Long on Rhetoric, Short on Argument

DeLong on “Adam’s Fallacy”

Dipankar Basu

Imagine for a moment that you are a student of mathematics; imagine further that you have just proved an important and non-trivial theorem. Having established the theorem, you then try to illustrate the result through several examples. This is not very uncommon in mathematics, as anybody working in that discipline will tell you. Now imagine someone reading through your work and discovering some mistake in one of your examples. Having made this brilliant discovery, this person then proceeds to pompously announce to the world that your theorem is false! That, in a nutshell, is my impression of macro-economist, economic historian and commentator J. Bradford DeLong’s comments on Duncan Foley’s recent book “Adam’s Fallacy: A Guide to Economic Theology”.(1) As it turns out, even that impression is only half correct; the example, in this case, seems to have no mistakes and certainly not the mistakes that DeLong makes his case on. As I will argue below, most of DeLong’s criticism of the example are baseless; some of them rest on misreadings, and some on what I call “imputation”; some of his claims are supported by quoting passages out of context or quoting them only partially.

What is Adam’s Fallacy?

To put DeLong’s comments in proper perspective it is necessary to first briefly talk about the main argument in Foley’s book. “Adam’s Fallacy” is at the same time an extremely erudite exposition of classical political economy and a sustained critical engagement with it. The book is organized around the notion of “Adam’s Fallacy” (hence the title of the book), a fallacy situated at the very foundation of political economy, a fallacy moreover that has been carried over right through to the present times. Foley’s narrative masterfully depicts the play of this fallacy in the works of all the great (political) economists of the last two centuries.

So what is Adam’s Fallacy? It is the claim, according to Foley, that the pursuit of self-interest, which is morally problematic in most human interactions, is unambiguously socially beneficial in the context of competitive market interactions.(2) This claim, which consolidated itself in the writings of Adam Smith, has been passed down from generation to generation in various forms and various guises, and has been accepted and used, according to Foley, by economists of virtually all political persuasions.

But why is this claim a fallacy? It is a fallacy in three senses. First, it is a logical fallacy because neither Adam Smith nor any of his followers who use this claim – often implicitly – have ever managed to prove it rigorously and robustly; at best it remains an unproved assertion. Second, it is a moral fallacy because it “urges us to accept direct and concrete evil in order that indirect and abstract good may come of it”. Third, it is a psychological fallacy because it leads us to deny the iniquitous manner in which costs and benefits are distributed in society under capitalism.

Having laid out the fallacy in the first few pages of the book, Foley then goes on to illustrate how it is active in the thoughts of almost all the great figures in the history of political economy. Discussing the work of each of these great thinkers, right from Adam Smith to John Maynard Keynes, Foley gives several examples to show Adam’s Fallacy in operation. For instance, while discussing Adam Smith’s “Wealth of Nations”, Foley refers to the use of Say’s Law in the context of technological unemployment as an example of Adam’s Fallacy in operation. While discussing Malthus and Ricardo, Foley shows how their attitude towards poverty is an example of Adam’s Fallacy in action. While discussing Marx’s critical political economy, Foley indicates how Adam’s Fallacy seduces even this “severest” critic of capitalism.  One could mention the other thinkers too that Foley discusses, but I think this much will suffice for the purposes of this article.

It is interesting to note, first of all, that DeLong has nothing to say about the general arguments establishing Adam’s Fallacy, arguments which are devastating in their simplicity. Neither has DeLong anything to say about the masterful treatment of all the great issues of political economy and the demonstration of their continuing relevance today. Instead of looking at the “proof” of the “theorem”, DeLong focuses his attention on one particular example in the discussion about one of the many thinkers that Foley talks about. This particular example of the operation of Adam’s Fallacy, the cause of evident discomfiture in DeLong, is about technological unemployment. So let me briefly summarize the issues involved in this example before taking up DeLong’s wayward comments on it.

