"Marginal Productivity Theory and the Mainstream"

Thomas Palley with more on the acceptance of heterodox views within the economics profession. Thomas discusses the neoclassical theory of income distribution. Since this theory was first published by J. B. Clark over a century ago, it has been controversial. J.M. Clark, his son and also a prominent economist, wrote that his father's ideas on marginal productivity and the distribution of income are ethical statements "oriented at Marx, and are best construed as an earnest, and not meticulously qualified, rebuttal of Marxian exploitation theory." Clark believed that each factor of production would receive a return just equal to its contribution to society (the value of its marginal product) and hence the distribution of income was ethically correct.

There have been many, many objections to the normative conclusions drawn by Clark from his marginal product theory. To name a few, it requires perfect competition, it rewards factors, not individuals (owning capital and land gives the owners income, but the income is for the contribution of the factors, not for the contributions of individuals receiving the income), there is no meaningful way to separate the contributions of factors to total product (when crops grow, was it the hoe used to weed the plot of land, or the person operating it?)

Marginal productivity theory gives an explanation for why income is distributed in a particular way. The question is whether it provides an adequate theory for understanding the flow of income to the factors of production, and I think it does provide such a basis. So it helps us to understand why income is distributed as it is, but arguing about whether the result is equitable or not is fruitless since it will depend upon the definition of equity used to evaluate the outcome, and the definition can differ across individuals.

My own view is that we don't need, at least not yet, to discard the standard theoretical model. Instead, and there has been a lot of work in this direction, we should first be sure that incorporating market and political power relationships into the standard structures won't explain what we observe in the world. Dani Rodrik states this as "I think the best antidote against the blind spots is neoclassical economics itself." When someone invokes marginal productivity theory's ethical implications, as they do, say, when they argue that CEO pay is justified because it reflects CEO's contributions to the firm's output (and hence to society more generally), they are implicitly assuming that the assumptions of the perfectly competitive market hold when it's doubtful that they do. Markets can fail in a variety of ways and the neoclassical model of pure competition can serve as a useful benchmark for understanding how departures from an idealized structure will resolve themselves as people interact in the marketplace. Too many people argue from the standpoint of the perfectly competitive model when it simply doesn't apply and I would be happy, at least as a start, if the discussions of these issues would do a better job at recognizing that the competitive model is an idealized benchmark to evaluate actual markets, but not an outcome we should necessarily expect in the real world. Here's Thomas [Note: Thomas has updated this post - see the new edition that is posted below]:

Marginal Productivity Theory and the Mainstream, by Thomas I. Palley: Last week as part of a discussion on the state of orthodox economics hosted by TPM Café, I posted an article excavating the microeconomic foundations of neo-classical economics. In that article I wrote:

There is one place that even orthodox lefties dare not go. That untouchable place is marginal product theory of income distribution, which basically says that competitive markets ensure that people are paid their contribution to production. This theory provides both a justification and an explanation of income distribution.

Dani Rodrik has challenged this claim in a blog titled “Do heterodox lefties have a better grasp on reality than neo-classical lefties?” As one would expect anything Dani writes is thoughtful and constructive. Let me try and develop some further insights.

Dani’s counter is that the mainstream empirical literature on pay determination is full of references to the impact of institutions, particularly unions. He also points to his own paper titled “Democracies Pay Higher Wages” that was published in the Quarterly Journal of Economics (QJE). That paper shows that bargaining environment, and especially the degree to which the political system is democratic, exerts a significant impact on labor’s share of the surplus. Democracy can raise wages by as much as fifty percent, controlling for productivity. I agree with this, although in a separate paper titled “Democracy, Labor Standards and Wages” I have argued democracies work their wage effects through better labor standards that they promote.

On the basis of this Dani concludes that (1) neo-classical economics is not averse to going beyond marginal productivity theory, and (2) the fact his paper was published in the mainstream establishment QJE suggests orthodox economics is not closed.

Dani is absolutely right about the empirical literature on pay, which is broad and multi-faceted. The question is what is the significance of that literature for marginal productivity theory and the economics profession?

When it comes to unions, it is well documented that unions raise wages. Union supporters argue that (especially in highly unionized economies) unions do so with no adverse employment effects. Marginal productivity theorists argue they do so at the cost of lower employment.

