Schumpeter on Marx and Tautological Social Scientific Explanation
Updated: Jul 18, 2020
The first part of Joseph Schumpeter’s Capitalism, Socialism, and Democracy is devoted to a critical engagement with the ideas of Karl Marx--first his theory of history and sociology, second his theory of economics, and third of the overarching Marxist system of social explanation which emerges from the marriage of these two ‘Karl Marxes’ as an overall theory of everything.
In that third sub-part, Schumpeter introduces a somewhat heterodox usage of the term ‘tautological,’ one that refers not to a formal logical relationship between statements, but to a fundamentally broken form of social scientific explanation, a form which obscures more than it illuminates because it utilizes spurious causal mechanisms to interpret readily observable social scientific fact. I intend to conceptualize the social scientific tautology as a general form of pseudo-mechanism, a broken, general form of causal mechanism which generally recur in the humanities and social sciences, leading, ultimately, to impoverished understandings of the underlying social phenomena. This essay is thus successful if it illuminates some ways in which intended causal interpretations of social phenomena can misfire.
Schumpeter’s purpose, largely through an interpretation of Marx’s theory of imperialism as a continuation of labor exploitation, is to show how the Marxian system is prone to using social scientific tautologies to provide a veneer of understanding to phenomena which Marxian analysis is, in fact, powerless to explain. In Marxian theory, imperialism is largely explained by two general features of economic theory derived from the domestic case: the tendency toward capital accumulation and class struggle. Accumulation is relevant because of the persistent tendency of capitalism to self-destruct: in the domestic case, the accumulation of capital into increasingly small numbers of hands and the decline in return to capital incessantly and increasingly threaten the profits on which the economic system is predicated. As the return on fixed capital diminishes, so too does the capacity to exploit domestic labor to derive profits.
Hence, imperialism: One way that capitalists can seek to temporarily escape from the self-destructive features of capitalism is to find new markets for exploitation. New labor markets in which surplus labor value has not already been exploited must be found to keep late-stage capitalist systems alive. This now-internationalized system of exploitation turns to colonialism once the ‘new’ market begins to offer organized resistance to its newfound system of capitalistic exploitation. We thus, apparently, have a causal explanation of colonial oppression which is derived from an analysis of domestic capitalism, something that at least looks like the ‘risky predictions’ of Popperian philosophy of science.
Now, note that the process of capitalistic imperialism is driven by the self-defeating logic of capitalism itself. It would be beside the point to here explore the mechanisms which, under the Marxian system, lead to the self-destruction of capitalism. However, it is precisely this self-destruction which prompts colonialist ventures, something of a way of allowing steam to escape a boiling pot before it can cause the pot to explode. There should thus be little incentive to engage in colonialism in a relatively early stage of capitalism, in which domestic markets still allow for capitalistic accumulation and exploitation with still-high rates of return of investment in fixed capital. On the other hand, more developed capitalist countries--those in which the pot is closest to exploding, lest some outlet for the steam to escape be found--should become increasingly desperate to find outlets for their colonialist ventures before it is too late. A high degree of capitalistic development should thus serve not only as an enabling condition for, but, instead, a causal explanation of participation in colonialism.
As Schumpeter notes, the Marxian theory of imperialism seems to explain a great deal of international politics, at least if one sticks to a superficial level of detail. Colonialism is a process in which more ‘advanced’ (in a capitalistic sense) polities come to invest in, and eventually politically dominate, less advanced polities. A pervasive feature is the use of ‘cheap’ colonial labor to increase profits from return to investment. Thus, even freedom-loving liberal (but capitalist!) states aggressively pursue both dominion over foreign markets, and, well into the twenty-first century, exploitation of cheap labor to enhance profits at home, in what would likely be stagnant economic conditions without the economic exploitation of foreign labor. Here’s where the Marxian theory appears to have some predictive power: the colonial market, too, will undertake capitalistic development, creating increasing friction between mother country and colony. Capital export to the colony will begin to decline. The colony, as it develops economically, will attempt to compete with the mother country in markets for manufactured goods, creating friction between the two countries. The mother country may undertake mercantilist policies toward the colony, and the resulting friction will lead to wars of independence and the like. Thus (Schumpter):
Colonial doors will eventually be closed to domestic capital which will no longer be able to flee from vanishing profits at home into richer pastures abroad. Lack of outlets, excess capacity, complete deadlock, in the end regular recurrence of national bankruptcies and other disasters--perhaps world wars from sheer capitalist despair--may confidently be anticipated. History is as simple as that.
