

Why would they want it to be true? If cockshott was right we wouldn't even have to talk about the transformation problem.

I always thought that according to Marx price is a result of value interacting with various other factors such as demand, monopoly/oligopoly etc.. Since I've read Kapital quite a lot of time has passed so my memory most certainly has a lot of holes in it. Could someone explain to me what according to Kliman the ltv actually entails?

So basically in a simple economy with equalization of profit rates or disruption of supply and demand price is just the expression of the value of a commodity in money. But we don't have a perfect economy and in real life competition leads to a tendency for profit rates to equalize so that everyone has the same rate of return. This however seems to contradict the theory of surplus value because if the theory of surplus value were true more labor intensive industrys should produce more value and get more profit (because dead labor can't produce new value). So Marx came up with the idea of the transformation of values into prices of production. Basically the "price of production" is the cost price or cost of production of the commodity plus the percentage of the total surplus value each capitalist gets. So while value rarely equals price the total value is equal to the total price and the total profit is equal to the total surplus value.

Thanks. Is that the argument that is elaborated upon in reclaiming marx's capital? I'm thinking about reading it.

After reading the parody, I'm now 95 % sure Kliman got it wrong. The joke goes like this: Guy who prides himself in doing SCIENCE reports a correlation between Christianity and total income in an area in the US, Christianity measured in total number of Christians. He comes to the conclusion that there is some significant correlation. He also tests that for other religions and concludes that Christianity is special in predicting total income.

Of course, this happens because we are not talking about income per head here, and a group's income correlates with the size of the group and being Christian is common, so there is good correlation between the number of Christians and the number of people, and the ratios of the amount of people of a minority religion in different areas doesn't correlate quite as well with the respective ratios of the amount of people in general in these areas.

So, why isn't that knee-slapper a sound rebuttal? Because the correlation is there. We can tell the economist in this hypothetical story that he is a bit dull and there are better measures (just use total population), but the economist is still correct that it works better with Christians than with e.g. Jews, who are less evenly spread across the country. This brings us back to the aforementioned weakness of a contrarian: being unable to see a statement as true when it is also dull. I also strongly suspect that, as the paper goes on rephrasing some particular statistical maneuvers like that double-log shoopdawoop thing, Kliman isn't enough of a math guy to actually understand those (nor his fans, nor the fans of the other side, people just seem to pick based on the conclusions they like to hear).

In the Christian story, mass conversions to Islam all over the country would reduce the quality of the Christianity-based predictor of income in an area and increase the quality of the Islam-based predictor, which is certainly an argument for how superficial the theory of that hypothetical economist is. But what is the faith-conversion analog here for the labour-time based theory of price estimates? Does the analog show in a similar manner how superficial the theory is? Does it even exist?

No. This is explained in capital volume 3 what he explains in RMC is this:
There is also a huge controversy over this transformation problem basically set in place by one guy. So after Marx death a russian economist named Bortkiewicz thought he found a contradiction in Marx's math. The inputs to the production period in his math were in values not prices (at least in the example math) and capitalists actually buy things at prices. So what he did is set up a table that operates in simple reproduction. So basically the exact quantitys of physical goods are reproduced each production period. He assumes that the price of inputs must be equal to the price of outputs and tried to solve. When he did this he found that the value's and price's of all the commoditys changed (and because he had set input prices to equal output prices total value =/= total price) he created a solution but it could only uphold one of the equalitys at a time and only be arbitrarily enforcing it. It also made the falling rate of profit incoherent. The TSSI and kliman attempt to save Marx's value theory by making 2 assumptions. 1. There is only one cost price. The inputs into the production period are in the "value of the price" not the value of the commodity, basically price is determined from value, they are not dual seperate systems. and 2. Input prices =/= output prices and can change over a production period. The book is an explanation of the controversy and the TSSI solution. Its definitely a good read.

Because it's contrary to Marx, who said prices of production are what real-world prices actually are closer to. And you get the result that industries where for each dollar of the owner spends on it a rather large percentage goes into the machinery and there isn't much fresh living labour that sets into motion this dead labour, there wold be low profit rates systematically, which leads to the question why anybody would run such a business. (It could still happen though, even in a world where everybody knows that systematic discrepancy, under the condition that people running a business don't have good foresight about whether they have to put a lot of money into dead labour or not, so that they basically stumble into situations where they have to put more money into dead labour and accept a lower profit rate over losing even more money.)

Have you read Capital I–III and if not, how can you be sure it's right?


Okay, thanks.

You are welcome

What are they aiming for then? If someone wanted to show a correspondence between labor-values and prices, surely one would want to do this on a commodity-by-commodity basis.

This is, of course, irrelevant. The causation is evidently not there, which is the heart of the matter.

The purpose of Kliman's counterargument, as far as I understand it, is to show that the existence of these kinds of aggregated correlations do not imply correlation on an individual level. Moreover, he claims that, in both the parody case and the value/price case, the correlation does in fact disappear when looking at the per-unit values. I cannot whether his (or anyone else's) computations on the econometric side of things are correct, as I am unfamiliar with these things.

Why does there need to be a labor/price analogue to see the fault in the theory? It suffices to check that the correlation disappears in the per-unit case, just like the "Christianity causes higher income" theory collapses when the data is viewed per capita.

I really am not sure what you are trying to say with this. Tautological statements are of course true, but they do not give any new information because they beg the question (place the conclusion in the hypothesis).

The correlation in the hypothetical story is there, and for having predictive power correlation is enough.
Counterargument to whom? Shaikh et all do not claim that for any pair of commodities you are guaranteed to find good correlation. That was already said in the thread. Are you or any other people who tend to be more in Kliman's camp on the questions of the thread actually familiar with the other side, or is that all based on how Kliman talks about them?
Why does the key thing of a story that is claimed to be a metaphor need to correspond well to the concept the story is claimed to be a metaphor for, hmm?

Question for those who don't believe there is correlation between labour values and prices: Which of the following statements do you disagree with?
1. Firms don't have access to infinite credit, so to survive you usually have to sell at a price which at least covers your costs.
2. There is correlation between prices of products and services of a firm and what it costs to run that firm (of course, there can be a few extreme deviations because of stuff like patents).
3. A firm's wage bill is usually a big part of these costs, and of course that's also usually true for the other firms it buys inputs from.

You understand what follows from that, right? So, if Kliman says that the correlation disappears when "corrected for firm size", could it be that he is doing something nonsensical here?

Here is a "hilarious response" to Kliman (hilarity levels controlled for being hilarious to an economist, in other words not hilarious at all): "House size is said to correlate with wealth. However, once you control house size for number of windows the correlation shrinks drastically. Haters say that by my approach I'm actually left with comparing houses that are usually not much different in size to begin with and so shouldn't expect much difference according to that same theory, but I don't understand their more complex statistical tools and math is bourgeois anyway, so it's them who must be wrong."

Did you?