What about democracy? I suggest democracy raises wages by altering the division of the cake, but there is no adverse employment effect. Furthermore, by raising wages and improving income distribution, democracy strengthens consumer demand and may increase employment through a Keynesian channel. This is a non-marginal productivity theory argument that cannot be rendered consistent with marginal productivity theory.

What does neo-classical marginal productivity theory imply about democracy? Holding productivity constant, democracy would tend to lower employment because it raises wages. However, there is a potential escape hatch (as there always is with neo-classical economics) in that one could argue democracy lowers the monopsony (buyer monopoly) power of employers, thereby raising both wages and employment.

The bottom line is there are two stories about the wage effects of democracy (and unions). My view is the institutionalist – Keynesian story is more plausible. I am not sure what Dani’s view is, as he seems to support both. However, that seems theoretically problematic.

When it comes to marginal productivity theory of income distribution you are either in or out with regard to the labor demand schedule, which tightly determines the relationship between wages and employment as a technological relationship. The neo-classical labor demand schedule is the essence of neo-classical economics, and most of its analysis collapses without it.

The bigger story is that the empirical data settle nothing and can be interpreted to be consistent with either theory. That suggests both should be taught with equal prominence in all economics departments, including top departments. Yet they are not, and only the neo-classical marginal productivity theory is taught as part of the core curriculum, while heterodox theory is essentially suppressed.

Recently MIT economists Peter Temin and Frank Levy have published a paper about the role of institutions in explaining inequality in 20th century America. Their paper is welcome – and (to be self-promoting) expands analytical themes developed in my 1998 book, Plenty of Nothing: The Downsizing of the American Dream and the Case for Structural Keynesianism. The engagement of these economists may be a sign that the economics profession’s thinking about income distribution is headed for change. If that is so, neo-classical economics will be in serious trouble because marginal productivity theory figures critically in its macroeconomics (both new classical and new Keynesian), its microeconomics, and it approach to trade and globalization.

 

Lastly, Dani argues the publication of his paper in the QJE is evidence of mainstream openness. I am delighted the QJE published his paper because it is a leading journal, which means the paper got wide circulation. However, the paper was issued earlier by the National Bureau of Economic Research and Dani is a Harvard professor. Both of those facts matter. The scuttlebutt is that the QJE is the Harvard/MIT working papers series, as evidenced by the extraordinarily large proportion of articles accepted from those faculties. This is just another example of the sociology of economics that my TPM Café discussion also emphasized.

Here's the second edition:

Marginal Productivity Theory and the Mainstream  - 2^nd Edition, by Thomas I. Palley: [Author note: I am re-posting this article because I think the original did not hit its target cleanly. I have left the original posting on my website so that readers can judge for themselves. The edits to paragraph 11 are especially important]

Last week as part of a discussion on the state of orthodox economics hosted by TPM Café, I posted an  article   excavating the microeconomic foundations of neo-classical economics. In that article I wrote:

There is one place that even orthodox lefties dare not go. That untouchable place is marginal product theory of income distribution, which basically says that competitive markets ensure that people are paid their contribution to production. This theory provides both a justification and an explanation of income distribution.

Dani Rodrik has challenged this claim in a blog titled  “Do heterodox lefties have a better grasp on reality than neo-classical lefties?”   As one would expect anything Dani writes is thoughtful and constructive. Let me try and develop some further insights.

Dani’s counter is that the mainstream empirical literature on pay determination is full of references to the impact of institutions, particularly unions. He also points to his own paper titled  “Democracies Pay Higher Wages”   that was published in the Quarterly Journal of Economics (QJE). That paper shows that bargaining environment, and especially the degree to which the political system is democratic, exerts a significant impact on labor’s share of the surplus. Democracy can raise wages by as much as fifty percent, controlling for productivity. I agree with this, although in a separate paper titled  “Democracy, Labor Standards and Wages”   I have argued democracies work their wage effects through better labor standards that they promote.

On the basis of this Dani concludes that (1) neo-classical economics is not averse to going beyond marginal productivity theory, and (2) the fact his paper was published in the mainstream establishment QJE suggests orthodox economics is not closed.

Dani is absolutely right about the empirical literature on pay, which is broad and multi-faceted. The question is what is the relation of that literature to marginal product theory and does it move beyond marginal product theory? I suggest not.

When it comes to unions, it is well documented that unions raise wages. Union supporters argue that (especially in highly unionized economies) unions do so with no adverse employment effects. Neo-classicals interpret unions through a marginal product labor demand lens, and generally argue they raise wages at the cost of lower employment. However, it is also the case that if employers have monopsony (buyer monopoly) power unions can also raise both wages and employment in a marginal product framework.

What about democracy? I suggest democracy raises wages by altering the division of the cake, but there is no adverse employment effect. Furthermore, by raising wages and improving income distribution, democracy strengthens consumer demand and may increase employment through a Keynesian channel. This is an argument based on a non-marginal product approach to income distribution.

What does neo-classical marginal productivity theory imply about democracy? Holding productivity constant, democracy would tend to lower employment because it raises wages. However, as with unions one could argue democracy lowers the monopsony power of employers, thereby raising both wages and employment. Under this neo-classical interpretation, democracy serves to move the economy closer to the idealized state of perfect competition in which the idealized version of marginal productivity theory of income distribution holds.

The bottom line is there are two stories about the wage effects of democracy (and unions) - one rooted in marginal product theory, the other not. My view is the institutionalist – Keynesian story is more plausible. I am not sure what Dani’s view is, as he seems to support both. However, that seems theoretically problematic.

When it comes to the neo-classical theory of income distribution you are either in or out with regard to the concept of marginal product. Under conditions of perfect competition, the marginal product of labor is the labor demand schedule, which tightly determines the relationship between wages and employment as a technological relationship. Under conditions other than perfect competition, the marginal product of labor remains ever-present and provides the reference point for determination of wages. However, if marginal product is an incoherent or unusable concept most of neo-classical economics (including its approach to income distribution) disintegrates: hence, the unwillingness to question marginal product analysis.   

The bigger story is that the empirical data settle nothing and can be interpreted to be consistent with either theory. That suggests both should be taught with equal prominence in all economics departments, including top departments. Yet they are not. Instead, only the neo-classical marginal productivity theory is taught as part of the core curriculum, the concept of marginal product is never questioned, and heterodox theory is essentially suppressed.

Recently MIT economists Peter Temin and Frank Levy have published a paper about the role of institutions in explaining inequality in 20th century America. Their paper is welcome – and (to be self-promoting) expands analytical themes developed in my 1998 book,  Plenty of Nothing: The Downsizing of the American Dream   and the Case for Structural Keynesianism. The engagement of these economists may be a sign that the economics profession’s thinking about income distribution is headed for change. If that is so, neo-classical economics will be in serious trouble because marginal productivity theory figures critically in its macroeconomics (both new classical and new Keynesian), its microeconomics, and it approach to trade and globalization.

Lastly, Dani argues the publication of his paper in the QJE is evidence of mainstream openness. I am delighted the QJE published his paper because it is a leading journal, which means the paper got wide circulation. However, the paper was issued earlier by the National Bureau of Economic Research and Dani is a Harvard professor. Both of those facts matter. The scuttlebutt is that the QJE is the Harvard/MIT working papers series, as evidenced by the extraordinarily large proportion of articles accepted from those faculties. This is just another example of the sociology of economics that my  TPM Café discussion   also emphasized. 

Comments

says...

(owning capital and land gives the owners income, but the income is for the contribution of the factors, not for the contributions of individuals receiving the income)

Strictly speaking, this is only true in the case of land.

The return to capital goods is shared between the laborers who originally created the good as a product of their own labor, and the capitalists who subsequently abstained from liquidating the good.

there is no meaningful way to separate the contributions of factors to total product (when crops grow, was it the hoe used to weed the plot of land, or the person operating it?)

This is a misconception. Hoes and other capital goods are not valued for their contribution to productivity, rather for the labor and abstinence involved in their production.

However, there is a potential escape hatch (as there always is with neo-classical economics) in that one could argue democracy lowers the monopsony (buyer monopoly) power of employers, thereby raising both wages and employment.

This paragraph is quite misleading, because there's no reason to limiting analysis to monopsony. If democracy lowers the overall amount of parasitic rent extraction (such as by lowering monopsony power), wages and/or the interest of capital will rise. How does this disprove marginal productivity theory?

says...

A critical view of Clark's marginal productivity theory, from the American Journal of Economics and Sociology.

[Bruce Wilder]

This topic and the sloppy way economists typically approach it is one of my hobby horses. I promise not to produce a lengthy dissertation, but I would like to point out some serious conceptual limitations.

First, is marginal productivity an "adequate" theory? No.

An adequate theory identifies the necessary and sufficient elements of a system and the functional relationships among those elements in that system. I am sure that could be said better. But, it is kind of basic philosophy of science to say that the point of analysis is to identify the necessary and sufficient elements. Not having all the elements is like trying to do solid geometry with the conceptual apparatus of plane geometry.

Marginal productivity, as a theory of income distribution, fails that test. It is the plane geometry of a 3-D world.

Plane geometry for a 3-D world can be useful. It is a projection, and if you assume a fixed point of view, it can work, just as a flat computer screen can be a window onto a pseudo 3-D computer game.

Marginal productivity analysis depends on output being a function of inputs -- and guess what? output is not a function of inputs. (A particular value for output does not have a unique correspondence with particular values for inputs; in practice the same inputs can produce lots of different outputs, and different inputs can produce the same outputs. It doesn't take any more than a few seconds of thought to confirm this is true.) The usual dodge is to assert that "maximum" output is a function of inputs. This dodge only works if "maximum" has some definition or meaning, other than "I don't want to actually think about the relationship of output and input."

In another context, economists would readily recognize that labor compensation is often contingent, in some way, on behavior, and like compensation for capital, has some implicit or explicit risk-conditionals built-in.

And, anyone in any actual firm, would recognize that the firm's production process is "managed" in some way, with output contingent. Compensation is part of the managing. (It really doesn't make much sense to assume maximum output, in such a context.)

Am I saying that the richer and more complicated world of risk and uncertainty and managed production can never be usefully reduced to the world of outputs as a function of inputs in marginal productivity analysis? No, I wouldn't go quite that far.

Risk and uncertainty and management are necessary elements. In an important sense, they can not eliminated from a proper analysis. To do so, is to create a projection. In some circumstances, a projection can be a useful picture.

But, it remains fundamentally true that the distribution of income is inextricably linked to the distribution of risk and the management of behavior. This is, or should be, recognized as a fundamental insight of economics. Risk, uncertainty, and management of behavior are necessary elements, which can never be disregarded.

Using marginal productivity to "justify" a CEO's compensation doesn't just involve questionable ethical reasoning. It is a kind of analytical malpractice, to abstract away from the necessary elements of risk and management, in this context.

A financial economist would not think of analyzing the returns to equity without reference to leverage. But, somehow it is OK, analytically, to treat labor as if it were various ingredients specified in a recipe.

[2slugbaits]

Bruce,

I'm not quite following you so I won't guarantee that my comments are fully responsive. If we look at factor productivity very little growth is due to capital. On the other hand, a lot of the growth of the 90s was due to technology, which would include management control over the production process, smarter workforce, etc. What seems to be happening is that a lot of the profits due to technology growth go to management types (which is appropriate) and to owners of capital (which is less defensible).

Also, while it's true that inputs can be applied to different outputs, it is also true that once an input is committed to particular output it is essentially unavailable for other uses. So I guess I don't quite understand your point.

[h.e.]

It's disappointing to see a discussion on the validity of the neoclassical theory of distribution without reference to the Cambridge Controversies. The theory has problems that are more fundamental than the potential need for alternative market structures and the like. There is a reason why macroeconomists carefully refer to the "one good" model.

[Bruce Wilder]

If I were an automotive engineer comparing the performance of two engines, I would use a theoretical analysis of thermodynamics to help sort out differences. This engine is consuming more fuel, that engine is wasting more heat -- stuff like that.

Same kind of thing, if I were an aeronautical engineer, trying to understand differences in the performance of two airfoils. I'd use a system of equations that analyze the forces or pressures, which add up to aerodynamic force. (They're called Euler's equations, I think).

On a deep level, analysis lets us identify what relates to what. You can only observe some of what is going on, and you have to infer the rest. Sound analysis lets you build a sound model, which lets you correctly identify and relate important observables and disregard the irrelevant. You're not likely to think the color of the airfoil is important, for example.

I know a lot of people like to think you can somehow directly observe "cause and effect", but it's not true, that's not something you can observe.

So, when people are comparing income distribution in 1990 with income distribution in 2000, say, their ability to figure things out and attribute "causality" (if that's what you want to call it) depends on the quality of theoretical analysis they bring to the problem.

An aeronautical engineer, who considered pressure and velocity, but forgot the density and viscosity of the fluid thru which his airfoil was moving would be doing less than stellar work. He'd get the wrong answer.

I'm saying that using marginal productivity analysis is like the work of that forgetful engineer. It leaves out important, maybe critical factors.

I think of marginal productivity analysis as being kind of the "big shovel" theory of income distribution. That is, it implies that having better technology and more capital tends to raise the marginal productivity of labor and therefore the wages (i.e. income) of labor. A guy operating a back hoe makes more money than a guy with a hand shovel, because he can and does dig a lot more hole -- hence, big shovel theory. That analysis establishes a relationship, which we can look for in the data on income distribution: is there more capital per worker? Is the capital (the worker's tools, education) more productive, more productivity-enhancing because of changes in technology?

As far as it goes, I am fine with that analysis, but I don't think it covers all the necessary elements affecting income distribution we know about.

And, it leaves me with some big questions about what is meant by "technology". The theory of production, from which the marginal productivity analysis derives just assumes, for no other reason, really, other than analytical convenience, that output is at a maximum in some meaningful sense. That makes maximum output a function of inputs, and reduces the whole problem of efficient production to a problem of allocative efficiency -- the firm chooses its input mix on the basis of the market prices of factors and that's it. Essentially, all of the problems of management and technology are simply assumed to be solved by the simple expedient of assuming that output is a function of inputs, because output is at a maximum.

Think about that: all the problems of management and technology are simply assumed to be have been solved.

I am not saying it is crazy, as an exercise in analysis, to make such an assumption. It lets the analyst focus on the problems of allocative efficiency, while setting aside potentially knotty problems of technical or administrative efficiency. Assuming that competitive pressure has brought output to a maximum at equilibrium is useful, conceptually, to gaining insight into what allocative efficiency means and requires.

But, to turn around, without relaxing that assumption and expanding the analysis to argue for some kind of technology-determinism in the distribution of income is just b.s.

Show me an analysis of production, where the problems of technology and management are not "solved" by assumption, where output is a function of technology and management and human behavior. Derive your analysis of income distribution from that theory of production. Then, we'll talk.

Economists, generally, at least when speaking to the general public, are very undisciplined when talking about income distribution issues. It tends to involve a lot of hand-waving, a lot ideological prejudice and sloppy thinking. There'll be pious hectoring on the value of education, and loose references to "skills".
To my thinking, there's way too little discussion about how changing the distribution of risk relates to changes in the distribution of income.

One thing a professional economist ought to grok is that the distribution of income and the distribution of risk are intimately related; a policy to increase the uninsured risk exposure of the poor and middle class will have the predictable consequence of redistributing income upward toward wealthier and more powerful people. That's a relationship I expect an economist to identify, and when it doesn't happen in a context where I think it ought to (say, in a discussion of private accounts and Social Security), I get suspicious and grumpy.

[Anarcho]

From the same paragraph. First:

"My own view is that we don't need, at least not yet, to discard the standard theoretical model." "

Then we have:

"When someone invokes marginal productivity theory's ethical implications . . . they are implicitly assuming that the assumptions of the perfectly competitive market hold when it's doubtful that they do.

So we should not discard the standard model even though it is "doubtful" (understatement!) that its assumptions hold in the real world? Sorry, I though economics was meant to be a science...

The only reason why marginal productivity theory survives is its political utility, namely the fact that it can be used to justify inequality and returns to capital. Which was why Clark invented it in the first place, to refute socialist/anarchist arguments that labour was exploited under capitalism.

Joan Robinson and Piero Sraffa had successfully debunked the theory in the 1950/60s (the Cambridge Capital Controversies) and their critique was admitted as valid by leading neo-classical economists! Yet this made no difference. To quote James K. Galbraith:

"Yet for psychological and political reasons rather than for logical and mathematical ones, the capital critique has not penetrated mainstream economics. It likely never will. Today only a handful of economists seem aware of it." ("The distribution of income", pp. 32-41, Richard P. F. Holt and Steven Pressman (eds.), A New Guide to Post Keynesian Economics, p. 34)

[Mark Thoma]

I actually got beyond the chapter on perfect competition.

You are equating "standard theoretical model" with pure competition. My definition is more encompassing - the standard model can handle relaxing the assumptions of pure competition.

And yes, I'm fully aware of Robinson and Sraffa, we all are. But thanks.

In any case, I think you missed my point. I'll assume it's unclear writing.

[ndd]

Prof. Thoma:
Your writing was clear enough to me. I am sure you are frustrated when laypeople with passionate opinions jump in with comments which appear ignorant, and I'm sure a couple of mine might qualify.
Nevertheless, I wish all economists who declaim on public policy issues were exposed to such feedback from laypeople, as you have done on this blog. If they were, they would understand the seething resentment towards so much of economic theory as it appears in the mainstream media, which appears to be social darwinism dressed up in math.
The right track/wrong track polls for the direction of the US are at record low readings. The public, or large swathes of it, no longer believes economic numbers. And twice this week, on very big issues, we find out the public's distrust may have been well placed: first with the admission that "core" CPI numbers are reducing signal, not noise. And via the Mark Mandel column you have posted today, we find out that GDP has been meaningfully and systematically overstated.
To the point you reply to above, it is simply a manifestation of the fact that the right-wing punditocracy and political establishment have successfully conflated econ 101 with real-world economics in the public's mind. Unfortunately, you are having some firsthand experience with the blowback -- but of course, even that is a teaching/learning experience.
Thank you again for all the effort you put into this blog.

[cm]

"Markets can fail in a variety of ways ..."

First of all, the concept of a market is an abstraction, usually with the connotation that they produce some kind of equilibrium or best/desirable outcome. Most claims about market performance and indeed operation require sufficient liquidity, and substitutability of the effects traded. It is highly doubtful that every back-room dealing situation fits that bill.

It may not be a matter of market failure, but that the imputation of a market mechanism is misplaced.

[Winslow R.]

bw wrote: "One thing a professional economist ought to grok is that the distribution of income and the distribution of risk are intimately related; a policy to increase the uninsured risk exposure of the poor and middle class will have the predictable consequence of redistributing income upward toward wealthier and more powerful people."

I'm not sure what qualifies as 'uninsured risk exposure' but I find the willingness to take on 'risk' as the path to higher income.

The willingness to spend 4 years and quite a bit of money for a college education is a risk.

The willingness to leverage a home 20 to 1 by our host was a risk.

I find that it is the opportunity to take on 'reasonable risk' that concentrates income. Those with the greatest access to 'reasonable risk' will have the largest returns.

Borrowing short at the interbank rate and lending long at the 10 year tsy is a 'reasonable risk' open only to a select few like Goldman Sachs.

While the list is long, the ability to access 'reasonable risk' is limited, concentrating income to those with access.

[Bruce Wilder]

Winslow R. "I find the willingness to take on 'risk' as the path to higher income."

Yes indeed, it commonly is.

Sunk cost investments are often critical elements in economic competition and growth, but recovery of such investments is never certain -- hence the risk.

Broad distribution of risk-bearing capacity in the economy is necessary to assure that such sunk-cost investments are efficiently made.

[reason]

Bruce Wilder

I think of marginal productivity analysis as being kind of the "big shovel" theory of income distribution. That is, it implies that having better technology and more capital tends to raise the marginal productivity of labor and therefore the wages (i.e. income) of labor. A guy operating a back hoe makes more money than a guy with a hand shovel, because he can and does dig a lot more hole -- hence, big shovel theory.

Um.. Bruce actually, that is not exactly what drives higher wages if you think about it. The key word in "marginal". If the next guy employed only has a small shovel - he sets the price of labour. And that really shows the problem. The more the labour market is disagregated, the more likely it is that in many of those individual labour markets, labour is in surplus and marginal productivity is low (WAY below the measured average productivity).

And then there is the problem with saturated markets (falling marginal revenue curves). The best labour market market conditions for workers are in periods of rapid growth, with lots of new opportunities and markets (and hence marginal revenue) growing. The problem is that situation tends to be unstable and in SOME markets produces major bottlenecks and inflation. It is hard to sustain consistant high levels of employment without inflation (as modern central banks do) without creating large sectors of the economy where labour has very low bargaining power.

There is no floor under wages (except minimum or award wages), where it is in excess supply REGARDLESS of average (i.e. measured) productivity.

[RueTheDay]

First of all, marginal productivity theory is not a "theory of distribution". It is AT MOST a theory of factor demand. It says nothing whatsoever about factor supply, which would be necessary for it to be a complete theory of distribution. Second, the reason I emphasize "at most" above is that marginal productivity theory only works under constant returns to scale; under increasing or decreasing returns to scale, the sum of the factor payments will not equal the total product.