The Marxian system appears to have predictive social scientific power after all! Large parts of international history fall into place as a mere consequence of the Marxian concepts of accumulation and class conflict, transported into the realm of international competition.
The precondition for a social scientific tautology here, it seems, is that the fit between the causal mechanism posited by Marxian analysis and the underlying phenomena is entirely superficial. It explains a few cases of a general phenomenon (i.e. imperialism) sort of well, at first glance. However, it fails to comport with other cases of what should be the general phenomenon, and, even where the fit appears to be good, the relationship does not weather careful scrutiny.
First of all, it may be true that some of the most aggressive colonizers of Marx’s era were the most advanced capitalist states. (I’m looking at you, Britain!) But a closer look at the underlying phenomena suggests that the fit is poor. To start with, Britain was nowhere near end-state capitalism of the sort where gains from continued exploitation of foreign labor market were impossible. Thanks to continued advances in mechanization and, eventually, computerization, it turns out that the Britain which was fervently exploiting the Indian subcontinent (among other places) had generations to go before anything which could be considered the ‘final’ stage of capitalism, contradicting Marx’s thesis. At most, it would appear that Britain’s higher level of industrialization and economic development enabled the conquest of India, not that that caused it as some sort of machination of the capitalist class.
Second, as Schumpter pointed out, many examples of European colonialism could not be justified on the posited economic grounds. Marx’s theory of imperialism completely fails to explain France’s adventures in northern Africa as well as Italy’s conquest of Ethiopia. (We might add Belgium’s colonial adventures in the Congo to Schumpeter’s examples.) The sorts of economic windfalls which Britain reaped from India simply were not available. Epicycles must be added onto the theory--most plausibly, that the capitalist classes which pushed for war in northern Africa or Ethiopia were mistaken as to the economic gains which they thought would result from conquest. But was this historically true with regard to any critical mass of capitalists, particularly ones responsible for organizing the efforts at colonization? Were capitalists even behind these colonial ventures in the first place? Even if so, we are adding complexity to the theory which damage its power as a general social scientific mechanism: We are no longer looking at the interests of dominant economic classes, as such (the focus of Marxian theory), but at something closer to a methodologically individualistic economic neoclassicalism involving imperfect information (or whatever else) which is largely hostile to a Marxian framework.
Third, again, as Schumpter himself recognized, it would take an enormous (and almost literally tautological--consistent with any empirical evidence which could possibly emerge, so as to vitiate the empirical explanatory power of the theory) invocation of false consciousness to understand colonialism along nakedly class lines. Colonialism was, we might say, a nationalistic project, one which bound together members of all classes into a national project of conquest (or resistance?), one which fully contradicts a class-based understanding of the underlying phenomena. The rich and poor of European nation-states collaborated with one another to dominate their colonies; there were, conversely, no significant lines of solidarity between, say, the working poor of Britain and of India. Again, there are ways that we can fix the lack of fit between theory and data, but the fact that we must do so does not bode well as the theory for a general framework of social science explanation.
Enter the idea of the social scientific tautology. The Marxian economist, confronted with the fact that his theory perhaps fails to explain more of the data than it actually explains (i.e. France and Italy compared to Britain) and explains poorly that data which it can sort of explain (i.e. in positing that the Britain of the nineteenth century was in ‘end stage’ capitalism where the forces inherent in capitalism which would eventually bring it to its knees were near their final form) can always rescue the theory by playing with what counts as evidence. Here is the full footnote in which Schumpter explains what he means by solving the problem of Marxian interpretation ‘by tautology':
The danger of empty tautologies being put over on us is best illustrated by individual cases. Thus, France conquered Algeria, Tunisia, and Morocco, and Italy conquered Abyssinia, by military force without there being any significant capitalist interests to press for it. As a matter of fact, presence of such interests was a pretense that was very difficult to establish, and the subsequent development of such interests was a slow process that went on, unsatisfactorily enough, under government pressure. If that should not look very Marxist, it will be replied that action was taken under pressure of potential or anticipated capitalist interests or that in the last analysis some capitalist interest or objective necessity “must” have been at the bottom of it. And we can then hunt for corroboratory evidence that will never be entirely lacking, since capitalist interests, like any others, will in fact be affected by, and take advantage of, any situation whatsoever, and since the particular conditions of the capitalist organism will always present some features which may without absurdity be liked up with those policies of national expansion. Evidently it is preconceived conviction and nothing else that keeps us going in a task as desperate as this; without such conviction it would never occur to us to embark upon it. And we really need not take the trouble; we might just as well say that “it must be so” and leave it at that. This is what I mean by tautological explanation.
Analogies between the philosophy of science in general and the philosophy of social science are always dangerous, due to the radically different modes of inquiry in, say, the natural sciences and the social sciences, but we’re again left with something like an analogue to the Popperian degenerating research paradigm: Rather than using our theory to explain the data, we are desperate to pigeonhole the data so that it somehow manages to comport with the theory. A few pages later, Schumpter continues with the idea that it’s tautological to explain all macro-level social phenomena through the lens of the class interests of the bourgeoisie, as the dominant class in capitalist society:
[I]n practically all cases the theory can be made tautologically true. For there is no policy short of exterminating the bourgeoisie that could not be held to serve some economic or extra-economic, short-run or long-run, bourgeois interest, at least in the sense that it wards off still worse things. This, however, does not make the theory any more valuable.
Schumpter’s point is thus that, if our theory of imperialism is that there will be some capitalists who perceive an advantage of the sort central to Marxian economic analysis who will seek to take advantage of a state drive to imperialism in order to advance ‘class’ interests, we will never be lacking for confirmation of the theory, because some capitalist, somewhere will be trying to take advantage of any economic opportunity which arises in order to earn a profit through competitive advantage--here, through favorable access to cheap labor markets. If this is to count as evidence of the theory, then it is virtually inconceivable that the real world will ever throw up any falsification of it. The tautology exists in the fact that the scientific theory can seek (and find!) its confirmatory evidence in virtually any fact pattern which the social universe can throw at it, and all the worse for the theory! It stands as no more than an ex post rationalization of the facts, whatever the facts may be; the proper relationship between causal mechanism and phenomenon does not exist.
In fact, this feature makes the theory empirically useless, in something like the way that rational choice theory is useless as an empirical theory unless fleshed out with contingently plausible assumptions. If we manipulate the discount factors which specify the agents’ time horizons, the utility functions which define their goals, the information sets which specify the structure of the game as they understand it, and so on, we can make virtually any behavior look like a best response to the environment, given the desires and beliefs we have stipulated for the agent. A much thicker set of constraints are needed to make the theory a theory at all. In the case of both Marxism and rational choice theory, this feature risks making the theory vacuous, not useful. At best, each often serves as a sort of methodological lens: what can we learn about phenomena from the special cases in which, on a closer look, Marxian analysis (or rational choice theory) breaks down? Is there a systematic failure which itself points to a better theory of social behavior? Fact patterns in which behavior appears to be irrational or contrary to group economic interests potentially make for more interesting case studies than those which fit well.
If the social scientific tautology is to be a recurrent problem in the social sciences (and not merely a special feature of the Marxian system), it should have some generalizable features which encompass other misfirings of social scientific explanation. Schumpter provides the basic framework for a social scientific tautology himself--situations in which we see “[a]n apparent verification by prima facie favorable cases which are not analyzed in detail.” Appeal is made to familiar cases where “the facts in question [are] superficially known to everyone and...thoroughly understood by very few.” These conditions are important: the fact that the structure of the social scientific problem is superficially understood by virtually all educated people provides a yardstick for ‘successful’ causal explanation in the imaginations of the ‘educated’ class, a standard which is too easy to meet because the deeper structure of the phenomena is insufficiently understood. The search for fit between theory and facts thus suffers from a fatal case of confirmation bias; the analyst ‘discards’ uncomfortable facts from the perspective of the theory because, most often, he has no need to become aware of the existence of these uncomfortable facts in the first place. We thus have, potentially, the intellectual misfiring which results from a cognitive bias well known to the psychological literature. The end result is that theories are confirmed based on ‘evidence’ that would be woefully inadequate to convince a properly trained scientist approaching the topic in good faith.
Social scientific tautologies often do, but need not, be used to defend functional explanations in the social sciences. A critique of functionalism has been a near-constant feature in the work of Jon Elster, who attacked not only the writings of Marxists (including most notably Marx himself), but also Foucauldians, psychoanalysts, sociologists such as Bourdieu, and others. We can understand functionalism as causal explanation of a phenomenon in terms of its effects, particularly when there is no adequately specified mechanism (evolutionary adaptation, feedback loops of various sorts, purposive attempt to bring about the effect by specific agents, and so on) which allows the effects to serve as a legitimate causal explanation for the underlying phenomenon. Functionalism, in this pejorative sense, is a pervasive feature of Marxian explanation, but it can also appear in more ‘respectable’ domains of the social sciences, such as the ‘rational choice functionalism’ of explaining e.g. sociological phenomena (say, norms of etiquette) based on their useful effects for a discrete ingroup (say, social elites) without a causal mechanism tying the development and continuation of, here, norms of etiquette to the actions purposefully undertaken based upon the desires and beliefs of the relevant agents. It’s a cheap facsimile of a valid causal explanation in that it provides no real understanding of the underlying phenomenon but can, in someone sufficiently ill-informed to not see the lack of fit between explanandum and explanans, provide a misfiring of the “Aha, I understand it now!” satisfaction with which we are all familiar. This phenomenological misfiring is important: the subjective light bulb moment (for lack of a better term) we experience when we have discovered something new can be triggered just as easily from spurious associations as correct ones. Pseudo-mechanisms, such as social scientific tautologies, provide the pleasure of genuine understanding (which is intense, at least for those of us who spend our professional lives seeking it) without requiring one to put in the work of uncovering a true scientific explanation. The attempt to trigger light bulb moments in others, in order to convert another to one’s own understanding of the universe, is also easy to understand both from the perspective of the agent and the object of the conversion: to the agent, it offers whatever benefits inhere in convincing others to the agent’s chosen ideological framework; to the object, it offers an endless succession of near-zero cost lightbulb moments, which are both pleasurable in themselves and a justification for intellectual laziness with regard to developing a true understanding of the underlying social phenomena. It is thus not difficult to understand how Marxism might have found generations of both apostles and converts. (I leave to the reader whether this is itself a functionalist pseudo-mechanism!)
Social scientific tautologies are, again, a paradigmatic example of what I think of as pseudo-mechanisms. If a mechanism is a regularity in individual or group-level social behavior which provides a causal explanation for some subset of social phenomena, a pseudo-mechanism is a recurring way in which misfiring of the processes of causal explanation leads to false explanations. Functionalism without rigorous specification into individual level beliefs and desires, in which cui bono serves as a question which, when answered, provides a fully adequate causal explanation in itself, is also a form of pseudo-mechanism. Thus, if we postulate that the social habits of the old rich are intended to keep out new rich who are unfamiliar with their customs, we invoke the pseudo-explanation of functionalism if we do not specify the actual agents who create and sustain the norms against free riding.
I do not think that tautological explanation is anything like a dead problem in the social sciences, one which faded away with the decline of Marxism as an intellectual framework taken seriously by anything like a critical mass of western elites. Cursory analyses appear to suggest that various schools of Neo-Marxism, and, perhaps more pervasively, Neo-Foucauldianism are on the rise in various domains of the social sciences. The only defense against pseudo-mechanisms, of course, is the doing of rigorous social science. In the overwhelming majority of cases (although perhaps not all of them) this involves a commitment to ultimately explaining social phenomena in terms of well-specified beliefs and desires held by individual agents and the decision environments in which they act, a weak form of methodological individualism. (I specify 'weak form,' because I understand methodological individualism to be fully consistent with, for example, the limited 'holism' of List and Spiekermann summarized here The development of such rigorously laid out micro-foundations is an ideal which is perhaps often impossible to reach in practice, but seeking to fulfill it (and occasionally finding oneself frustrated by the various constraints, practical and theoretical, which impose themselves on the social scientist) is likely the only way to develop the capacity for judgement which separates the competent social scientist from the pseudoscientist who happily falls into tautological explanation. And, if it is sometimes difficult to tell good social science from bad, it seems to me to be quite often apparent when one is reading the work of someone who is not even trying to do social science well.