The Example of Technological Unemployment

Adam Smith’s vision of the virtuous spiral of economic development rests on highlighting the link between the division of labour, labour productivity and the extent of the market; this link, when and where it can take hold, operates as a positive feedback loop (to use Foley’s terminology). Widening extent of the market (i.e., growth in effective demand) supports an increasing division of labour, which increases the productivity of labour, leading to falling prices and rising real incomes. Growth in real income increases the extent of the market in turn, completing the virtuous spiral. This story of economic growth fueled by technological change (merely another name for the increasing division of labour) has, according to Foley, one blind spot: unemployment created by rapid productivity growth. And talking about this blind spot is what draws DeLong’s misconceived ire.

DeLong’s Claims About the Example

DeLong makes his argument in the form of six “follies” of Foley that he claims to have discovered in this example. Let me mention them as DeLong states them before taking up each in greater detail.

(F1): “Foley’s Folly #1: The assertion that the costs of higher labor productivity are “direct, concrete” while the benefits of higher labor productivity are “indirect, abstract.” Lower prices that give consumers higher real incomes are exactly as concrete and as direct as are income losses from unemployment.”

(F2): “Foley’s Folly #2: The claim that technological unemployment is the rule…”

(F3): “Foley’s Folly #3: “The assertion that a belief in the theoretical truth of Say’s Law–in the efficiency of financial markets–is necessary to support the claim that the market system is for the general good”.

(F4): “Foley’s Folly #4: Duncan Foley’s calling a belief in efficient financial institutions “Adam [Smith]’s Fallacy in action.”

(F5): “Foley’s Folly #5: It turns out that Duncan Foley doesn’t believe that technological unemployment is the rule.”

(F6): “Foley’s Folly #6. I don’t think I understand high Western European unemployment. But I do not think that the first-order cause of high Western European unemployment was rapid labor productivity growth.”

Rebuttal of DeLong’s Claims

If you have not already noticed it then let me point this out: look once again at F2 and compare it to F5; they are diametrically opposed claims! DeLong would have us believe that within a few pages, Foley has reversed his position on technological unemployment dramatically, claiming first that it is the rule and then claiming that it is not. Faced with this rather strong claim by DeLong I started looking for the passage in the book where Foley asserts that “technological unemployment is the rule”; I am still searching. Probably DeLong would be so kind as to point to this passage that Foley seems to have forgotten to insert in his book. I will take up the issue of “technological unemployment” later, but here it is important to notice the trick that DeLong has used in the course of his argumentation: impute to your opponent something that (s)he hasn’t stated and then in a flourish of glory make the discovery that (s)he has in fact stated exactly the opposite. This way one can “generate” numerous “contradictions” in the text under scrutiny!

F3 and F4 are really not worth spending much time on; they are silly misreadings or intentional misleadings. Foley does not claim, as mentioned in F3, that “a belief in the theoretical truth of Say’s Law–in the efficiency of financial markets-is necessary to support the claim that the market system is for the general good”; it might very well not be necessary. But when push comes to shove, pro-market arguments almost always take the support of some version of Say’s Law (the claim that supply creates its own demand); that is all Foley claims. Examples abound: benefits of free trade, technological progress, privatization policies, so-called development policies like building dams and highways by displacing people without proper compensation, etc.

F4 is even more surprising because Foley refers to the whole argument about technological unemployment as an example of Adam’s Fallacy in action and not to a belief in efficient financial institutions (page 11), as F4 claims. Foley points out that proponents of Say’s Law (who argue that technological unemployment is impossible) seem to implicitly assume the existence of efficient financial institutions; but that is only one, even though crucial, piece of the argument. That by itself is not Adam’s Fallacy in action.

In F6, DeLong tries to convey the impression that talking about European unemployment and technical change in the same breath is outrageous. Here he seems to merely display his ignorance about a large and growing literature on “skill-biased technical change” as a cause of unemployment and inequality. This literature (both theoretical and empirical) attempts to explain, at the same time, unemployment in Europe and increasing wage inequality in the US as the result of skill-biased technological change in different institutional contexts of social support mechanisms.(3) One might very well disagree with the conclusions of this literature; but to behave as if linking European unemployment to technical change is outlandish is to willfully deny the existence of this literature altogether.

