Productivity Does Not Explain Wages

Does productivity explain income? I asked this question in a previous post. My answer was a bombastic no. In this post, I’ll dig deeper into the reasons that productivity doesn’t explain income. I’ll focus on wages.

The evidence

Let’s start with the evidence trumpeted as proof that productivity explains wages. Looking across firms, we find that sales per worker correlates with average wages. Figure 1 shows this correlation for about 50,000 US firms over the years 1950 to 2015.

sales_wages
Figure 1: The correlation between a firm’s average wages and its sales per worker. Data comes from Compustat. To adjust for inflation, I’ve divided wages and sales per worker by their respective averages (in the firm sample) in each year. I’ve shown stock tickers for select firms.

Mainstream economists take this correlation as evidence that productivity explains wages. Sales, they say, measure firms’ output. So sales per worker indicates firms’ labor productivity. Thus the evidence in Figure 1 indicates that productivity explains (much of) workers’ income. Case closed.

The problem

Yes, sales per worker correlates with average wages. No one disputes this fact. What I dispute is that this correlation says anything about productivity. The problem is simple. Sales per worker doesn’t measure productivity.

To understand the problem, let’s do some basic accounting. A firm’s sales equal the unit price of the firm’s product times the quantity of this product:

Sales = Unit Price × Unit Quantity

Dividing both sides by the number of workers gives:

Sales per Worker = Unit Price × Unit Quantity per Worker

Let’s unpack this equation. The ‘unit quantity per worker’ measures labor productivity. It tells us the firm’s output per worker. For instance, a farm might grow 10 tons of potatoes per worker. If another farm grows 15 tons of potatoes per worker, it unambiguously produces more potatoes per worker (assuming the potatoes are the same).

The problem with using sales to measure productivity is that prices get in the way. Imagine that two farms, Old McDonald’s and Spuds-R-Us, both produce 10 tons of potatoes per worker. Next, imagine that Old McDonald’s sells their potatoes for $100 per ton. Spuds-R-Us, however, sells their potatoes for $200 per ton. The result is that Spuds-R-Us has double the sales per worker as Old McDonald’s. When we equate sales with productivity, it appears that workers at Spuds-R-Us are twice as productive as workers at Old McDonald’s. But they’re not. We’ve been fooled by prices.

The solution to this problem seems simple. Rather than use sales to measure output, we should measure a firm’s output directly. Count up what the firm produces, and that’s its output. Problem solved.

So why don’t economists measure output directly? Because the restrictions needed to do so are severe. In fact, they’re so severe that they’re almost never met in the real world. Let’s go through these restrictions.

Restriction 1: Firms must produce identical commodities

To objectively compare productivity, you have to find firms that produce the same commodity. You could, for instance, compare the productivity of two farms that produce (the same) potatoes. But if the farms produce different things, you’re out of luck.

Here’s why. When firms produce different commodities, we need a common dimension to compare their outputs. The problem is that the choice of dimension affects our measure of output.

To see the problem, let’s return to our two farms, Old McDonald’s and Spuds-R-Us. Suppose that Spuds-R-Us produces 10 tons of potatoes per worker. Tired of growing potatoes, Old McDonald’s instead grows 5 tons of corn per worker. Which workers are more productive?

The answer depends on our dimension of analysis.

Suppose we compare potatoes and corn using mass. We find that Spuds-R-Us workers (who produce 10 tons per worker) are more productive than Old McDonald’s workers (who produce 5 tons per worker).

Now suppose we compare potatoes and corn using energy. Furthermore, imagine that corn has twice the caloric density of potatoes. Now we find that workers at Spuds-R-Us (who produce half the mass of food at twice the caloric density) have the same labor productivity as Old McDonald’s workers.

The lesson? Unless two firms produce the same commodity, productivity comparisons are subjective. They depend on the choice of dimension.

Restriction 2: Firm output must be countable

When you read economic textbooks, it’s clear that the discipline of economics is stuck in the 19th century. Firms, the textbooks say, produce stuff.

But what about all those other firms that don’t produce stuff? What is their output? What, for instance, is the output of Goldman Sacks? What is the output of a high school? What is the output of a hospital? What is the output of a legal firm?

Yes, these institutions do things. But it defies reason to give these activities a ‘unit quantity’. In other words, it defies reason to quantify the output of these institutions.

Restriction 3: Firms must produce a single commodity

Complicating things further, we can objectively measure output only when firms produce a single commodity. If a firm produces two (or more) commodities, its output is affected by how we add the commodities together.

To see the problem, let’s return to Old McDonald’s and Spuds-R-Us. Suppose that both farms have diversified their production. Spuds-R-Us produces 5 tons of potatoes and 1 ton of corn per worker. Old McDonald’s produces 1 ton of potatoes and 5 tons of corn. Which workers are more productive?

The answer depends on our dimension of analysis. In terms of mass, both farms produce 6 tons of food per worker. So labor productivity appears the same. But suppose we measure the output of energy. Again, we’ll assume that corn has double the caloric density of potatoes. Suppose corn contains 2 GJ (gigajoule) per ton, while potatoes contain 1 GJ per ton. Now we find that Old McDonald’s workers are about 60% more productive than workers at Spuds-R-Us. Here’s the calculation:

Spuds-R-Us:
5 tons potato × 1 GJ / ton + 1 ton corn × 2 GJ / ton = 7 GJ

Old McDonald’s:
1 ton potato × 1 GJ / ton + 5 ton corn × 2 GJ / ton = 11 GJ

This ‘aggregation problem’ is why the neoclassical theory of income distribution assumes a single-commodity world — a world in which everyone produces and consumes the same thing. In this one-commodity world, we can measure productivity unambiguously. In the real world (with many commodities) productivity depends on our choice of dimension.

The severity of the problem

Let’s take stock. If we want to measure productivity objectively, the restrictions are severe:

  1. Firms must produce the same commodity
  2. This commodity must be countable
  3. Firms must produce only one commodity

These conditions are so stringent that they’re rarely met in the real world. This is a bit of a problem for neoclassical theory. It proposes that everyone’s income is explained by their productivity. But only in the rarest of circumstances can we measure productivity objectively.

It’s hard not to laugh at this predicament. It’s like Newton proclaiming that gravitational force is proportional to mass. But in the next sentence he realizes that mass can be measured only in the rarest of circumstances.

The neoclassical sleight of hand

Neoclassical economists don’t think of themselves as Newtons who can’t measure mass. Instead, economics textbooks don’t even mention the problems with measuring productivity. In these textbooks, all seems well in neoclassical land.

But all is not well. Neoclassical economists perpetuate their fantasy by relying on a sleight of hand. Here’s what they do.

First, economists argue that the purpose of all economic activity is to give consumers utility. Buy a potato and you get utility. Buy a cigarette and you get utility. Utility, economists say, is the universal dimension of output. By measuring utility, we can compare the output of any and all firms (no matter what they produce).

After proclaiming that utility is the universal dimension of output, economists pull their trick. Utility, they say, is revealed through prices. So a painting worth $1000 gives the buyer 1000 times the utility as a $1 potato.

With this thinking in hand, economists see that a firm’s sales measure its output of utility:

Sales = Unit Price × Unit Quantity

Sales = Unit Utility × Unit Quantity = Gross Utility

So sales become a universal measure of utility, and utility is the universal measure of output. Now, when we compare sales per worker to wages (as in Figure 1), economists proclaim that we’re comparing productivity to wages.

Except we’re not.

The problem is that this whole operation is circular. The idea that prices reveal utility is a hypothesis. And as every good scientist knows, you can’t use your hypothesis to test your hypothesis. But that’s what neoclassical economists do. They assume that one aspect of their theory is true (the link between prices and utility) to test another aspect of their theory (the link between productivity and income). This is a big no no.

Why do economists use this circular reasoning? Probably because they don’t know they’re doing it. Economists take as received wisdom the idea that prices reveal utility. But this is just a hypothesis. In fact, it’s a bad hypothesis. Why? Because we can never measure utility independently of prices.

Why are sales related to wages?

Whenever I go through the logic above, mainstream economists will retort: “But look at the correlation between wages and sales! How can this not show that productivity explains wages?” Their reasoning seems to be that, absent an alternative explanation, this correlation must support their hypothesis.

In No, Productivity Does Not Explain Income, I gave an alternative explanation. The correlation between wages and sales per worker, I argued, follows from accounting principles.

Sales isn’t a measure of output. It’s an income stream. Once earned, this income gets split by the firm into different categories. Some of it goes to workers. Some of it goes to other firms (as non-labor costs). And some of it goes to the firm’s owners as profit.

Dividing an income stream

Figure 2: Dividing a firm’s income stream. Accounting principles dictate that a firm’s sales get divided into profits and wages.

By definition, the terms on the left must sum to the terms on the right. So it’s not surprising that we find a correlation between wages and sales. They’re related by an accounting identity.

In comments on No, Productivity Does Not Explain Income (and on other sites), some economists pounced on this argument, saying it was fatally flawed. And in hindsight, I admit that I wasn’t clear enough about my reasoning. I was thinking about the real world. But the economists who critiqued my reasoning were thinking in terms of pure mathematics.

To frame the debate, let’s think about something more concrete than income. Let’s think about volume. In rough terms, the volume of an object is the product of its length, width and height:

V = L × W × H

Now, let’s pick a dimension — say length. Will the length of an object correlate with its volume? In general terms, no. I can make an object with any volume using any length. I just have to adjust the other dimensions appropriately. By doing so, I can make a cube have the same volume as a box that is long and thin.

So in pure mathematical terms, the accounting definition of volume doesn’t lead to a correlation between length and volume.

But when we look at real-world objects — like animals — we will find a correlation. If we took all the species on earth and plotted their length against their volume, we’d expect a tight correlation. A bacteria has a small length and a small volume. A blue whale has a big length and a big volume. Fill in the gaps between and we should get a nice tight line.

The reason for this correlation is that animals cannot take any shape. You’ll never find an animal that is a mile long and a few micrometers wide. Such a beast doesn’t exist. Yes, the shapes of animals vary. But in the grand scheme, this varation is small. As a first approximation, animals are roughly cubes. Or, if you’re a physicist, they’re spheres.

With this shape restriction, it follows from the definition of volume that animal length should correlate with animal volume. We’d be astonished if it didn’t.

So too with the correlation between sales per worker and wages. True, this correlation doesn’t follow purely from accounting principles. It follows jointly from accounting principles, and the fact that firms can’t take any form. We don’t find firms that pay their workers nothing. That’s slavery and its illegal. Similarly, we don’t find (many) firms that pay their workers the entirety of sales. That leaves no room for profit.

So in the real world, there are restrictions on how firms can divide their income stream. Here’s what these restrictions look like. In Figure 3, I’ve plotted the distribution of firms’ payroll as a portion of sales. This is the portion of sales that goes to workers. Across all firms, it’s a pretty tight distribution, clustered around 25%.

payroll_frac

Figure 3: The distribution of firm payrolls as a fraction of sales. Data is for US firms in the Compustat database over the years 1950–2015.

Yes, it’s theoretically possible for a firm to give any portion of its sales to workers. But this isn’t what happens in reality. In the real world, most firms give between 10% and 50% of their sales to workers. Just like with the shape of animals, there are real-world restrictions on the ‘shape’ that firms can take.

Given these restrictions, it’s not surprising that we find a correlation between sales per worker and wages. When a firm’s income stream grows, so does the amount going to workers.

None of this has anything to do with productivity. It’s all about income. Sales are the firm’s income. And wages are the portion of this income given to workers.

Prices … the elephant in the room

Let’s conclude this foray into neoclassical thinking. The reason that sales don’t measure firm output is because they mix unit prices with unit quantities. Yes, sales per worker correlates with wages. But the elephant in the room is prices. Greater sales may be due to greater output. But it can also be due to greater unit prices.

In many cases, price differences are everything.

Imagine that a lawyer and a janitor both work 40 hours a week as self-employed contractors. The lawyer charges $1000 per hour, while the janitor charges $20. At the end of the week, the lawyer has 50 times the sales as the janitor. This difference comes down solely to price. The lawyer charges 50 times more for their hourly services than the janitor.

The question is why?

Neoclassical economists proclaim they have the answer. The lawyer, they say, produces 50 times the utility as the janitor. Ask economists how they know this, and they’ll answer with a straight face: “Prices revealed it.”

It’s time to recognize this sleight of hand for what it is: a farce. The reality is that we know virtually nothing about what causes prices. And we will continue to know nothing as long as researchers believe the neoclassical farce.

Further reading

The Aggregation Problem: Implications for Ecological and Biophysical Economics. BioPhysical Economics and Resource Quality. 4(1), 1-15. SocArxiv Preprint.

30 comments

  1. I am encouraged to see that you accept that your previous post, and the main claim in a peer-reviewed article, was wrong. Yet, you are doubling down, with an even flimsier explanation, all in the sacred goal of making neo-classical economists look like a bunch of idiots.

    Let me first follow your metaphor of the correlation between length and volume of organisms.

    The correlation that exists between length and volume in organisms has absolutely nothing to do with the fact that length is one dimension of volume. For example, there’s no correlation between the width of a hair on a person head (or among all persons head) and its volume. That is b/c width is more or less constant, while length vary, and it is the length that drives the variation in volume. Moving to two dimensions for a second – there’s no correlation between a bridge width and its area, since a bridge area is overwhelmingly determined by its length, not its width. And so on. The accounting here has zero relevance here. Zero.

    The strong correlation between an organism length and volume has everything to do with the underlying forces that shape organisms, namely, evolution. A very disproportional organism will not be able to survive very well, b/c its motion will be limited, it will be fragile to shocks, etc.

    To conclude, the fact that the correlation exists is meaningful, has nothing to do with accounting truisms, and has everything to do with the forces that shape the size of the variables in question. It’s a super interesting, and important, fact that this correlation exists, and a theory to explain it is very very much called for.

    Back to firms. You write:

    “Yes, it’s theoretically possible for a firm to give any portion of its sales to workers. But this isn’t what happens in reality. In the real world, most firms give between 10% and 50% of their sales to workers. Just like with the shape of animals, there are real-world restrictions on the ‘shape’ that firms can take.”

    And what are these “real-world restrictions”? You mention the extremes, like that a firm can’t pay zero and won’t pay all (or more than all) of its revenue.* But the correlation is much tighter than what only these very general bounds would imply. And any observer should ask themselves – why is it the case that this strong correlation exists? What are these “real world restrictions”? Neoclassical economics has one explanation, which you seem to not like very much. But an explanation is very much called for.

    I will not get into the whole debate of measuring utility (all agree that utility is not directly measurable), its relation to prices (you are between way oversimplifying and wrong about this issue).

    But it seems to me that your argument really boils down to claiming that the word “productivity” is a misnomer. Fine by me. Let’s call this – the contribution of a worker to a firm added value – a banana.

    Is a theory that explains the relationship between a worker’s wage and a worker’s banana circular or useless? Of course not. For one, it contradicts at least the simplest (and still common) interpretation of Marx theory of wages. For another, it has policy implications: if it is true, then increasing firms revenue (e.g. by improving infrastructure) will increase wages. The theory also relates a worker education and training to her banana, and therefore to her wage. This also is not circular, contradicts some theories, and has policy implications. Whether we should use the word productivity here seems to me like a completely unimportant point.

    Firms have different added value per worker. Firms also pay different wages. This, like many related patterns, are some of the most important patterns in economic life. A coherent theory that relates the to each other isn’t circular by any sense of the term “circular”, and it’s immensely useful both for pure understanding, and for policy. Your only critique, really, is that you don’t like the use of the word “productivity” in that context. Fine. use any other. It’s neither a valid, nor, pardon my french, very interesting critique of neo-classical economics.

    *(You are actually wrong about the latter – a very large number of firms pay in wages more than their current revenue, by borrowing from capital markets. But this is borderline gotcha. In the ling run it is true that firms will not pay all of their revenue in wages)

    Like

  2. Neoclassical economist: You earn more if you’re more productive.

    Scientist: How do you know if you’re more productive?

    Neoclassical economist: You’re more productive if you earn more income.

    Scientist: Isn’t that circular reasoning?

    Neoclassical economist: It has policy implications.

    Scientist: I’m going to walk away now.

    Like

    • This is frustrating…. your new comment here is of course a silly straw man. How can anyone believe that a whole field of study would make such obviously idiotic argument is beyond me. Not directly, not indirectly, not implicitly, this is simply preposterous to think such an obvious folly will escape the eyes of a whole field.

      More importantly, it’s – again – a completely different argument from anything you said before. Let me rephrase this conversation in terms of what you said before:

      neoclassical economist: you earn more if you are more productive
      scientist: how do you know if you are more productive?
      neoclassical economist: you are more productive if the firm you work for has higher revenue (or value added).
      scientist: isn’t that circular reasoning?
      neoclassical economist: what? why? how is that possibly circular? What on earth are you talking about?! Are you a real scientist?

      what is frustrating here is that you, yourself, in the post we are all talking about, presented a completely different kind of evidence about the relationship between productivity and wages. Why are you going now to this silly straw man?

      Like

      • Assafzim:

        You’re just arguing by definition. You have defined productivity to be “higher revenues,” and so when you see higher revenues you claim the firm is more productive. That’s the circular logic.

        There’s a bigger picture here. The real domain can only serve as a causal explanation for the nominal domain assuming you have a way of measuring the real domain INDEPENDENTLY of the nominal domain. If your definition of the real domain (like productivity) depends on the nominal domain (like revenues), then clearly you cannot use your “real” variable as a causal explainer.

        Let me put in more obvious terms: the independent variable cannot be dependent on the dependent variable, otherwise it explains nothing.

        This is the basic logic neoclassical theory ignores (they’re aware of it, at least some economists, but they tuck it aside).

        Like

      • EK

        No, I am not.

        Your first point does not make any sense. The discussion is about the correlation between a worker’s wage and her contribution to the revenues of the firm. It is not about the relationship between a firm’s sales and its “productivity”. OF COURSE if you define productivity as higher revenue then it would be incredibly stupid to say that the higher revenue implies higher productivity. OF COURSE!! Just how stupid do you think I, or any other neo-classical economist, are?

        What I did say is that there is nothing circular about
        (1) a theory that relates a workers wage to the worker’s contribution to the firm’s revenue
        (2) there’s absolutely nothing in the definitions of any of the above variables that makes a correlation between wages and revenue an “accounting truism”.
        (3) That I don’t care if you call “contribution to a firm revenue” by any name you wish. I have no attachment to the term productivity. Call it a banana if you like, and we still have useful, logical, non-circular causal theory with evidence that support it and this evidence is not in any way trivial or an “accounting truism”.

        Big picture:
        This goes far outside the discussion here, so I will not get into it in detail. I dislike the tendency of discussions of a specific topic to divulge into “everything I don’t like about neoclassics” discussion. Blair made a very specific point (that the way “productivity” is defined makes it an accounting truism that wages will be correlated with “productivity”), on the way ridiculing neoclassical economist. This point is 100% wrong, and his argumentation sloppy. That it was published in a peer-review publication is troubling.

        I mostly disagree with the “real as causal to nominal” point. Actually, it’s more precise to say that I find it nitpicky and lacking in genuine insight, as it is not wrong. It is undoubtedly true that the infinite diversity of economic products, production methods, and consumption patterns, makes it impossible to make any general statements on the real variables without using some aggregator, and prices are a very common one**. It is also true that prices are set within the model. But I don’t see how this insight is very useful. If one insists that only “pure real” variables should be allowed as causal explanation of nominal variables, you basically rule out the possibility to ever say anything about economic activity.
        More importantly – you can have excellent, non-circular, and very useful models that use one nominal variable to explain the other. Testing these theories is in no way a foregone conclusion, and the tests are valid, and sophisticated even when both the dependent and the independent variables are nominal. I think you are useing the term “independent” here a bit vaguely, mixing the more standard meaning of “is not directly, causaly, affected by” with a more vague “set in a completely separate system. A theory with empirical tests realting the number of years you smoke with the number of years you live is valid, and useful, even though both dependent (years you live) and independent (years you smoke) are measured in the same unit, and are determined in the same large model.

        But, I promised I won’t go on this tangent, and I already went way more than I planned. So with this – if you want to discuss the original point – I am here. If you want to discuss real-as-causal-for-nominal, go for it, but I will not respond any more.

        **BTW, this is true, of course, for Blair’s model of levels of hierarchy as well. It goes without saying that managerial relationships are vastly different between different firms, and just counting how many people, or how many levels of people, are under someone make no sense just like comparing the number of oranges I pick to the number of software code lines Joe writes.

        Like

      • assafzim,

        I want to thank you for commenting on this post. Most neoclassical economists simply ignore critiques of their theory. However, you have taken the time to respond.

        Your responses are fascinating and depressing. They are fascinating because they continuously highlight the problems I am trying to point out. You continuously use circular reasoning. The depressing part is that you seem to be incapable of seeing what you are doing. In fact, you seem not to understand the very theory that you are defending.

        You are defending something called “marginal productivity theory”. This theory makes it absolutely clear that income stems from productivity. Moreover, this theory defines productivity in a very reasonable way. Productivity is measured in terms of the output of a *single* commodity.

        By evoking a single-commodity world, marginal productivity theory achieves logical consistency. There is no aggregation problem and productivity is objectively defined. This is the logically consistent theory that you must defend.

        The problem is that it is indefensible, because the theory imagines a non-existent world. We live in a world with many commodities. In this world, productivity is ill defined.

        But what you do (and what all neoclassical economists do) is bait and switch. You evoke a logically consistent theory whose premise is systematically violated in the real world. But you are not deterred. Instead, you redefine productivity in terms of sales.

        Look through your comments and you’ll find you repeatedly do this. But once you do this, you’ve abandoned the theory you are defending. Instead, you insist we test a different theory, one that is completely circular.

        Sales are a type of income. But you insist that we call them productivity. You refuse to acknowledge the importance of prices. Instead, you are implicitly saying that firms are more productive if they can charge more for their product. Of course, you are free to define productivity any way you want. But this definition violates most people’s common sense definition of productivity. And, just to reiterate (again), it defines productivity in terms of income.

        That you cannot see this circularity is fascinating. It is a depressing testament to neoclassical religiosity.

        Like

      • Assafzim:

        You wrote:

        “The discussion is about the correlation between a worker’s wage and her contribution to the revenues of the firm. It is not about the relationship between a firm’s sales and its “productivity”.”

        No. That’s not what the central issue is about. That’s at best a tangential problem that’s not as important as you think it is. The central concern in this debate is the following simple question:

        What is productivity?

        If you can’t grapple with this question, then you have nothing meaningful to say about a worker’s salary and her “contribution to the revenues of the firm,” which is your roundabout way of saying “productivity.” If you can’t understand the aggregation issues with explaining a firm’s higher revenue by invoking productivity, then you need to step back and read a few more things about economics. I suggest starting with the Cambridge capital controversy.

        “OF COURSE if you define productivity as higher revenue then it would be incredibly stupid to say that the higher revenue implies higher productivity. OF COURSE!!”

        But that’s exactly what you did. Go and read your own post. You wrote, “you are more productive if the firm you work for has higher revenue.” By holding productivity as a metaphysical dangler, you are giving a de facto definition with these kinds of statements.

        “Just how stupid do you think I, or any other neo-classical economist, are?”

        You should not have asked me that. Considering that Solow claimed a great discovery and won a Nobel Prize on the basis of a useless identity equation that proved nothing about GDP or production, pretty damn stupid. Merton and Scholes ran LTCM into the ground, or at least their model partly did, only a year after winning their Nobel. Irving Fisher, and one cannot even make this stuff up, actually made a direct analogy between people and particles and utility and energy in one of his books. Much of early neoclassical theory was based on spurious and ludicrous analogies with physics, completely ignoring issues about emergence and aggregation along the way. In 1938, Samuleson presented his revealed preference theory as an alternative to ordinal utility theory. In 1950, the two were shown to be equivalent and Samuelson celebrated that fact as some kind of vindication. 12 years ago is just so passe.

        I ask this in all honesty: is there any meaningful about the world that neoclassical theory adequately explains? I certainly can’t think of anything. Prices? Nope. Supply and demand cycles are almost epiphenomenal; they’re just the product of other social forces and political decisions. When oil prices quadrupled from October 1973 to January 1974, it’s not because the world suddenly ran out of oil. It’s because OPEC cut production to protest Western support for Israel in the Yom Kippur War. The MSRP of the car at the dealership is based less on demand and more on upper managers and capitalists in some corporate boardroom hashing out what they need to do to hit certain profit rates, all based on considerations about what their competitors are up to. Can it explain trade? Doubtful. Nearly all powerful nations got wealthy in part through strategic and protectionist trade policies that allowed them to develop their domestic industries.

        Neoclassical theory is capitalist propaganda decorated with useless mathematical abstractions. It made a fundamental mistake with its ontology by hyper-focusing on the individual and ignoring social relations and interactions. As a result, it can’t explain anything. So what you’ve seen in economics recently are things like the “institutional turn” and the “cultural turn” and the “behavioral turn.” Much of what used to neoclassical theory is shading into Post-Keynesian thought or other economic frameworks. The holdouts, and they are still powerful, continue dabbling in futile projects.

        “(1) a theory that relates a workers wage to the worker’s contribution to the firm’s revenue
        (2) there’s absolutely nothing in the definitions of any of the above variables that makes a correlation between wages and revenue an “accounting truism”.
        (3) That I don’t care if you call “contribution to a firm revenue” by any name you wish. I have no attachment to the term productivity. Call it a banana if you like, and we still have useful, logical, non-circular causal theory with evidence that support it and this evidence is not in any way trivial or an “accounting truism”.”

        On 1: it’s extremely difficult, if not impossible, to measure a “worker’s contribution to the firm’s revenue.” I mean, that’s kind of what this debate is all about. How do you define or measure productivity? If you measure different things, you will get different answers.

        On 2: that’s not what the main issue is for me. You might be having a different debate with you and yourself, but I’m concerned about figuring the link between greater productivity, whatever that means, and a firm’s revenue.

        On 3: that’s absurd. Of course it’s circular reasoning. You still have not provided any conception of productivity that exists independently of the nominal domain (wages, revenues, etc). Until you do, it is and will forever remain circular reasoning to say that productivity explains higher revenues.

        “Actually, it’s more precise to say that I find it nitpicky and lacking in genuine insight, as it is not wrong. It is undoubtedly true that the infinite diversity of economic products, production methods, and consumption patterns, makes it impossible to make any general statements on the real variables without using some aggregator, and prices are a very common one**.”

        But there are a million aggregation problems with using prices to indicate something about scale. Meanwhile, there are far fewer if you use other units, like energy. So there are alternatives to prices.

        “If one insists that only “pure real” variables should be allowed as causal explanation of nominal variables, you basically rule out the possibility to ever say anything about economic activity.”

        Well that’s clearly false. Look at energy consumption over the last two centuries around the world. It has skyrocketed. That says something important right there: about the Industrial Revolution, the rise of capitalism, etc. Blair himself has done important work on this issue, looking at the links between energy consumption and hierarchical organization. All of this has been done by looking purely at real variables. Meanwhile, the nominal domain reflects the real domain only indirectly and through highly complex causal processes. The nominal domain is largely a product of social power and hierarchy, and these are in turn more reflections of biophysical activities. We can get into all of this you like.

        “you can have excellent, non-circular, and very useful models that use one nominal variable to explain the other. Testing these theories is in no way a foregone conclusion, and the tests are valid, and sophisticated even when both the dependent and the independent variables are nominal.”

        Perhaps you could find those models, especially if you mix them up with some other variables, but you won’t find them in the canonical versions and elements of neoclassical theory (marginal productivity, time preference, utility, etc).

        Like

      • I also wanted to address your example about an empirical result relating the number of smoking years to the number of years alive. You claimed that such a result is “valid” and “useful.”

        But this is a very faulty and deceptive analogy. You see, we actually have a compelling causal narrative about why those two things are statistically related, and this causation can be established on the basis of other evidence besides this particular correlation itself. For example, scientists and doctors can actually probe inside the lungs of people who regularly smoked and can observe the physical deterioration of their lungs first-hand.

        Now let’s turn to our example. What exactly is the causal implication of noticing that employee salaries are correlated with firm revenues? You haven’t really addressed this question, not in any substantive way at least. You seem to think it has something to do with productivity, but you can’t quite put your finger on it because you refuse to tell us what productivity means.

        Like

      • As always… EK flying off to the Cambridge capital controversy (Yigal – I hope you are reading this!), and LTCM, and Solow and what not. Like I said – I do not like it when specific concrete discussion become everything I dislike about neoclassics. So – not going there.

        Blair –
        I understand perfectly well. All your current obfuscation, however, does not change a very clear fact: your post, and your peer-reviewed published article made a point that is 100% wrong. Not almost, not “I wasn’t clear enough about my reasoning”. Wrong. 100%.

        Let’s recap here for a second with some tidbits that are crystal clear (if nothing else, you are clear in your writing):

        “Fortunately (for themselves), neoclassical economists don’t play by the normal rules of science. If you browse the economics literature, you’ll find an endless stream of studies claiming that wages are proportional to productivity. Under the hood of these studies is a trick that allows productivity to be universally compared. And even better, ***it guarantees that income will be proportional to productivity***.” and

        “Thus sales per worker will obviously correlate with wages. Given our accounting definition, it has too.” and

        “so economists test their theory of income distribution by assuming it is true.”

        This is dead wrong. 1+1=3 kind of wrong. The definition does not guarantee it any more than the definition of “spending on groceries” guarantees that it will correlate with income. That’s what I commented on. nothing more, nothing less.
        Your later move from “accounting truism” to accounting principles and real-world restrictions was just silly and I don’t believe even you yourself think you made a valid point there. You are basically saying “given that firms in real world have a strong correlation between sales and wages (which is basically what the data about paying 10%-50% of sales in wages mean), then it’s meaningless that/has to be by definition that this correlation exists”. How is that an argument for anything?

        And please, spare me the “it’s fascinating” nonsense. I understand everything you are saying perfectly well. I don’t want to reopen everything anyone ever said about productivity. I commented on a specific, and dead wrong argument.

        Like

      • No. It’s not “The point”. It’s a completely different point from what you originally said. You are constantly changing your arguments, throwing in utility theory here, and “restrictions” there (whatever that means), and now you just change the question completely, saying the issue is what we can or cannot learn from the strong correlation between wages and sales to the notion of “productivity”.

        But – do I need to copy paste it all over again? – your argument was that this correlation “has to” exist because of “accounting truism”. Which is (1) completely different from what you say now is the point and (2) dead wrong.

        Ok, I think we both made our point. assafzim out.

        Like

      • If you read and understood my writing, you’d know that this was the point all along. If you want to test marginal productivity theory, you can’t define productivity in terms of income.

        You’ve offered no arguments addressing this critique. Literally nothing except heckling.

        Like

      • Assafzim:

        You wrote:

        “As always… EK flying off to the Cambridge capital controversy (Yigal – I hope you are reading this!), and LTCM, and Solow and what not. Like I said – I do not like it when specific concrete discussion become everything I dislike about neoclassics. So – not going there.”

        But I concretely addressed your point, separately from the Cambridge capital controversy. You have absolutely no idea how a worker’s “contribution” (you know, not productivity) is causally related to the revenue of the firm. And you never will until you specify what you mean by productivity (sorry, “contribution”). CCC is only relevant to the extent that it illuminates some important themes in this discussion, hence why I mentioned it.

        Like

      • EK –

        A worker’s contribution to the firm revenue is defined, quite as it sounds, as “the difference between the firm’s revenue when this worker is employed and the firm’s revenue when all else is equal, but the worker’s is not employed”. A prediction of the theory is that this will be equal to the worker’s wage (I am being loose here about revenues/profits/added value. not important). Nothing circular here. Nothing. Tehre’s a clear causal direction going one way.

        Now, what we all agree to is that this – contribution of a worker to revenue – is very hard, usually impossible for a researcher to measure. This is a very lively debate in the profession as well. This is an empirical challenge, even if a hard one. Much literature about it. But it is not a conceptual problem, it does not imply circular reasoning, and it is not what Blair original post was about. And now for real – assafzim out.

        Like

      • Assafzim writes:

        “A worker’s contribution to the firm revenue is defined, quite as it sounds, as “the difference between the firm’s revenue when this worker is employed and the firm’s revenue when all else is equal, but the worker’s is not employed”. A prediction of the theory is that this will be equal to the worker’s wage”

        Right, so this is exactly the claim of marginal “contribution” theory. A worker’s “contribution” is equal to the marginal revenue “contribution” of labor. In other words, contribution is being defined in terms of the amount of money that the worker brings to the firm. How do I know which worker contributed the most? The one that brought the most money to the firm. Not circular at all!

        I think what you need is not a lecture on economics, but one on logic and philosophy. Look up the descriptivist theory of names (not a great theory of reference in general, but applicable here). Here’s what you’re doing:

        Premise 1: Define contribution as X (where X is “the difference between the firm’s revenue….”)

        Premise 2: Claim that X is equal to the worker’s wage

        Note: don’t flatter yourself, there’s no prediction whatsoever here. You can’t actually measure X. You can only measure the worker’s wage, and so you’re asserting, by fiat, that the two are equal.

        Conclusion: Because contribution is X, and X is the worker’s wage, then contribution is equal to the worker’s wage by transitivity.

        Like I said in the beginning, all you’re doing is arguing by definition.

        Also, the 21st century called: marginality is an absolutely useless concept. In reality, productivity is integrated between multiple groups of people using highly complex and variable tools, machines, and computers. Productivity is not exclusively an individual trait, and thus it’s impossible to measure, in any way that’s not laughable and absurd, the “net amount of money” that a worker brings in for a company.

        Like

      • That we cannot directly observe a firm’s contribution to a firm’s revenue does not mean there’s no prediction here. There is a prediction here, and it’s hard to test it. That’s all very well known and widely discussed. Let me conclude (this time for real!) with this from Richard Feynman:

        “What I had intended to do was to find out whether they thought theoretical constructs were essential objects. The electron is a theory that we use; it is so useful in understanding the way nature works that we can almost call it real. I wanted to make the idea of a theory clear by analogy. In the case of the brick, my next question was going to be, “What about the inside of the brick?”–and I would then point out that no one has ever seen the inside of a brick. Every time you break the brick, you only see the surface. That the brick has an inside is a simple theory which helps us understand things better. The theory of electrons is analogous. So I began by asking, “Is a brick an essential object?”

        Like

      • “That we cannot directly observe a firm’s contribution to a firm’s revenue does not mean there’s no prediction here. There is a prediction here, and it’s hard to test it.”

        In order to say that a worker’s marginal contribution, as you defined it, is equal to her wage, you must actually demonstrate that something like “marginal contribution” actually exists. Like, in external reality, not in someone’s imagination. You have no way of measuring it, yet you’re sure it exists (you must be, otherwise all your claims fall apart like a series of dominoes). Now, I certainly don’t think that something only exists if you can measure it. I admit, and I say this is a naturalist, that there might be things out there that simply cannot be experimentally confirmed, but which may nonetheless exist in external reality. Strings and the multiverse come to mind.

        But I wouldn’t build a scientific theory on these kinds of metaphysical constructs.

        If you want to do philosophy, I’m happy to talk. I love philosophy.

        PS I also love Feynman. He was a brilliant physicist. But he was not a brilliant philosopher. We have very good experimental reasons to think that electrons exist, even if we can’t see them directly with our eyes.

        Like

  3. ‘This “equilibrium” graph (Figure 3) and the ideas behind it have been re-iterated so many times in the past half-century that many observes assume they represent one of the few firmly proven facts in economics. Not at all. There is no empirical evidence whatsoever that demand equals supply in any market and that, indeed, markets work in the way this story narrates.
    We know this by simply paying attention to the details of the narrative presented. The innocuous assumptions briefly mentioned at the outset are in fact necessary joint conditions in order for the result of equilibrium to be obtained. There are at least eight of these result-critical necessary assumptions: Firstly, all market participants have to have “perfect information”, aware of all existing information (thus not needing lecture rooms, books, television or the internet to gather information in a time-consuming manner; there are no lawyers, consultants or estate agents in the economy). Secondly, there are markets trading everything (and their grandmother). Thirdly, all markets are characterized by millions of small firms that compete fiercely so that there are no profits at all in the corporate sector (and certainly there are no oligopolies or monopolies; computer software is produced by so many firms, one hardly knows what operating system to choose…). Fourthly, prices change all the time, even during the course of each day, to reflect changed circumstances (no labels are to be found on the wares offered in supermarkets as a result, except in LCD-form). Fifthly, there are no transaction costs (it costs no petrol to drive to the supermarket, stock brokers charge no commission, estate agents work for free – actually, don’t exist, due to perfect information!). Sixthly, everyone has an infinite amount of time and lives infinitely long lives. Seventhly, market participants are solely interested in increasing their own material benefit and do not care for others (so there are no babies, human reproduction has stopped – since babies have all died of neglect; this is where the eternal life of the grown-ups helps). Eighthly, nobody can be influenced by others in any way (so trillion-dollar advertising industry does not exist, just like the legal services and estate agent industries).
    It is only in this theoretical dreamworld defined by this conflagration of wholly unrealistic assumptions that markets can be expected to clear, delivering equilibrium and rendering prices the important variable in the economy – including the price of money as the key variable in the macroeconomy. This is the origin of the idea that interest rates are the key variable driving the economy: it is the price of money that determines economic outcomes, since quantities fall into place.
    But how likely are these assumptions that are needed for equilibrium to pertain? We know that none of them hold. Yet, if we generously assumed, for sake of argument (in good economists’ style), that the probability of each assumption holding true is 55% – i.e. the assumptions are more likely to be true than not – even then we find the mainstream result is elusive: Because all assumptions need to hold at the same time, the probability of obtaining equilibrium in that case is 0.55 to the power of 8 – i.e. less than 1%! In other words, neoclassical economics has demonstrated to us that the circumstances required for equilibrium to occur in any market are so unlikely that we can be sure there is no equilibrium anywhere. Thus we know that markets are rationed, and rationed markets are determined by quantities, not prices.
    On our planet earth – as opposed to the very different planet that economists seem to be on – all markets are rationed. In rationed markets a simple rule applies: the short side principle. It says that whichever quantity of demand or supply is smaller (the ‘short side’) will be transacted (it is the only quantity that can be transacted). Meanwhile, the rest will remain unserved, and thus the short side wields power: the power to pick and choose with whom to do business. Examples abound. For instance, when applying for a job, there tend to be more applicants than jobs, resulting in a selection procedure that may involve a number of activities and demands that can only be described as being of a non-market nature (think about how Hollywood actresses are selected), but does not usually include the question: what is the lowest wage you are prepared to work for?
    Thus the theoretical dream world of “market equilibrium” allows economists to avoid talking about the reality of pervasive rationing, and with it, power being exerted by the short side in every market. Thus the entire power hiring starlets for Hollywood films, can exploit his power of being able to pick and choose with whom to do business, by extracting ‘non-market benefits’ of all kinds. The pretence of ‘equilibrium’ not only keeps this real power dimension hidden. It also helps to deflect the public discourse onto the politically more convenient alleged role of ‘prices’, such as the price of money, the interest rate. The emphasis on prices then also helps to justify the charging of usury (interest), which until about 300 years ago was illegal in most countries, including throughout Europe.
    However, this narrative has suffered an abductio ad absurdum by the long period of near zero interest rates, so that it became obvious that the true monetary policy action takes place in terms of quantities, not the interest rate.
    Thus it can be plainly seen today that the most important macroeconomic variable cannot be the price of money. Instead, it is its quantity. Is the quantity of money rationed by the demand or supply side? Asked differently, what is larger – the demand for money or its supply? Since money – and this includes bank money – is so useful, there is always some demand for it by someone. As a result, the short side is always the supply of money and credit. Banks ration credit even at the best of times in order to ensure that borrowers with sensible investment projects stay among the loan applicants – if rates are raised to equilibrate demand and supply, the resulting interest rate would be so high that only speculative projects would remain and banks’ loan portfolios would be too risky.
    The banks thus occupy a pivotal role in the economy as they undertake the task of creating and allocating the new purchasing power that is added to the money supply and they decide what projects will get this newly created funding, and what projects will have to be abandoned due to a ‘lack of money’.
    It is for this reason that we need the right type of banks that take the right decisions concerning the important question of how much money should be created, for what purpose and given into whose hands. These decisions will reshape the economic landscape within a short time period.
    Moreover, it is for this reason that central banks have always monitored bank credit creation and allocation closely and most have intervened directly – if often secretly or ‘informally’ – in order to manage or control bank credit creation. Guidance of bank credit is in fact the only monetary policy tool with a strong track record of preventing asset bubbles and thus avoiding the subsequent banking crises. But credit guidance has always been undertaken in secrecy by central banks, since awareness of its existence and effectiveness gives away the truth that the official central banking narrative is smokescreen.’
    https://professorwerner.org/shifting-from-central-planning-to-a-decentralised-economy-do-we-need-central-banks/

    Like

  4. I spent many years helping businesses and government departments solve operational problems, including productivity problems. The key question here is “do academic economists have anything useful to say about business productivity”? From what I can see, both here and elsewhere, the answer appears to be No. A few comments.

    Productivity in economics (ProdE) is not the same concept as productivity in the real world (ProdR). It doesn’t even have the same units of measure. This causes endless confusion, particularly when many economists seem unable to appreciate the difference.

    It is not enough to measure productivity (of either variant). The only useful purpose of measuring ProdR is as part of an improvement process. However, improvement also requires an understanding of the underlying business to identify productivity bottlenecks and improvement opportunities. ProdR can then be improved by redesigning the business to remove the bottlenecks and implement the opportunities.

    When I read economists talking about ProdE, the arguments seem to go as follows:

    Economists have been measuring ProdE
    It is not improving
    We think that this is a major issue
    We don’t know what is holding ProdE back
    We don’t know what to do to improve ProdE.

    This is a complete waste of time. Your profession needs to challenge itself by asking some difficult questions such as:

    What would economists need to do in order to make a useful and distinctive contribution to society regarding business productivity improvement?

    Why do academic economists study businesses the way I might study the surface of Mars? i.e. I have never been there; I don’t know anything about it; I can’t be bothered to talk to anyone who does understand it; but I’m smart enough to fool other people who also don’t know anything about it

    Why do economists not use relevant examples to challenge their own thinking?

    For example, if a business automates its production line it is perfectly possible for ProdR per worker to increase without any individual worker improving their personal productivity. It is the automated production line which is more productive. Why would any residual workers deserve a rise in their wages? Who deserves to benefit from this productivity improvement? What would happen to ProdE if the business shared the benefits of improving ProdR by reducing the price to customers by an equivalent amount?

    A final question to Blair. Why do young, smart heterodox economists spend so much time debunking “neo-classical” economics? In a decade of reading about economics, I have never understood this. Neo-classical economists don’t listen. No-one in the real world cares. Why not do something useful instead? One of the best ways for anyone to improve their personal productivity is to cut out any tasks or processes that don’t add any value to anyone.

    Like

    • Hi Jamie,

      Thank you for commenting. Nice that you brought up automation. In my own research, I’ve argued that virtually all increases in productivity have come from the increasing use of machines, which are basically a tool for converting energy into a useful form of work. This has very little to do with workers.

      Regarding debunking neoclassical economics. I don’t think heterodox economists get into these debates to try to convince neoclassical economists that they’re wrong. The hope is that the debate will be helpful for people outside the economics discipline. I have recently lamented the time I spend arguing against neoclassical economics: https://twitter.com/blair_fix/status/1196428740273823744

      As scientists, though, a major part of our job is arguing against theories we feel are wrong. That’s the only way things will change.

      Like

      • Blair,

        Thanks for replying. As I said, I have been reading about economics for about ten years. I rarely comment on academic blogs because many of the debates are sterile and endlessly repetitive. Also, most economists don’t listen to any perspectives other than their own anyway. This is equally true of heterodox and neo-classical economists, so I appreciate your reply.

        Blair: “The hope is that the debate will be helpful for people outside the economics discipline”

        I’d say that you can split non-economists into three groups:

        Group A: by far the largest group. They are not interested in economics. They are not listening to anything you say so nothing you say will be helpful to them

        Group B: the next largest group. They think that academic economists are con men who know very little about anything but regard themselves as self-styled “public intellectuals” who should offer opinions on almost everything even when no-one asks for their opinions. This type of non-economist will not listen to you until your profession has a more realistic view of its own role in society and learns to present itself with some humility. Until then, nothing you say will be helpful to them

        Group C: a VERY small group of geeks like me who read about economics. They already believe that most economics is very poor. You don’t need to persuade them. They are the only non-economists who will read your blog, but what they want is your ideas on how to improve economics, not endless criticism of others. And your ideas need to be more substantial than the views that the non-economists already hold. Otherwise, you are not adding anything.

        Blair: “As scientists, though, a major part of our job is arguing against theories we feel are wrong. That’s the only way things will change”

        Economists are not scientists. “Scientist” is a mark of merit allocated by other people who recognise that merit. It is not a title you assume for yourself.

        For example, I first came across the use of mathematical forecasting in the economy in the 1980s when I worked for one of the oil majors. The oil majors have mathematical forecasting models that are so successful at forecasting demand that you can almost always refuel your car just by turning up at a service station whenever you feel like it. These mathematical models are successful and fulfil a useful economic purpose. However, no-one thinks they represent “science”.

        Meanwhile, many academic economists seem to imagine that the ability to refuel your car whenever you feel like it is down to the supply and demand fairies. They don’t often even acknowledge the existence of the useful mathematical models in business. However, when an academic economist produces a mathematical model, it is supposed to represent “science” even when it doesn’t have any useful purpose. That’s not credible.

        You seem like a smart guy and you write well. However, our social systems are built on what I might call heterogeneous merit / accountability. In other words, you are only as good as how you are viewed by other people. This discipline applies to all businesses that want to win or retain customers; all lawyers who want to persuade juries; all politicians who want to be elected; and all scientists who want to be believed.

        I think that there is an idea in economics that supply creates its own demand (Say’s Law?). That is not correct. It is not enough to supply heterodox economic insights unless you can identify a demand for those insights. You (heterodox economists – not just Blair) have not done that. If you were a business, you would have gone bust years ago.

        Like

      • Hi Jamie,

        I agree with your split of people. Very few people care about the particulars of economic theory. But you could say the same for any branch of science.

        That being said, I don’t subscribe to your idea that heterodox economists must create demand for their ideas. First and foremost, I consider science to be the pursuit of truth. Whether that truth is useful (and in demand) or not is a different matter.

        There’s plenty of examples of true theories that were not useful, at least immediately. General relativity, for instance, had no obvious applications beyond astronomy for many years. But then when GPS was developed, general relativity was absolutely essential because Earth’s gravity slows time on Earth enough that the positioning systems would not work if this isn’t corrected for.

        Now, I’m all for communicating scientific ideas to the public. That’s why I have this blog. But the truth is, I don’t care if there is a demand for my research. I care about finding out the way the world works.

        Like

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s