In F1, DeLong displays his predilection for playing on words. He objects to Foley’s claim that the costs of productivity growth are “direct and concrete” while the benefits are “indirect and abstract”. He feels, instead that “Lower prices that give consumers higher real incomes are exactly as concrete and as direct as are income losses from unemployment”. Instead of playing on words like “direct” and “indirect”, I think we can understand Foley’s claim in a very simple manner: costs of capitalist development are largely borne by a group which does not enjoy the lion’s share of the benefits of that development. That, in my opinion, is the sense in which Foley juxtaposes the directness of costs with the indirectness of benefits. Here is what Foley has to say on this matter: “The immediate effect of increases in labour productivity is to impose costs (unemployment) on a group (workers) who are in a weak position to protect themselves from these costs” (page 11). That this has always happened and is still happening is no great mystery for anyone who cares to so much as glance at the periphery of the capitalist world.

Technological Unemployment Once Again

This, of course, brings me to the main claim that DeLong wants so desperately to defend: technological unemployment is impossible. There are three apparently plausible arguments that DeLong makes in his rather rude piece. One, he points to some historical examples where productivity growth has been accompanied by rising rather than falling demand for labour. These examples do not constitute an argument against thepossibility of technological unemployment for two reasons. First, they are mere examples (as DeLong also realizes), and one can come up with other examples where the opposite has happened. An immediate example is the case of India after the onset of the so-called liberalization of the economy, starting from the mid-1980s: increasing productivity has been accompanied by the falling growth elasticity of demand for labour. For every unit increase in the growth rate of GDP, the percentage increase in the demand for labour has fallen.(4) Second, there have been other factors, like government policies, that have been important in shoring up the demand for labour during periods of rapid productivity growth. For instance, in the case of the “steam-machinery-cotton complex in Manchester at the start of the industrial revolution”, the accumulation of capital in England rested on the policy-driven destruction of artisanal production in India, high tariffs for cotton products coming into England and a safe, protected market for Manchester’s products in India under the stern gaze of the colonial government. So, in this particular case at least, it seems that a steady, sure and growing market was what contributed to the rising demand for labour and dampened the possible effects of technological unemployment.

And this brings us to DeLong’s second point: if the (income) elasticity of demand (5) is high then demand for labour might increase along with rising productivity, states DeLong. There is hardly any reason to deny that. But the question remains: what if the elasticity of demand is low? If DeLong wants us to believe that he has made a novel point, he is mistaken. For Foley already takes account of this possibility when he states that if market demand can proceed at a faster pace than productivity growth, labour demand might grow. The point is that there is no necessity that this will always happen. Either DeLong has to establish theoretically that the income elasticity of demand for products of industries under going rapid productivity growth will always be high, or one will have to establish the facts empirically in each case. Foley merely seems to advocate a cautionary approach as opposed to baseless optimism of market enthusiasts.

The third point that DeLong makes in this context is to point to an apparent contradiction in Foley’s stance on Say’s Law as it plays out in the argument on “technological unemployment”. Between F2 and F5, DeLong would have us believe, Foley changes his position on “technological unemployment” completely. Here is what DeLong writes:

“Still worse, in my view, is Foley’s Folly #5: It turns out that Duncan Foley doesn’t believe that technological unemployment is the rule. Immediately after this song-and-dance about the “direct, concrete” costs that labor productivity growth generates via increased unemployment, costs that “ordinary moral reasoning would regard… as a bad thing,” Foley writes:

Over long periods of time… something like Say’s Law does operate… there is no long-term drift towards constantly increasing unemployment as a result of technological change…

Unless Foley wants to maintain–which I don’t think he does–that in the absence of technological progress we would have steadily falling unemployment, what he is saying here is that unemployment is not higher as a result of technological progress. There is no cost to charge against the benefit of higher productivity. Technological unemployment is a non-issue. And I cannot understand why Foley raises it–let alone introduces technological unemployment as his first example in his book of the “real costs of capitalist development” about which Adam Smith is in “wholesale denial.”

This is an example of misreading supported by partial quotations from the text. Foley adopts a more nuanced stance towards Say’s Law than DeLong can give him credit for. He agrees that something akin to Say’s Law seems to be in operation in the long run but stresses that it is a folly to argue for its presence in the short-run. And it is precisely in this distinction that the costs of “technological unemployment” resides. I can do no better than to let Foley speak:

Say’s Law comes up again and again in the story, and it will help to keep two points in mind. Over longer periods of time, it appears that something like Say’s Law does operate: at least there is no long-term drift toward constantly increasing unemployment as a result of technological change and rising labor productivity. On the other hand, over shorter periods, the absorption of technologically unemployed workers into new jobs can be quite slow, creating real social, economic and political problems…one important issue about Say’s Law is what time scale we are looking at…”

The important issue is the time scale of the analysis. Admitting that there is no upward drift in the unemployment time series is not the same as admitting that “there is no cost to charge against the benefit of higher productivity” (as DeLong claims) because the costs operate in the short run, and the costs are imposed on those who are hardly ever able to partake of the benefits of that technological progress even in the long run to the degree that they had to bear the costs.

The deeper problem with DeLong’s arguments is that he seems blind to some simple facts that were pointed out a few years ago by Baumol and Wolff (1998).(6) In periods of rapid technological growth, demand for new skills grow while demand for several classes of existing skills diminish. So, even when productivity growth leads to an overall increase in the demand for labour (the best possible scenario for DeLong), rising unemployment can be observed for several reasons. First, firms will, on average, need to shut down for longer periods for retooling; this will lead to increases in layoffs as well as increases in the average duration of unemployment spells. Second, firms will decrease hiring workers with some kinds of skills and increase their employment of workers with other kinds of skills. This will also lead to some layoffs. Third, firms will not re-train those workers for the newly required skills for whom the costs of training are the highest, the older workers and the younger, less-educated workers. Hence, workers who face the greatest difficulties in making themselves suitable for the new situation might gradually slide into the pool of long-term unemployed. Moreover, these effects will be in operation over all the phases of the business cycle. In their study, Baumol and Wolff (1998) find that the average duration of unemployment spells have doubled since the 1940s. And all the problems associated with joblessness – suicide, illness, divorce, broken families and destroyed childhoods, criminal activity, substance abuse – increase with the duration of joblessness.

As far as I can understand, Foley has nowhere asserted that technological unemployment is the rule; he has merely called our attention to its very real possibility. Denying this possibility a priori, which is the substance of DeLong’s argument, is to precisely commit Adam’s Fallacy all over again. Therefore, DeLong, by arguing against the possibilities of technological unemployment, has most spectacularly vindicated Foley’s claim that Adam’s Fallacy is very much alive and kicking in the economics profession in the twenty-first century.

The argument against DeLong’s evasion is actually very similar to what we have already encountered: the costs of technological progress are disproportionately borne by those who never share in the major part of that benefit. To my mind, Adam’s Fallacy, in most cases, boils down to the following: failure to recognize the fact that the institutions of capitalism make it impossible for the costs and benefits of technological progress and economic development to be shared equitably among the members of society. Private property and markets – the defining institutions of capitalism – can only distribute these costs and benefits inequitably, and that is precisely what Adam’s Fallacy tries to blind us to.


(1) This can be accessed at: http://DeLong.typepad.com/sdj/2006/09/the_childish_ba.html

(2) There is another, deeper sense in which Foley uses this term: the fallacious idea that there is a separate economic sphere of life governed by its specific laws.

(3) Weiss, Matthias and Garloff, Alfred, “Skill Biased Technological Change and Endogenous Benefits: The Dynamics of Unemployment and Wage Inequality” (2005). ZEW Discussion Paper No. 05-79. Available at SSRN: http://ssrn.com/abstract=862064

(4) Bhattacharya, B. B., and S. Sakthivel, “Economic Reforms and Jobless Growth in India in the 1990s”, Working Paper, Institute of Economic Growth, New Delhi, India. Available at: (http://ieg.nic.in/worksakthi245.pdf)

(5) The income elasticity of demand measures the responsiveness of demand to changes in income; formally, it is defined as the ratio of the percentage change in demand to the percentage change of income. Economists also work with other elasticities, like the price elasticity of demand or the cross-elasticity of demand, which are defined in a similar manner.

(6) Baumol, Willim J., and Edward N. Wolff, “Side Effects of Progress: How Technological Change Increases the Duration of Unemployment” (1998). Public Policy Brief No. 41, The Jerome Levy Economics Institute of the Bard College.

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 )

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

%d bloggers like this: