. 4
( 7)


(4) The scent theory: Women use fragrances which “beguile wasps, but
which then sting because they become aroused and then aggrieved
when they discover that the bouquet stems not from ¬‚owers and react
to frustration by aggression” (1998: 77). Critical to this mechanism
are assumptions about wasp behavior.
Each of these could explain the outcome. But which, if any, is true?
Hernes points to standard methodology: take the Rambo theory. If
women are more tender, then they should be less tolerant of pain. Does
any research support this? For each of these theories, we can test the truth
of assumptions with evidence and argument. Very often this requires
drawing out the implications of the assumptions, and accordingly, it
requires a strenuous effort to see exactly what those assumptions are.
Unfortunately, not only will this not be easy, but it is easy to fail to notice
that assumptions which may be critical are being made.
Lawson (1997) also provides a wonderful example. He cites Leamer™s
account (1983) of the predicament of the applied econometrician:
The applied econometrician is like a farmer who notices that the yield is somewhat
higher under the trees where birds roost, and he uses this for evidence that bird
droppings increase the yield. However, when he presents his ¬ndings . . . another
farmer . . . objects that he used the same data but came up with the conclusion
that moderate amounts of shade increase the yields . . . A bright chap . . . then
observes that these two hypotheses are indistinguishable, given the available data.
(1997: 214)
Lawson answers:
The obvious response of course, albeit one that econometricians occupied with
¬tting a line to given sets of data rarely contemplate, is to add to the ˜available data.™
Speci¬cally, the aim must be to draw consequences for, and seek out observations
on, actual phenomena which allow the causal factor responsible to be identi¬ed.
If, for example, bird droppings are a relevant causal factor then we could expect
higher yields wherever birds roost. Perhaps there is a telegraph wire that crosses
the ¬eld which is heavily populated with roosting birds, but which provides only
Agents and generative social mechanisms 101

negligible shade . . . Perhaps too there is a plot of land somewhere close to the
farm house which is shaded by a protruding roof, but which birds avoid because
of a patrolling cat . . . The fact that it is not possible to state categorically at
this abstract level the precise conditions under which substantive theories can be
selected amongst, i.e., without knowing the contents of the theories themselves
or the nature or context of the conditions upon which they bear, is an unfortunate
fact of all science. (1997: 214)

Lawson™s more general conclusion deserves quoting(1997: 214): “Sci-
ence is a messy business. It requires an abundance of ingenuity, as well as
patience, along with skills that may need to be developed on the job.”16
One ¬nal point needs to be considered here. It is the question of the
role, limits and / or advantages of verstehen in the construction and eval-
uation of social mechanisms. In a well-known passage, Weber asserted
that in the human sciences, “we can accomplish something which is never
attainable in the natural sciences, namely the subjective understanding of
the component individuals” (1968: 15). In one sense, this merely points
to a huge difference in the nature of theorized mechanisms in the nat-
ural and human sciences. Since persons are the critical causes of what
happens in society, a social mechanism must appeal to their beliefs and
motivations. But is this an advantage?
Some writers, perhaps including Weber, have thought so. There is no
argument, perhaps, that it is an advantage when it comes to building a
model of a mechanism. As Bhaskar notes,
How, then, given the mishmash nature of social reality, is theory-construction
accomplished in social science? Fortunately, most of the phenomena with
which the social scientist has to deal will already be identi¬ed, thanks to the
concept-dependent nature of social activities, under certain descriptions. (Bhaskar,
1979: 63)

Thus, we know what it is to cash a check and to need a job. As already
suggested, theory construction must, inevitably, draw on common stocks
of knowledge. As Schutz (1970) rightly noted, the explanation must be
comprehensible to the lay person.
One might also argue that considerations of plausibility enter into our
assessment of hypothesized social mechanisms. An account may be plau-
sible in the sense that it con¬rms widely available beliefs and understand-
ings (prejudices?). Clearly there is a manifest danger here. Fifty thousand
Frenchmen can be wrong. On the other hand, in order to understand one
another, we must have some understanding of the motivations and goals
of others even if we can be mistaken on any given instance. Still, if the

16 See also Sayer, 1992: chapter 7.
102 A Realist Philosophy of Social Science

account is plausible and can be sustained evidentially, we may legitimately
have some con¬dence in it.
But there remains the problem that actors can be mistaken about the
social world which their activities sustain. Does not this suggest a dis-
advantage for theory construction in the human sciences? This objec-
tion would seem to rest on a misunderstanding of how the mechanism
explains. There are two problems to be solved. There is the problem of
understanding the beliefs of members. This obviously requires evidence.
But there is also the problem of explaining the beliefs and actions of mem-
bers, of discovering the existing conditions and consequences of action “
which may or may not be known or acknowledged. This, too, requires
evidence. It may be that the members do grasp reasonably well the social
world that their action sustains. As in the work of Willis and Goffman, to
take two outstanding examples, the mechanism may be able to show that
the outcome would not be what it is unless the actors had false or partial
beliefs about the conditions and consequences of their actions. That is, in
these cases, explaining the outcome requires seeing that the actors failed
to have an adequate grasp of the social world that is the ongoing product
of their actions.
I have so far argued that in both the physical and the social sciences
understanding requires identi¬cation of the generative mechanisms of
outcomes. In chapter 3, it was argued that it was implausible to believe
that we could improve on our ordinary capacities to understand the
actions of concrete persons. Without social science, we can have an
entirely adequate understanding of why your boss expects punctuality,
of why your spouse wants to visit family at Christmas. In this chapter, it
was argued that theory abstracts from the concrete reality of the actors
and situations to get at the logic of a social process, for example, to achieve
an understanding of an increased divorce or crime rate. But more needs
to be said about how social mechanisms function in explanatory contexts.
This takes us back to considering the goals of social science and forces
us to address the question of the relation of history to social science.
The question of “historical sociology” has been the battleground for this
debate. The next chapter takes this debate, historically and currently, as
its point of departure.
5 Social science and history

One could argue that the classical sociologists, Montesquieu, Comte,
de Tocqueville, Marx, Weber, even Durkheim, were historical sociolo-
gists. All of them worked with historical materials and all shared in the
idea that this was essential to what they were doing. They did their work
before sociology emerged in the twentieth century as a distinct discipline.
Even though Marx disapproved of much work which had been done by
people who identi¬ed themselves as historians, he believed, indeed, that
history was the human science.1 Similarly, Weber seems to have believed
that sociology was, as he put it, a propaedeutic for historical work. That is,
sociology could provide the tools and concepts for good historical inquiry,
but was not itself an independent body of knowledge.2
Beginning in the so-called Methodenstreit (battle of methods) toward
the end of the nineteenth century, the literature has tended to employ
a distinction formulated by Windelband in 1894 between two kinds of
inquiry: “nomothetic” versus “idiographic.” It is generally held that the
natural sciences are nomothetic “ they are engaged in the search of laws,
while the human sciences, including history, are idiographic “ their object
of concern is the concrete particular in its uniqueness. But there is con-
siderable disagreement as to whether the grounds for this difference are
methodological, epistemological or ontological. In what is perhaps its
most recurring form, two kinds of explanation are at issue: on the nomo-
thetic view, explanation is in the form of the covering law model. It is just
this that is rejected by defenders of the idiographic view. For them, the

1 In the German Ideology, Marx and Engels write: “We know only one science; the science
of history.” Quoted in Simon, 1994: 107.
2 On Weber, there is a vast literature, of course. For some critical and historical background,
see Manicas, 1987: 127“140, and more recently, Fritz Ringer, 1997. Also see notes 7 and
9 below. My effort pursues themes set out by Weber.
Durkheim had a different take. For him, “history can only be a science on condition that
it raises itself above the particular; but then it ceases to be itself, and becomes a branch
of sociology. It merges [as Comte would have said] with dynamic sociology” (1972: 78).

104 A Realist Philosophy of Social Science

human sciences, including history, require a distinct form of explanation
in which the goal is to grasp the meaning of human action. For the nomo-
thetic sciences, general theory is essential, and induction and deduction
are the primary tools; for the idiographic sciences, the structure of expla-
nation, as in history, is narrative and there is no role for theory in the
sense assumed in the natural sciences.
Except for some very important exceptions, historical sociology was
very nearly extinguished by the middle of the twentieth century. Although
the story would be complicated, this was largely a consequence of dis-
ciplinary specialization coupled with the vigorous effort on the part of
modern sociologists to model their work on lessons presumably learned
from natural science. From this perspective, since it did not seek “gen-
eral explanatory variables,” history was not nomothetic, hence not a sci-
ence. Sociologists, accordingly, could leave narrative to historians and get
on with their own important efforts at discovering “general explanatory
Such a view was, at best, a distortion of genuine science “ a distor-
tion which, fortunately, the classic writers did not need to confront. The
writers who continued to practice historical sociology were not in the
mainstream of academic sociology, even when their work was recognized
as signi¬cant. One thinks here of some of the work of writers with variant
understandings of the work of Weber, for example, Benjamin Nelson,
Reinhard Bendix, C. Wright Mills and Barrington Moore. One thinks
also here of some Marxists who were historians, including Christopher
Hill, Eric Hobsbawm and M. I. Finley, and of historians in¬‚uenced by
Marxism, for example, Marc Bloch, and of a few others even harder to
classify, such as Norbert Elias and Karl Polanyi.
Since the 1970s, however, sociologists have again begun “to reach for
history.” As Tilly writes:
Historical analyses of industrialization, of rebellion, of family structure began to
appear in the journals that sociologists read. Departments of sociology began
hiring specialists in something called “historical and comparative analysis.”

3 It is also important to examine the role of Talcott Parsons in the current
(mis)understandings of Weber. See Grathoff (ed.), 1978; Wagner 1983; and Camic 1987.
In his 1979 New York Times sympathetic re¬‚ection on the legacy of Parsons, Daniel Bell
noted that charges against Parsons™s style of sociology rested on a misunderstanding of
the difference between history and sociology. Following a version of Durkheim, but cer-
tainly not Weber, Bell asserted, “there is no science of the particular; it is necessary to
generalize. In the sciences the aim is to establish the invariant features of phenomena.”
Parsons repeatedly warned his readers that he was not dealing with concrete phenomena
but with “ ˜analytical abstractions,™ a set of logical categories into which all social actions
would ¬t.” Parsons™s “integration” of Weber, accordingly, vitiated the core of Weber™s
views on social science (see below).
Social science and history 105

Sociological authors began to write as if when something happened seriously
affected how it happened. Some few sociologists actually began to learn the basic
historical skills: archival exploration, textual analysis, and the like. History began
to matter. (Tilly, 1982: 38)

One suspects that this rediscovery of history was due, in part at least,
to the onslaught against the conventional wisdom in the philosophy of
science, to the eclipse “ at least of¬cially “ of the dominating work of
Talcott Parsons and of structural functionalism, and perhaps also, as
Tilly suggests, to the new skepticism regarding progressivist theories of
modernization and development.4

The recent past
Books about historical sociology by Stinchcombe (1978), Tilly (1982,
1984), Abrams (1983), Smith (1991) and Skocpol (1984, 1994), not to
mention a host of important books in historical sociology by both younger
writers and well-established authors, demonstrate a continuing vitality
and, I hasten to add, a continuing disagreement over strategies for joining
history and sociology. There was even a “counter-revolution” against the
very idea of joining the two. Goldthorpe (1991) strenuously defended the
covering law model and insisted on a sharp division between history and
sociology. Most recently, we have two important collections of essays: one,
edited by McDonald (1996), The Historic Turn in the Human Sciences, has
a broad scope; another, edited by Mahoney and Rueschemeyer (2003),
Comparative Historical Analysis in the Social Sciences, focuses on compara-
tive methods. There is also a new and useful handbook on historical soci-
ology (Delanty et al., 2003). This volume covers most of the ground and
concludes that “historical sociology is deeply divided between explana-
tory ˜sociological™ approaches and more empirical and interpretative ˜his-
torical™ approaches.” Finally, there is lively debate in both The American
Journal of Sociology (1998) and in a recent book (Gould, 2004) between
Margaret Somers and advocates of rational choice theory regarding the
role, if any, of “general theory” in historical sociology. (See appendix C,

4 Craig Calhoun (1996) provides a useful sociology of historical sociology. Important here
is his observation that the work of the 1970s and 1980s made the effort to legitimize
historical sociology by arguing that it could be as rigorous as other forms of sociology.
Presumably “other forms of sociology” more nearly approximated the methods assumed
to be true of natural science. Theda Skocpol™s effort to employ Mill™s methods is a good
example. See appendix B for critical discussion. For additional re¬‚ections on the sociology
of historical sociology, see Skocpol, 1994 and Delanty et al., 2003.
106 A Realist Philosophy of Social Science

Symptomatic of confusion over the relevant issues, writers disagree
even on their understanding of what some of the better-known stud-
ies were doing. At one extreme we ¬nd Stinchcombe arguing that “the
difference between Trotsky™s Marxism, Smelser™s functionalism, and de
Tocqueville™s conservative despair makes hardly any difference to any
important question of sociological theory” (1978: 2). Thus, “when they
do a good job of historical interpretation, Marx and Weber and Parsons
and Trotsky and Smelser all operate in the same way.” This is surely
puzzling: perhaps they rarely “do a good job” or what counts as “the
same way” is very fuzzy “ despite Stinchcombe™s interesting readings of
the authors he discusses.
Although entirely different in aim and approach, Dennis Smith™s The
Rise of Historical Sociology (1991), which gives an account of eighteen
noteworthy writers, puts him, ultimately, near to Stinchcombe as regards
differences between historical sociologists. He gives even-handed descrip-
tions of each of his eighteen selected writers, usually in pairs, usually
arranged by topics: for example, under “old empires, new nations,” he
discusses Eisenstadt and Lipset, and under “two critical rationalists,” we
¬nd Barrington Moore and E. P. Thompson. In his last chapter he iden-
ti¬es four relevant issues: “whether historical sociologists have operated
as ˜outsiders™, or as members of the relevant ˜establishment™; the way
they handle problems of involvement and detachment; third, their orien-
tations toward theory, empirical generalization and primary exploration
of historical data; and fourth, the strategies of explanation they adopt.”
Important as these notions are, they seem to be almost afterthoughts: they
do not drive the accounts of the writers he discusses. More importantly,
perhaps, he is very uncritical of notions of theory and explanation. And
on all four of the issues, he ¬nds no clear lines at all, offering instead four
“strategies of explanation”: competitive selection, system contradictions,
infrastructural capacities and dominant routes of social change.
However, these are not best construed as “strategies of explanation,”
but as theoretical orientations, predicated very much on very different
notions of the nature of history and of society. In any case, he ¬nds
some combination of these in all the writers he has discussed. Presumably
Smelzer™s work exempli¬es the dominant route strategy “ a stage theory “
but in a less pure form, it is also found in Runciman, Wallerstein, Lenski,
Moore and Anderson. With Moore and Anderson, however, it is comple-
mented with evolutionist assumptions, and, in the case of Moore (Ander-
son is not mentioned), “great attention to the infrastructural capacities
of dominant and subordinate classes within agrarian polities.” He thus
meets Mann, who, as it turns out, shares with Lenski “a location between
infrastructural capacities and dominant routes.” One could go on. The
Social science and history 107

upshot is the feeling that historical sociologists are quite messy beasts and
that as far as method or strategy is concerned, an eclectic anarchy is to
be recommended. On the other hand, one wonders whether the prob-
lem resides more in the way that the material is being conceptualized by
these commentators. It may be that, as with much talk about science, the
problem stems from an inapt theorizing of the actual practices.

Modes of comparison: individualizing, universalizing
and variation ¬nding
A similar generosity is found in Tilly™s suggestive, but ultimately unhelp-
ful, classi¬cation. Tilly identi¬ed four approaches to historical sociology.
For him the key difference is the mode of comparison: individualizing,
universalizing, variation ¬nding, and encompassing comparisons. Thus,
a purely individualizing comparison treats each case as unique, taking up one
instance at a time, and minimizing its common properties with other instances. A
pure universalizing comparison, on the other hand, identi¬es common properties
among all instances of a phenomenon . . . [Variation ¬nding] is supposed to
establish a principle of variation in the character or intensity of a phenomenon
by examining systemic differences among instances. (Tilly, 1984: 81“82)5

While Tilly™s account of comparison is provocative, it suffers, as he seems
to acknowledge, in part, because writers are anything but clear or con-
sistent as to whether they are individualizing, universalizing or variation
¬nding. Indeed, he undermines his own classi¬cation by remarking:
If we needed a pedigree for individualizing comparison, its use by Max Weber
would suf¬ce. When Weber started elaborating his great taxonomies, he bowed
toward generalization. When he spoke of rationalization and charisma, he ges-
tured toward universalizing comparison. But his wide comparisons of religious
systems served mainly to specify the uniqueness of the achieving, accumulating,
rationalizing bureaucratic West. To a large degree, Max Weber used comparison
for the purpose of individualizing. (1984: 88)

Weber is a critical ¬gure in the literature of historical sociology and his
work has been put to a number of uses. In what follows, we will argue
that his work remains fundamental for any plausible version of a histori-
cal sociology. However, “individualizing” cannot serve as a way to distin-
guish these often inconsistent efforts. When Reinhard Bendix and Perry
Anderson are both identi¬ed as individualizers we may suspect that we

5 “Encompassing” is suf¬ciently unlike “comparison” to omit it from discussion here.
Tilly de¬nes it as follows: “[Encompassing] places different instances at various locations
within the same system, on the way to explaining their characteristics as a function of
their varying relationships within the system as whole” (1984: 83).
108 A Realist Philosophy of Social Science

have missed what is essential in comparing their work. Moreover, as later
chapters in Tilly™s very useful book show, no writer fails to do some indi-
vidualizing, some universalizing and some variation-¬nding. As is plain
enough from Tilly™s own formulation of these strategies, one is tempted
to say that Weber could not help both individualizing and generalizing
since individualizing presupposes generalizing. Thus, we need to see what
counts as a bureaucracy “ a step in generalizing “ and then to see how the
bureaucracies of China differed from bureaucracies in modern capitalist
societies “ an individualizing step.
Much confusion, unfortunately, attends the idea of generalization. In
the ¬rst place, it is quite indispensable. Whenever we use an abstract
noun, we are committed to a generalization, however vague, however
open textured or ideal-typical. Thus, when we identify an institution as
a bureaucracy, we assume that there are some properties connoted by
the term which enables us to call institutions of China and institutions in
modern capitalist states bureaucracies. On the other hand, it is true, but
trivially so, that every concrete particular is unique, so the real question
is whether the individualizing is non-trivial “ essential to our interest in
understanding the particular concrete under study. Thus, what features
of pertinence distinguish bureaucracies in capitalist societies from those
of China? Similarly, variation ¬nding requires both individualizing and
generalizing, ¬nding differences along a continuum.
There are several real questions here. One, clearly seen by Tilly, is the
question of whether our interest is in what is distinctive or in what is
common. Weber insisted that physical theories, for example, the physics
of masses, apply to all masses at any time and place, and such highly
abstract knowledge is interesting to us. When it comes to the human
sciences, it is the concrete in all its individuality which interests us. Thus,
he wrote famously: “the type of social science in which we are interested
is an empirical science of concrete reality (Wirklichkeitswissenschaft). Our
aim is the understanding of the characteristic uniqueness of the reality
in which we move” (1949: 72). For him, even if social science were to
model itself as an abstract science, like physics, and offer general theories
and propositions true of all human groupings, such knowledge would be
neither useful nor interesting.
This turns out to be no small matter, especially as regards compar-
ison. On this view, comparisons are not employed primarily to ¬nd
generalizations “ what is true, for example, of all bureaucracies “ but
rather to sharpen one™s understanding, for example, of “the unique-
ness of the achieving, accumulating, rationalizing bureaucratic West.”
More importantly, on Weber™s view of the matter, the goal is not to settle
Social science and history 109

for the generalization, for example, that in pre-revolutionary France and
pre-revolutionary China, “peasants were the critical class component,”
but to see as precisely as one can, the differences in peasant relations
in these two cases. Finally, while there can be little doubt that at some
level of abstraction, there will be resemblances both in the types and in
the sequences characteristic of, say, commercialization or state-building,
differences in outcomes can be explained only by identifying different
causes in these similar sequences.6 The comparative method becomes,
on this view, both a method of discovery and a way to test hypotheses
about causes. I will return to this below.
Second, and as important, there is the question of what one does with
generalizations. Tilly™s classi¬cation suffers mainly from the fact that it
gives us no help in seeing what comparison is doing for us. In particular,
is it the main goal of a historical sociology to seek generalizations (and
law-like statements) in order to explain by covering laws? Presumably, it
is just this that distinguishes sociology as a science and separates it from
history “ the familiar “nomothetic / idiographic” divide.7
It is important to see here that Weber (1975) revised the prevailing
nomothetic / idiographic bifurcation and insisted that “the logical pecu-
liarity of ˜historical™ knowledge in contrast to ˜natural-scienti¬c™ knowl-
edge . . . has nothing at all to do with the distinction between the ˜psy-
chical™ and the ˜physical,™ the ˜personality™ and ˜action,™ on the one hand,
and the dead ˜natural object™ and the ˜mechanical processes of nature,™ on
the other (1975: 184“185). The key difference is in the goals of two kinds
of science: the nomological or abstract sciences employ laws which are
unconditionally and universally valid. The sciences of concrete reality aim
at knowledge of the particular. “Because of the logical impossibility of an
exhaustive reproduction of even a limited aspect of reality . . . this must
mean the following: knowledge of those aspects of reality which we regard
as essential because of their individual peculiarities (Weber, 1975: 57).
In turn, we can then identify what in the in¬nitely complex causal history
of the concrete explains it (see below).
6 In The Protestant Ethic, Weber notes that he “treated one side of the causal chain,” while
in his more extensive studies of religion, in order to “¬nd points of comparison with the
Occidental development,” he aimed at ¬nding causal relationships “to economic life and
social strati¬cation.” “For only in this way is it possible to attempt a causal evaluation
of those elements of economic ethics of the Western religions which differentiate them
from others, with a hope of attaining even at tolerable degree of approximation” (1958:
27). Too often Weber is read as giving a “culturalist” (and monocausal) explanation in
response to “socio-economic” (and also monocausal) Marxist accounts.
7 Charles Ragin and David Zaret (1983) have argued for a version of Weber which focuses
on the particular features of concrete cases and rejects as impertinent a Durkheimian
search for general explanatory variables.
110 A Realist Philosophy of Social Science

There are, as already noted, at least two other functions of general-
ization. First, it is presupposed in individualizing: we identify critical
differences in a set of outcomes or sequences which at some level of
abstraction resemble one another. For example, we see that in Case 1,
unlike Case 2, the peasants had property rights and that this made a
difference in the outcomes, exactly because different choices were pos-
sible. Second, hypotheses generated by comparative study will serve as
promissory notes calling for explanation. Thus, comparative study may
yield the generalization: “In a highly bureaucratic absolutist state, the
landed nobility has little political power.” Such generalizations, if true,
may provide some explanation, depending on the question of interest.
But, in turn, they demand explanation: what are the mechanisms that
explain the generalization? The analogy to the natural sciences is exact:
molecular theory begins with unexplained generalizations, most of which
are common to ordinary experience. Sugar dissolves in water; iron rusts.
Molecular theory tells us why. In the writings of historians and historical
sociologists, one encounters a host of generalizations, many indeed which
come directly from ordinary experience, for example, “It is not easy to
challenge a bureaucracy,” or “Bureaucracies are not easy to control.” If
true, there will be reasons for this and these will be given in terms of
causal mechanisms which regard the capacities, beliefs and behavior of
Tilly offers that the relative value of the strategies he identi¬es “depends
upon the intellectual task at hand.” This is certainly the case as to whether
the particular problem calls for individualizing, generalizing or variation
¬nding. But he also sees that the value of strategies “depends on the
nature of the social world and the limits to our knowledge of that world”
(Tilly, 1984: 145). Indeed, one of most powerful parts of his book is his
epistemologically and ontologically sensitive analysis of the “pernicious
postulates,” for example, that “a single recurrent social process governs
all social change” (1984: 33), or that “abstractly speci¬ed processes such
as differentiation or concentration, mark out the limits for intelligible
analysis” (1984: 50). As he would probably agree, if we want to iden-
tify strategies for inquiry, it is desirable to provide a classi¬cation which
takes for its criterion of demarcation a feature which cuts deeply into
methodological, epistemological and ontological issues. While there are
some serious problems with it, Theda Skocpol™s (1984) classi¬cation of
approaches in terms of explanation strategies is exactly what is called for.
A review of this will also allow us to see more clearly how the argument of
the present volume relates to existing literature on the question of history
and sociology.
Social science and history 111

A taxonomy of explanation types
Skocpol offers a trichotomy of strategies that might be termed “function-
alist universalist,” “analytical historical” and “interpretative historical.”
Skocpol is also unwilling to risk dogmatism and ¬nds that writers very
often mix strategies “ assuming, presumably, that no problems of coher-
ence arise?
The basic orientation of the ¬rst subtype, “functionalist universalist,”
is expressed well by S. M. Lipset:
From an ideal-typical point of view, the task of the sociologist is to formulate
general hypotheses, hopefully set within a larger theoretical framework, and to
test them. His interest in the way in which a nation such as the United States
formulated a national identity is to specify propositions about the general process
involved in the creation of national identities in new nations. Similarly, his concern
with changes in the pattern of American religious participation is to formulate
and test hypotheses about the function of religion for other institutions and for
the social system as a whole . . . These are clearly not problems of the historian.
History must be concerned with the analysis of the particular set of events or
processes. Where the sociologist looks for concepts which subsume a variety of
particular descriptive categories, the historian must remain close to the actual
happenings. (quoted in Tilly, 1982: 5)
The basic idea is clear. The task of the (historical) sociologist is to use
theory to generate some general hypotheses which, if true, would explain
the particular event under examination. For the functionalist, the theory
will be a version of structural functionalism.8
In Lipset™s example, we have hypotheses about “the general process” of
nation-building. We might argue that a key modernizing process is differ-
entiation, including the increasing division of labor in society, increased
institutional separation and thus accentuated individualism. For example:
“Whenever a society undergoes modernization, there is an increasing divi-
sion of labor in society.” Coupled with other hypotheses, for example,
“When religious institutions are weakened, there is a loss of normative
control,” we are led to the conclusion that an essential requirement for
the continuing stability of the social system is the development of organs
of “authoritative interpretation and enforcement” “ a legitimated legal
system and the coercive forces of the police. With this theory, then, one
goes into history and examines nation-building in a variety of contexts.
Skocpol provides some powerful criticism of functionalist universal-
ism. First, “the model itself has to be taken as given prior to its historical
8 Rational choice theory is another candidate for such a general theory. See below
appendix C.
112 A Realist Philosophy of Social Science

application” (Skocpol, 1984: 365). It is important to stress what this must
assume. It assumes that there are universal principles of social reproduc-
tion and social change, even if the particular forms they take are his-
torically variable. If, however, as Mills (1959) long ago insisted, we do
not know of any such universal principles, that these vary with the social
structure we are examining, then this criticism is fatal. Second, how can
we be sure that different investigators would concretize such abstract con-
cepts as “differentiation” or “mass organization” in the same way? Here
the problem is the wobbly character of critical general terms. Perhaps
indeed, in order to save functionalist assumptions about change, almost
anything can be made to count. Finally, and following on this, perhaps
historical facts are omitted or distorted to ¬t the preconceived theory.9
For the functional universalist, histories are case studies meant to elab-
orate and demonstrate the validity of universally valid theoretical ideas.
Case studies allow the theorist to move from the abstract to the concrete.
By contrast, the analytic sociologist “aspires to generate new explanatory
generalizations through comparative historical analysis,” or alternatively,
to “discover causal regularities that account for speci¬cally de¬ned histor-
ical processes or outcomes, and explore alternative hypotheses to achieve
that end” (Skocpol, 1984: 362). For example, while Skocpol hopes to
induce a general theory of modern revolution from studies of cases, Lipset
hopes to test his general theory against cases. They thus differ fundamen-
tally in the use to be made of history.
For Skocpol, analytic historical sociologists “acknowledge the desir-
ability of generalizable explanatory principles” (1984: 375), but they
stand between those who seek “a single overarching model” and those
who restrict themselves to “the meaningful exploration of the complex
particularities of each singular time and place” (1984: 374).10 This is
certainly plausible. But like the functionalist universalist, Skocpol is com-
mitted to the covering law model of explanation.11 We have argued that

9 This line of criticism applies also to a good deal of what goes under the name of Marx-
ism. There are other problems of functionalist theory (Marxist and non-Marxist), some
familiar since at least Nagel, 1961 and Hempel, 1965 scrutinized reigning Parsonian
theory. For criticism of functionalism in sociology, see Anthony Giddens, 1979, 1981.
Functionalist theory still commands attention. See Alexander, 1998; Munch 1987; and
Luhmann 1997.
10 Moore, 1966 and Anderson, 1974 take this route, but do not assume the covering law
model. Paige (1999) asserts that Skocpol, contrary to her claims to the contrary, is com-
mitted to the search for “universal causal laws” (1999: 791). But it is her commitment
to the covering law model which misleads Paige. See below.
11 For her, “the distinctive causes of the social-revolutionary situations in France, 1789,
Russia, 1917, China, 1911” (Skocpol, 1984: 154) reduce to two: if a state organization
susceptible to administrative and military collapse is subjected to intensi¬ed pressures
from developed countries abroad and there is widespread peasant revolt facilitated by
Social science and history 113

this mode of explanation cannot be sustained in any science and will not
repeat those arguments here. In any case, as a number of critics have
shown (Burawoy, 1989; Sewell, 1996), Skocpol™s use of Mill™s methods
does not allow her to produce the necessary explanatory generalizations “
“the suf¬cient distinctive causes” “ of the French, Russian and Chinese
Interpretative historical sociologists (on Skocpol™s taxonomy) eschew
causal explanation and seek what is called “a meaningful interpretation.”
This leads such writers to pay especial attention to ideas, and to the
intentions of actors. Since causal explanation is rejected, description and
explanation, which take the form of narrative, tend to collapse. Interpre-
tative historical sociologists are skeptical of the sort of theory employed
by both generalists and analytic historical sociologists, but many ¬nd
useful Weber™s conception of the ideal-type. In their comparative work,
interpretative historical sociologists are interested in individuating and in
establishing signi¬cant differences between what is compared. As Rein-
hard Bendix says:
By means of comparative analysis I want to preserve a sense of historical par-
ticularity, as far as I can, while still comparing different countries. Rather than
aim at broader generalizations and lose that sense, I ask the same or at least sim-
ilar questions of divergent materials and so leave room for divergent answers. I
want to make more transparent the divergence among structures of authority and
among the ways in which societies have responded to the challenges implicit in
the civilization accomplishments of other countries.13

We can illustrate this brie¬‚y with reference to his impressive Kings or
People (1978). Like Barrington Moore (below), Bendix is interested in
modernization, but the difference in orientation is clear from the very
¬rst pages:
It is easiest to de¬ne modernization as a breakdown of the ideal-typical traditional
order: Authority loses its sanctity, monarchy declines, hierarchical social order is
disrupted. Secular authority, rule in the name of the people, and an equalitarian
ethos are typical attributes of modern society. (Bendix, 1978: 10)

“The traditional order” is ideal-typically de¬ned in the sense that the
several features singled out are true more or less of pre-modern society,

agrarian sociopolitical structures, then there will be a social revolution. In 1789, France
was subjected to such pressures and had an agrarian social political structure. Hence
there was a social revolution in France (1984: 154). Similarly, we can substitute Russia
and China in the second premise and “explain” their revolutions.
12 See appendix B for detailed analysis of this.
13 Quoted by Theda Skocpol (1984: 370), from Reinhard Bendix, “The Mandate to Rule:
An Introduction,” Social Forces 55 (1976): 247.
114 A Realist Philosophy of Social Science

some may be present, some, sometimes absent. Ideal-types do not pre-
tend to be strict de¬nitions. Indeed, while they are not ¬ctions, nothing
really corresponds to them. Following Weber, they represent “some valid
point of view” that is culturally signi¬cant for us. In Bendix™s study, con-
siderable emphasis is put on Christianity, Hinduism and Confucianism.
This is a version of Weber, but without Weber™s fundamental concerns
with causality. Moreover, as Tilly remarks, “ordinary people disappear
from Bendix™s history, except as a breeding ground for new elites and as
a ¬eld in which those new elites sow their implicitly revolutionary ideas.”
Indeed, “the pivotal events are not alterations in the structure of produc-
tion or of power [as in Moore], but changes in prevailing ideas, beliefs,
and justi¬cations” (Tilly, 1984: 93). And these remain unexplained.
Critical problems can be anticipated with each of these strategies.
There is good reason to reject the idea of a general theory that can be
applied to historical instances,14 but there is also good reason to be clear
about one™s meta-theory “ one™s epistemological and ontological assump-
tions about inquiry in the human sciences. Thus, are there historical laws?
What explanatory role do individuals, Napoleon or George W. Bush, play
in the effort at explanation? Is culture an explanatory variable or does it
require explanation? Finally, there is good reason to insist, as Weber had
long ago argued, that attention to the particular and to identifying mean-
ing are, in the human sciences, an essential part of the causal problem.
But then we must be clear about our sense of causality.
It is possible to articulate a conception of a historical sociology which
responds to these questions, meets these desiderata, draws on the work of
Weber “ and is both coherent and plausible. Indeed, it turns out that this
is an answer to the more general problem of explaining concrete social
events or episodes. If it is a critical task of a social science to provide
explanations of events, then a good deal of sociology is historical, in the
sense that sequence is critical.15 We can suggest along the way that what
we take to be successes in the efforts of historical sociologists are suc-
cesses mainly because their work manifests, albeit unclearly and perhaps
even incoherently, the conception to be defended here. It must also be
emphasized that the issue is not that one cannot ¬nd valuable insights in
many of the major efforts in historical sociology. It is rather that these are
often accidents in the sense that they were not only not promoted by the

14 See appendix C for further discussion of this.
15 Of course, all sociology is historical in the sense that social forms are historical products.
But in addition to explaining events, there is also a sociological interest in understanding
in the sense of chapter 1. That is, as understanding in the physical sciences comes with a
theory of generative mechanisms, a theory of social mechanisms gives us understanding,
for example, of why working-class kids get working-class jobs, of capitalism or state-
building. This does not require history in the sense of historical sequence. See below.
Social science and history 115

explicit strategy of the author, but would in fact not have been there at
all had the author been clear about his or her commitments.16

A realist historical sociology
The conception to be defended here begins with two fundamental obser-
vations. First, we assume a realist conception of causality: causes are
productive powers that bring about outcomes. Second, the primary17
causal agents in history are persons. We need to consider the important
role of social structures “ as the ongoing product of activity “ but we
cannot say that social structures are causal. As virtually existing, they do
not determine action, even though, to be sure, they both constrain and
enable action. In every instance, persons must operate with “materials
at hand”, “rules and resources” in Giddens™s formulation. What powers
and capacities they have will be very much a function of the resources
available to them in acting. As Marx rightly insisted, we make history “
though not with materials of our choosing. But as C. Wright Mills (1956)
observed (probably with the Marx text in mind), “the fact is that although
we are all of us within history, we do not all possess equal powers to make
history.” For example, the decision of the US president to invade Iraq
was a decision of monumental importance exactly because it changed
the world in ways that would be quite impossible for most ordinary peo-
ple. He could make war, but wars begun cannot be undone. Options are
foreclosed and new choices are demanded.18 Nevertheless, this decision
was enabled by the actions of many persons who acted in terms of social
mechanisms they did not create.
There will then be an analogy in historical explanation to explaining
an event in the natural sciences. Consider the simplest case: we want to
explain the fact that this morning in Sam™s kitchen, a spoonful of salt
dissolved in a pot of water. We can appeal to the generalization, “salt
is water-soluble,” and to the fact that somebody put the salt into the
pot. Both implicate causes: the promissory note of a mechanism and the

16 Several writers have argued that Skocpol™s most in¬‚uential States and Social Revolutions
(1979) is an excellent example of an incoherently wrought success. Put brie¬‚y, she allows
us to believe that her use of Mill™s methods generated explanatory generalizations. In fact,
her explanations involved examining causes, understood in realist terms. In addition to
what follows, see appendix B for further discussion.
17 Again, natural events and processes very often play critical causal roles in history. Donald
McNeil Jr. recently asked, pertinently, “What follows in the wake of a tsunami? The death
of a nation? Secessionist warfare or, conversely, the unexpected drift of warring parties
toward a peace table? A surge in Islamic fundamentalism?” (New York Times, January 2,
18 There is, obviously, a parallel to biography. Some decisions are of major personal
importance: they alter “the path we are on,” foreclosing some options and opening
116 A Realist Philosophy of Social Science

effect of human action. Consider an analogous case in social science: why
didn™t Jones get the job for which she seemed to be eminently quali¬ed?
We notice that Jones is a female and that Sam is a sexist. We have an
explanation since “sexist” is functioning as promissory note.
Neither case is remarkable. Such problems as explaining the collapse
of the Twin Towers or the outcome of the American presidential elec-
tion in 2000 are more interesting. But the basic task remains the same.
One must identify and trace the ensemble of causes, both singular causal
statements regarding the eventful acts of key agents, for example, the
nineteen terrorists who commandeered the aircraft, and foreign policy
decisions of key actors in the US government. The narrative will include
an account of both past and currently existing social mechanisms, for
example, processes which generate enemies of the USA, including the
mechanism which produces terrorists,19 CIA intelligence practices and
processes of foreign policy decision-making in the USA. None of this will
be easy and much of it will be contestable. (Explaining the collapse of
the Twin Towers would require also identifying of the causal role of the
pertinent physical mechanisms, for example, the incapacity of the struc-
ture to resist the heat generated by explosions of jet fuel.) Taken together
these produced the outcome.
Time will be critical, since the sequencing of causes is essential. In the
case of the election, this nexus of causes will involve the acts of key agents
whose decisions and actions had identi¬able causal consequences, mostly
unintended, for example, the actions of the opposing candidates and their
advisors, the acts of Katherine Harris and the Supreme Court.20 But it

19 To engage a “war on terror” it is essential, of course, that the mechanism which pro-
duces terrorists be understood. Without attempting to even sketch this mechanism,
one may reasonably suppose that invading Iraq would not only not address this prob-
lem, but might, indeed, exacerbate it. Compare here the too often failure to consider
the mechanisms which produce criminals, or the failure to notice that the mechanisms
which produce youthful drug dealers are not the same as the mechanisms which produce
white-collar criminals.
20 Struggling with many of the same problems of this chapter, Marshall Sahlins (2004)
offers a distinction between two types of “structural agency,” “systemic agency” and
“conjunctural agency”. By virtue of institutional position, the acts of a systemic agent,
such as Napoleon, “are fateful whatever strategic decision he took” (2004: 158). Bobby
Thompson (whose home run in the bottom of the ninth inning won the 1951 World
Series of baseball) or Katherine Harris (whose decision as Florida Secretary of State,
powerfully in¬‚uenced the presidential victory of George W. Bush) are “conjunctural
agents.” Thompson and Harris were “circumstantially selected for [their] historic roles
by the relationships of a particular historical circumstance” (2004: 157). There may be
some utility to such distinctions as they are employed in the narrative, but as Sahlins
acknowledges, explaining outcomes requires seeing that the acts of both depend upon
capacities made available by their “positions” and that in both cases, their actions are
essential to explaining the outcome. Of course, as Sahlins notes, had either done differ-
ently, their names would have dropped out of history and this is not true of Napoleon.
Social science and history 117

will also involve the acts of theorized typical persons: rural whites in Mid-
dle America; media people; politicians; publicists and pundits; ministers;
gay and anti-gay activists; union members; and campaign ¬nancers; all
of whose beliefs, motives and situations explain their decisions. But the
social scienti¬c task does not end there: we also need an explanation of
why typical actors have the beliefs they do. Getting a handle on this will
not be easy and, as above, the account may well be contested. On the other
hand, why should we suppose that the causal story will be uncomplicated?
One needs to be reminded of the limits of meteorology and indeed, more
generally of physics “ the abstract science par excellence.
The basic outline of this view of inquiry is already present in the work
of Max Weber.21 Guenther Roth ably summarizes:
Both sociology and historiography proceed from causality inherent in human
action. When Weber de¬ned sociology as “a science concerning itself with the
interpretative understanding of social action and thereby with the causal expla-
nation of its course and consequences,” he meant to af¬rm that in history only
men act, not social organisms or rei¬ed collectivities.22 The construction of socio-
economic models, such as patrimonialism or rule by notables, is possible because,
in principle, we can understand the intentions of men and causally explain the
course and consequences of their actions. (Roth and Schlucter, 1979: 205)23

21 There are some important differences between Weber and contemporary “realisms,”
but the critical point here is that he was anti-positivist both in rejecting the Humean
analysis of causality for an account in which causes produce effects, and in rejecting the
idea that social science pursues laws which presumably explain the real concrete. See,
among many texts, “Objectivity” in Social Science and Social Policy (1949). For example,
“Where the individuality of a phenomenon is concerned, the question of causality is not
a question of laws but of concrete causal relationships; it is not a question of subsumption
of the event under some general rubric as a representative case but of its imputation as a
consequences of constellation” (Weber, 1949: 78f.). “The conclusion which follows . . . is
that an ˜objective™ analysis of cultural events, which proceeds according to the theses that
ideal of science is the reduction of empirical reality of ˜laws™ is meaningless” (1949: 80).
More generally Weber rejected the Laplacean metaphysics wherein (following Dilthey™s
critique) the “ideal goal” was “a sort of ˜astronomical knowledge™ ” (1949: 73).
While this cannot be argued here, despite positivist readings of Marx (via Engels),
Marx would agree to this. See Sayer, 1979, 1987. Similarly, while Weber would agree
that persons can (and often do) act on beliefs which are false or distorted, this is not
given the attention that Marx gives it.
22 Compare Marx: “History does nothing, it possesses no immense wealth, it wages no
battles. It is man, real living man, that does all that, that possesses and ¬ghts; history is
not a person apart, using man as a means for its own particular aims; history is nothing
but the activity of man pursuing his aims” (Marx, 1956: 125).
23 See also Fritz Ringer, 1997, 2002. Ringer offers a most useful account of Weber on causal
analysis. He focuses on Weber™s notions of “interpretation” and “adequate causation,”
for Weber, the effort to identify the change in the existing state of affairs which produced
an outcome. Verstehen is critical since persons are causal agents who act for reasons which
we need to identify: “historical agents envisage the results that they hope to achieve,
along with the means to achieve them, and that is what moves them to act. The speci¬c
characteristic of ˜this kind of cause™. . . is that we can ˜understand™ it” (Ringer, 2002:
118 A Realist Philosophy of Social Science

Roth distinguishes socio-economic models from what he terms “secular
theories.” On the present view, both are easily construed as social mech-
anisms. There is little to dispute as regards the main line of argument:
The socio-economic models as well as the secular theories are not intended to
explain what is happening in a given situation. One model alone cannot ade-
quately describe a given case: a battery of models or hyphenated types, such as a
patrimonial bureaucracy, can provide a better approximation. Their utility lies in
serving as base lines for identifying the distinctiveness of the case. While secular
theories attempt to trace a long line of causation, they too have limited useful-
ness as regards a given situation. Theories such as those of democratization and
industrialization diminish in explanatory value when we look at the relatively short
span of a few years or even two or three decades, because they are concerned with
long-range structural change. (Roth and Schlucter, 1979: 198)
The real concrete, whether it be a speci¬c exchange between two parties
or a civilizational whole, like Modern Western Capitalism, is complex
and the product of complex causes, some of very long historical genesis.
Our models are always abstract, even though, importantly, they will be of
varying degrees of abstraction. The causal account of nineteenth-century
democratization might well begin with the Greek polis, even if understand-
ing nineteenth-century democracy will require a causally linked series of
con¬gurations, each getting us closer to the real concrete which needs to
be explained.24 As Weber insisted:
Every individual constellation which it “explains” or predicts is causally explicable
only as the consequences of another equally individual constellation which has

169). Moreover, “as both Simmel and Weber showed, a singular relationship does not
and cannot be speci¬ed as a set of connections among the elementary constituents of two
successive total states . . . The logic of causal analysis does not change with the generality
of the historical developments and outcomes that are to be explained” (2002: 175). That
is, the same logic applies whether the explanandum is Modern Western Capitalism or
“the defenestration of Prague.”
Two problems with Ringer™s account may be noted. First, Ringer offers that Weber
employed counterfactual reasoning and comparative analysis, especially after 1909 in
developing causal arguments. No doubt counterfactual reasoning is useful in seeking to
identify causes, and is especially useful in getting an understanding of real possibilities
available at some time and place. Too often, actual choices and their outcomes are made
to seem inevitable. But there is no way to test a counterfactual in history, so that Ringer™s
analysis (graphically represented in diagrams) cannot be sustained. That is, one cannot
compare the actual course of events with an imagined difference. For example, while it
is clear that the rejection of the Confederation was critical as regards nation-building in
the USA, that the choices by the SPD in Germany were fundamental as regards ensuing
German history, and that Hitler™s decision to invade Russia was important as regards
the course of World War Two, we can only speculate on what might have occurred had
the US Constitution failed to be rati¬ed by the States, had the SPD not compromised
in 1918, and had Hitler not invaded Russia. See Manicas, 1989.
24 In his General Economic History (2003), of course, this is precisely what Weber sets out
to do. See Collins, 1980.
Social science and history 119

preceded it. As far back as we may go into the grey mist of the far-off past,
the reality . . . always remains equally individual, equally undeducible from laws.
(1949: 73)
Weber is no methodological individualist. Indeed, if anything, his sub-
stantive investigations (in contrast to his methodological arguments) are
very “macro-oriented” involving social mechanisms of “typical actors”
working with materials at hand. Moreover, he appeals infrequently to the
causal consequences of actions of key actors.25
Moore™s now classic account in Social Origins of Dictatorship and Democ-
racy (1966) is a ¬ne illustration. Moore writes:
To sum up as concisely as possible, we seek to understand the role of the landed
upper classes and the peasants in the bourgeois revolutions leading capitalist
democracy, the abortive bourgeois revolutions leading to fascism, and the peasant
revolutions leading to communism. The ways in which the landed upper classes
and the peasants reacted to the challenge of commercial agriculture were decisive
factors in determining the political outcome. (1966: xvii)
This is a highly abstract summary of what was theorized to be the key
mechanism “ the role and relations of lords and peasants. The analysis,
then, would aim at understanding precisely what these concretely were
and what were their outcomes. The analysis is in terms of typical actors
at particular times and places. Typical differences in their relations and
conditions give us an understanding of the problems that were set, the
beliefs which de¬ned their choices and the consequences of their actions,
intended and mostly unintended.
Moore™s work is fundamentally individualizing: as he moved closer to
the concrete, comparison led to his seeing differences in the relations of
lord and peasant in each of his cases.26 This involved bringing in details
and sometimes also seeing connections of causally related mechanisms,
for example, “market forms” versus “labor-repressive forms.” Thus, was
25 Roth suggests usefully that this is the third level of analysis, what he terms “situational
analysis,” typically in his important political writings. See especially Weber (1968), “Par-
liament and Government in Reconstructed Germany.”
26 In certainly one of the very best reviews of Moore™s book, Skocpol (1994) notes: “Moore
(rather unsystematically) elaborates and interrelates three key variables in order to explain
(a) differences among the sequences characteristic of the major Routes, and (b) differ-
ences among the ˜Bourgeois Revolution™ cases. His overall ˜explanation sketch™ seems
so unsystematic not only because he fails to de¬ne variables and spell out their roles
in explaining sequences of structures and events, but also because so much of Social
Origins is taken up with case accounts for individual countries. This fact has even led
one reviewer to assert that Moore™s method is ˜idiographic™!” (1994: 28). This is correct
and unobjectionable. The use of the term “variables” by Skocpol, so characteristic of
hard science orientations, may be excused here, except that one suspects that it was the
attractiveness of these methods which led Skocpol astray in her important States and
Social Revolutions. See appendix B, for extended discussion.
120 A Realist Philosophy of Social Science

surplus that accrued to the landed classes, derived from rents or from the
sale of produce? Moore concludes: “In comparison with their counter-
part in England during the eighteenth century, the French nobility lived
very largely from dues collected in kind or in cash from their peasants”
(1966: 41). The pertinent mechanism here involved differences in the
extent of commercialization, itself a promissory note and consequence
explained by differences in England™s early advantages in international
political economy. Similarly, what were the prevailing social relations,
the attitudes of typical actors, peasants and aristocrats?
Moore is self-conscious regarding the idea that explanation requires
identifying the actual motives and beliefs of actors. He writes: “We can-
not do without some conception of how people perceive the world and
what they do or want to do about what they see” (1966: 487).27 Thus,
in order to explain the reproduction and transformation of lord / peasant
relations in France (especially in section 5 of chapter II), Moore iden-
ti¬es the perceptions of peasants, the poorest and the less poor, as best
as his documents will allow, and offers an account of why they held the
beliefs they did. For example, “by 1789 the large majority of rural pro-
prietors did not have enough land to live on and had to work for others
or ¬nd some auxiliary trade” (1966: 71). That they expressed concern in
securing more land is easily explained. By contrast, the situation of the
richer peasants led them to acknowledge “the social position and spe-
cial privileges of the nobility,” a “fact which suggests that they could not
understand any general connection between the privileges of the nobil-
ity and their own problems” (1966: 73). Similarly, despite opinions to
the contrary, Moore offers that the data do not support the view that
the French ruling classes had beliefs which led them to forgo commerce.
But the mechanism which gave the nobility rental income had conse-
quences, including tenancy: “The best solution, at least for many, appears
to have been to throw the burden of cultivation as much as possible on
those tenants who would manage large units or, more directly on the

27 But he sees a large problem with making “ideas” causes. He continues: “To detach [the
conception of the] how people perceive the world from the way that people reach it, or to
take it out of its historical context and raise it to the status of an independent causal factor
in his own right, means that the supposedly impartial investigator succumbs to the justi-
¬cations that ruling groups generally offer for their most brutal conduct” (Moore, 1966:
487). This is, of course, Marx speaking to Hegel in the famous texts of the German Ide-
ology. Indeed, as Moore insists, “to maintain and transmit a value system, human beings
are punched, bullied, sent to jail, thrown into concentration camps, cajoled, bribed,
made into heroes, encouraged to read newspapers, stood up against the wall and shot,
and sometimes even taught sociology” (Moore, 1966: 486). Compare Bendix, above.
More generally, understanding both the reproduction and transformation of structure
requires specifying a social mechanism. This will include, probably, not only the uses of
violence, but if participants have false beliefs about their condition, why they have these
beliefs. One may wonder about the pedagogic role of too much social science?
Social science and history 121

peasant” (1966: 73). He offers that this compares neatly to the English
Tenancy, too, had consequences: “By the time of the Revolution, peas-
ants possessed close to de facto property rights” (1966: 42). While the
revolution “began with an offensive by the nobility” “ a fact also in need
of explanation “ “the three great popular upheavals” were provoked by
the sans-culottes of Paris, and “succeeded as long as it could draw on active
support from the countryside” (1966: 77). Here we need to understand
the situation and expectations of peasants in the countryside such that
they gave (or did not give) active support to the sans-culottes. Throughout
his book, Moore develops arguments about mechanisms, none as com-
plete as it might be, but mostly more than promissory notes. On the other
hand, as noted earlier, whether an explanation is satisfactory depends on
the question asked and the interests of those offered the explanation.
Moore™s explanation of the three different routes “ only hinted at here “
is a narrative (actually, each chapter offers one), but it is important to
emphasize what is intended by the use of the term “narrative.” If one
de¬nes “narrative” as “an account of some process or development as a
story, in which a series of events are depicted chronologically,”28 there
is a temptation to ignore causality or, more usually, to restrict causes to
singular causes, the speci¬c acts of speci¬able individuals. Such narra-
tives, accordingly, ignore causal mechanisms, the “materials” of action.
This is the typical historian™s approach. But as argued here, we need
both causal mechanisms and singular causes woven together chronolog-
ically in a story. Narrative in this sense is no mere chronicle, nor can it
ignore the context “ the ongoing causal mechanisms “ which enables and
constrains the decisions and actions of agents. Moore™s account gives an
understanding of three paths to three culturally signi¬cant outcomes and,
as in Weber, he makes the effort to do this in terms of “the causes of their
being historically so and not otherwise.”

History and sociology
The foregoing suggests that there are a number of tasks for sociology that
do not require history and there is at least one important task that does.
First, a great deal of very good and important sociological work is
descriptive: either qualitative or quantitative.29 A good ethnography or a

28 I follow Andrew Sayer, 1992: appendix.
29 On description, see Sayer, 1992: appendix. As he notes, “thick description need not be
seen as antithetical to theory, or synonymous with narrative. It could be the product of
a concrete research which combines and works up the insights of a range of theories
dealing with particular aspects of the object” (1992: 262). See also appendix A of the
present volume.
122 A Realist Philosophy of Social Science

good statistical study is an important achievement. But they do not offer
explanations of concrete events or episodes, nor do they give us an understanding
of the processes at work in society. One might say that a good ethnography
gives us an understanding of the system of belief and practices of a group,
but if it stops there, it does not give us an understanding of why these
particular beliefs and practices are what they are. To do this one requires
causal analysis “ an account of the mechanisms at work; and it requires
history “ an account of the genesis of those mechanisms.
This suggests another sort of inquiry “ one which can ignore history.
We may have a pattern which requires explanation, as illustrated in chap-
ter 4 with examples from a variety of writers. Understanding, as in the
physical sciences, aims at identifying the mechanism which explains a gen-
eralization, for example, a discovered correlation between drug use and
homicide, or between schooling and employment. Or we might be seeking
to understand a process or set of processes, such as economic develop-
ment, state-building or gentri¬cation. These parallel processes in the nat-
ural sciences like oxidation or growth, except that, to be sure, we cannot
assume that social processes will everywhere be the same. Or we might
look at abstract mechanisms which make intelligible a relatively bounded
system, including the reproduction of social institutions, for example,
capitalism, a health care delivery system or the international system.
Perhaps paradoxically, having identi¬ed the pertinent social mecha-
nism, one can achieve understanding without history and without appeal
to concrete agents.30 The typical actors of the mechanisms suf¬ce. While
what is to be understood is located in time and space, the passage of real
time is not relevant here. Thus, Tilly™s account of the pertinent mecha-
nisms gives us some understanding of modern state-building; but it does
not explain the genesis of the modern state in England or anywhere else.
That is not the goal of the inquiry. The concern is to identify the relevant
key mechanisms and not the particular trajectory of state-building as it
concretely occurred in a particular place. Similarly, Marx™s Capital is a
work of theory, aiming at identifying the mechanisms of capitalism and
capitalist reproduction. If we accept this account, it gives us understand-
ing, just as molecular chemistry enables our understanding of a host of
chemical outcomes. Just as molecular theory cannot of itself explain any
particular outcome, for example, a ¬re in a hotel in Las Vegas, Marx™s
Capital cannot of itself explain the success of Japanese capitalism in the
1980s “ or its more recent stagnation. In the present formulation, these
are (loosely) episodes “ the proper task of a historical sociology.

30 An early version of this argument can be found in Manicas, 1981, a review essay of
Skocpol™s States and Social Revolutions.
Social science and history 123

Again, we are close to Weber™s view of the matter as summarized by
Roth. He ¬nds in Weber three levels of historical analysis: sociological,
historical and situational, this last not of immediate concern here. Roth
The three levels are all historical in a general sense, but in Weber™s terminology
the ¬rst is that of sociology “ of type or model construction and of rules of
experience “ whereas the second level, the causal explanation of past events, is
labeled by him “historical” in quotation marks, or sometimes “developmental”
(entwicklungsgeschichtlich). (Roth and Schlucter, 1979:197)
To sum up: In Weber™s practiced methodology “sociology” is the generalized
aspect of the study of history and contrasts with the causal analysis of individual
phenomena “ the task of “history.” Both sociology and historiography proceed
from the causality inherent in social action. (Roth and Schlucter, 1979: 205)31

The generalized aspect which de¬nes the goals of sociology often takes
the form of generalizations: “When religious institutions are weakened,
there is a loss of normative control,” or “In a highly bureaucratic absolutist
state, the landed nobility has little political power.” Understanding comes
with the unpacking of these: the development of a model which explains
the generalization.32
As noted in chapter 1, the terms “understanding” and “explanation”
are often interchangeable, depending on the context. Allowing for consid-
erable arbitrariness, we can say that explanations in historical sociology
are paradigmatically of episodes (or events located in time and space)33 :
The Bolshevik Revolution; World War Two; the Great Depression; the
Civil Rights movement; a change in female participation in the labor
force; a rising crime rate; an immigrant pattern; the victory of the
Christian Democrats in the German elections of 2004. Explaining an
episode requires an understanding of the pertinent mechanisms, but since

31 In Economy and Society, Weber writes: “We have taken it for granted that sociology
seeks to formulate type concepts and generalized uniformities of empirical process. This
distinguishes it from history, which is oriented to the causal analysis and explanation of
individual actions, structures, and personalities possessing cultural signi¬cance” (1968:
vol. 1, 19).
32 This would include what Sayer calls “analysis.” “By analysis I mean the explanation of
concrete cases by the direct application of abstractions or theoretical models of what are
believed to be widely replicated structures and mechanisms. As such it tends to abstract
from particular historical sequences” (1992: 259).
33 “Event” suggests a sharply limited time-frame: the attack on the Twin Towers. An episode
is a relatively extended piece of history. Weber was quite right to see that the characteri-
zation of what is to be explained, whether it was a sharply limited “event” or a relatively
amorphous civilizational construct, was determined pragmatically, had political mean-
ing, and that a causal explanation was then called for. McAdam et al. (2001) would seem
here to follow Weber in holding, rightly, that the naming and labeling of an episode is
an interpretative, theoretical and political act.
124 A Realist Philosophy of Social Science

sequences in real time are here critical, it will also require history. Explain-
ing an episode calls for a narrative “ a history “ in which one needs to
show what actions, events and mechanisms combined sequentially to pro-
duce the outcome. Explaining episodes, like explaining the collapse of a
bridge, requires a narrative which identi¬es the causes open-systemically
at work in the world, where perhaps none is either a necessary or suf¬-
cient condition. Very much work in the social sciences is historical in this
It would be churlish to suppose that inquirers in the social sciences
do not also move sometimes somewhat uneasily between these two
paradigms. Just as much quite good social science offers us quite incom-
plete suggestions of mechanisms at work, much quite good social science
offers but sketches of causal histories. More generally, quite good social
science is dealing with problems which, as Weber insisted, were both con-
crete and terribly complex. Unlike the “pure” theorists of the physical
sciences who can deal with high abstractions, who can deal with causality
in a highly strati¬ed way, and who can subject their theories to rigorous
experimental test, this is not possible in the human sciences.34
Indeed, in the social sciences as in the natural sciences, we can very
often come to a quite satisfactory understanding of some process or sys-
tem. In none of the sciences (pace the D-N “ideal”) can we hope to ¬nd
a complete explanation of an episode since there will always be bits and
pieces in its unique causal history yet to be identi¬ed. Indeed, there is
a sense in which providing causal explanations in the human sciences is
both easier and harder. It is easier in that human action is absolutely crit-
ical to what happens in history. But it is more dif¬cult in that there are
immense (theoretical) dif¬culties in identifying the social mechanisms
which, taken together, enabled and constrained actors, a consequence of
the absence of the capacity to construct system closure: to experiment.
Similarly, the range of complexity will be a direct function of the
question asked. Thus, one can get an understanding of capitalism from
Marx™s Capital, but understanding American capitalism requires not only
an understanding of the mechanisms which de¬ne capitalism, but of a
complicated nest of other mechanisms true of contemporary American
society. One can understand capitalism without considering gender dis-
crimination or racism, but since these mechanisms function in capitalist
34 Lawson (1997) distinguishes “pure,” or “abstract” or “theoretical” explanation, from
“applied,” or “concrete” or “practical” explanation. The former task regards “the identi-
¬cation of underlying structures, powers, mechanisms and their tendencies” (1997: 220),
while the latter task “entails drawing upon antecedently established knowledge of relatively
enduring structures and mechanisms (rather than revealing them), and investigating the
manner of their joint articulation in the production of the novel event in question” (1997:
220). There is certainly no objection to his alternative formulation.
Social science and history 125

markets, they will of necessity be part of the account of American cap-
italism. Finally, explaining the global posture of American capitalism in
the past decade requires not only a grasp of the relevant mechanisms,
but a narrative of how contingent events, and the actions of key agents,
working with materials at hand, combined to produce that outcome.
In the next chapter, we turn to an examination of a family of social
mechanisms familiar to social science since at least Adam Smith. It is the
family of social mechanisms we call “markets.”
6 Markets as social mechanisms

There is a long history of theorizing markets as social mechanisms in
exactly the sense of the previous chapter. This began at least with Adam
Smith™s account in The Wealth of Nations (1776). As Smith saw, market
outcomes could be explained as the joint product of the actions of per-
sons interacting in society. It is easy enough to see that Smith constructed
his mechanism by making assumptions about persons, their beliefs, aims
and interests, about what they know, and what they can do. The out-
come of their activity was, for example, a market price. The idea was
beautifully developed in neo-classical micro-economic theory. This main-
tained the fundamental assumptions of the “classical” theory, but was
able to develop the analysis by generalizing the idea of the “marginal”
to cover both production and consumption. All of this could then be
articulated in terms of continuous variable mathematical models spec-
ifying the relationships between what are considered the key variables.
Following the assumptions of the deductivist (D-N) account of theory,
the consequences of the assumptions laid down by the theory would then
be rigorously deduced.1
But there is a paradox here of some importance. On the one hand, by
virtue of the sophistication of the models produced by economists, it is
1 Very brie¬‚y, the pertinent history is this: neo-classical theory is distinguished from classical
theory by virtue of the introduction of marginality which enabled theory to overcome the
puzzlement generated by the distinction between exchange value and use value. W. S.
Jevons, Carl Menger and Leon Walras each quite independently arrived at the main ideas
which became widely accepted with Alfred Marshall™s Principles of Economics (1890). It
is the heart of what today is called “micro-economics.” Walras, along with Pareto and
then Pigou, is generally credited with introducing into this body of theory the idea of
general equilibrium. See Schumpeter, 1954: 892“944, chapter 7. The ability to formalize
these models with mathematics was a decisive additional step, generally attributed to
Samuelson (1947), Kenneth Arrow, Gerald DeBreu and Frank Hahn. It ¬t beautifully
into the dominating empiricist notion of theory as a deductive system in which outcomes
are “explained” as entailments from the premises. For a valuable account of the contingent
facts leading to the use of mathematics in economic analysis, see Mirowski, 1991. For an
excellent account of developments from Hayek on, see Boettke, 1997.

Markets as social mechanisms 127

very often said that economics is the most advanced or scienti¬c of all the
social sciences. Most recently, a number of political scientists and some
sociologists have adopted the generalized version of this sort of model-
building under the heading of “rational choice theory” (RCT).2 But the
problem is not that markets are not social mechanisms which, if properly
modeled, could give us an understanding of outcomes by appeal to the
actions of persons “ the bogeyman of methodological individualism “ but
that the mathematical model-building of mainstream theory, encouraged
by a false idea of science, has lost nearly all touch with reality. This line
of criticism is not new. For example, in 1982, Nobel prize winner Wassily
Leontief had this to say:
Page after page of professional economic journals are ¬lled with mathematical for-
mulas leading the reader from sets of more or less plausible but entirely arbitrary
assumptions to precisely stated but irrelevant theoretical conclusions . . . Year
after year economic theorists continue to produce scores of mathematical models
and to explore in great detail their formal properties; and the econometricians
¬t algebraic functions of all possible shapes to essentially the same sets of data
without being able to advance, in any perceptible way, a systematic understanding
of the structures and the operations of a real economic system.3
These mathematical models are perfect examples of deductivist theory
construction. But if this idea of science is misconceived then these models
are, on their face, a poor choice for thinking that economics is an advanced
social science. Since a good deal of the foregoing has been a criticism of
this positivist or neo-positivist conception of science, we need not repeat
the arguments here. In what follows we concentrate on the assumptions
of the models themselves.
The account of social science (developed in chapters 3, 4 and 5), can be
brought to bear directly on the criticism that neo-classical model-building
has enormously oversimpli¬ed its analysis of markets. This problem was
identi¬ed by the earliest critics of the neo-classical model.4 They saw
it as failing to acknowledge that economic actors are social beings who
2 See appendix C. For extensive criticism of its use in political science, see Green and
Shapiro (eds.), 1996.
3 Leontief, 1982: 104, quoted by Lawson, 1977: 4.
4 The long history of criticism of the neo-classical model begins with Durkheim (see
Lukes, 1972), and in Germany with the Methodenstreit, conveniently dated from the
1893 publication of Carl Menger™s Untersuchungen uber die Method de Sozialwissenschaften
und der de Politischen Okonomie insbesondere. Weber, of course, played a key role, too
often misunderstood. One then needs to include Thorstein Veblen and a long line of
“institutionalists,” from John R. Commons to John Kenneth Galbraith to many contem-
porary “economic sociologists.” Useful anthologies of essays by representative writers
include: Etzioni and Lawrence, 1991; Granoveter and Swedburg, 1992; Swedburg,
1993; Smelzer and Swedburg, 1994; Biggart, 2002; Dobbin, 2004. See also Dugger,
1992. We exclude here any discussion of Marxist criticisms.
128 A Realist Philosophy of Social Science

are socially situated, that economic institutions, as all else in society, are
historical and social constructions, profoundly related to a host of other
institutions and that, unlike the example of celestial mechanics, processes
in time are critical. Thus, on the mainstream view, persons are conceived
as atomized individuals and as socialized exclusively as historically indif-
ferent “rational beings” who make choices in an unchanging environ-
ment. Not only do they have approximately similar motivations, but they
have more or less equal powers and capacities and they have broad scale
knowledge of the market conditions in which they act. Clearly there are a
number of problems here: CEOs of corporations, Mom and Pop Chinese
restaurateurs, heart surgeons, immigrant farm workers “ legal and illegal “
non-unionized plumbers and unionized auto workers, part-time female
salesclerks, NBA superstars, public school teachers and drug dealers (one
could go on), simply do not have the same beliefs or capacities “ either as
“producers” or as “consumers.” Nor, as importantly, are they atomized
(although one could argue that capitalism is doing the best it can to make
them so), nor do they have the knowledge and information required of
them by the theory.
In what follows, we draw on some of the now familiar criticisms of the
model. These are all well known, but in part, at least, because the model
has the authority of mainstream empiricist philosophy of science, the
economics profession has largely been content to reproduce their prac-
tices and ignore these critiques.5 In this chapter, having restored persons
as historically situated beings, it is also argued that there are different

For some exceptional doubt offered by the discipline™s most leading lights, see the AEA
Presidential Addresses of Leontief, 1971, Tobin, 1972 and Solow, 1980. Similar themes
have been expressed by other notable insiders, for example, Thurow, 1983, Balough,
1982, Hirshman, 1985 and Sen, 1977. For a variety of critical analyses, see also the Pro-
gressive Economics Forum (www.web.ca/∼pef).
Business school professionals are also critical as regards the usefulness of micro-
economic models for business decision-making. See Oxen¬eld (ed.), 1963. For example,
“market models admit time considerations only in a limited and contrived manner . . . But
investment represents the concern of major executives, rather than clerks, for the very rea-
son that markets are dynamic and are buffeted by many forces that vary over time . . . In
other words, executives who are estimating the pattern of revenues and costs over the
life of an investment “ and the length of its life “ get relatively little help from market
models of price theory” (Oxenfeld (ed.), 1963: 63). See also Lazonick, 1991 and Hayek™s
critique, below.
5 For most economists, the model is justi¬ed in terms of its putative predictive value or as
a “useful” approximation of concrete reality. See below. There is no argument as to the
usefulness of these models for ideological purposes. See Stiglitz, 2002, and for different
sort of “usefulness,” see Davis 2004. Davis writes that “a majority of AEA members”
who responded to a survey he conducted, admitted, “at least privately, that academic
research mainly bene¬ts academic researchers who use it to advance their own careers
and that journal articles have little impact on our understanding of the real world and the
practice of public policy” (2004: 359).
Markets as social mechanisms 129

kinds of markets which differ in fundamental ways, and that once the
effort is made to construct models which are closer to reality, it becomes
clear that, while these models must sacri¬ce the elegance of mathematical
models, they can aid in understanding market processes. But explaining
most outcomes, for example, current unemployment, requires engaging
history “ just as with the explanation of a terrorist attack or a war.
The problem, it must be emphasized, is not that abstraction, simpli-
fying assumptions and model-building are inappropriate for a human
science (Boettke, 1997: 13). As argued in chapter 1, to understand the
concrete one must offer abstractions from it: we understand why iron
rusts because abstraction has yielded a representation of the causal pow-
ers of the theoretical (but real) entity, Fe. Similarly, as regards the social
mechanisms which give us an understanding of our social world “ includ-
ing, then, the focus of this chapter: the mechanism of markets.

The neo-classical model of the market
The basic ontology of neo-classical theory postulates rational individu-
als engaged in interaction, either as consumers or producers (¬rms) (see
appendix D). Their attributes and situations are formally de¬ned in terms
of a familiar set of postulates. For example, if an agent prefers x to y and
y to z, then she prefers x to z: revealed preferences are transitive. The
general equilibrium model adds a number of further assumptions: that
there is perfect information available to all parties, that there are many
buyers and sellers in every market, that each may enter and leave easily,
that everyone has the relevant information, that there is an interdepen-
dence among the many markets, that commodities (including labor) are
in¬nitely divisible, etc.
The majority of economists would seem to agree that most of these
assumptions, at least without severe quali¬cations, are false.6 People are
not rational in the relevant sense, decisions about preferences are not
made pairwise, commodities are not indifferently substitutable in all situ-
ations,7 ¬rms do not always maximize pro¬ts, transaction costs are totally

6 There are important writers who take the very heroic stand of insisting that the model
suf¬ciently well approximates reality to be a valid description of it. These include the
so-called Chicago School economists, Milton Friedman, George Stigler, Gary Becker
and Robert Lucas. See Boettke, 1997.
7 That is, it is not true (as asserted by, for example, Debreu, 1984) that “commodity space
has the structure of a real vector space”“ a critical assumption for formalization. Mirowski
(1991) relates the story of the shepherd who agreed to accept two sticks of tobacco for one
sheep but became confused when given four sticks for a second sheep. For the economist,
this shows that the shepherd does not understand arithmetic. But indeed, it shows that
the economist does not know sheep!
130 A Realist Philosophy of Social Science

ignored,8 information is never anywhere near perfect, etc. Moreover, the
formalized theory leaves no room for the effects of the passage of time
and the fact that, in the real world, the economic situation is continually
changing because of decisions made by ordinary people, corporate heads,
managers and government of¬cials “ from heads of state to the secretary
of the treasury.

Starting from scratch
One could argue that even if the assumptions of the mechanism are not
true, then as Weber suggested, the model can nonetheless provide a use-
ful heuristic. But if the goal is to understand or explain outcomes, then it is
hard to see how the model can do this. Understanding, like explaining out-
comes, requires that the assumptions of the mechanism be more or less
true. Thus, for example, the equilibrium price of a commodity is that
price exactly because in the theorized causal mechanism, individuals are
making decisions in accordance with the assumptions of the model. If,
indeed, this is not true, then even if the model does has predictive capac-
ities, the outcomes are not explained.9 We must then ask, can we build
better models, even if, perhaps inevitably, they will lack the rigor provided
by formalization?
We should, at the outset, acknowledge that the neo-classical theory
may illuminate outcomes in some markets. We spoke earlier of Chinese
restaurateurs. In most cities in the USA, they are small ¬rms in a gen-
uinely competitive environment. Operators know what their costs and
their sales are and must make decisions which seek to produce pro¬ts.
While it is contestable that they think in terms of opportunity costs and
have the concept of marginal productivity, it is not implausible to hold
that these theoretical ideas help us to understand their actual behavior
and thus, assuming that their customers are seeking quality at the best
price, the actual prices in Honolulu of items on the menus of Chinese
restaurants.10 But even if this is true, this environment is not typical.

8 R. H. Coase (1995) has famously argued that the existence of transaction costs “implies
that methods of coordination alternative to the market, which are themselves costly and
in various ways imperfect, may nonetheless be preferable to relying on the pricing mech-
anism, the only method of co-ordination normally analysed by economists” (1995: 8).
This was Coase™s 1991 Nobel Laureate Address. Transaction costs, for example, con-
tracts to be drawn up, inspections to be made, arrangements to settle disputes, processing
costs, are the least of it. See below.
9 It is highly contestable whether neo-classical theory survives the test of “good predic-
tions.” Fundamentally at issue is the quality of the “empirical” tests. See appendix A.
10 We say “may illuminate” outcomes since we need to say here that if operators do not
think in terms of opportunity costs as de¬ned by the theory, they are thinking something
quite similar. As Schutz insisted, to explain action one must identify actual beliefs and
motivations, not ones imposed on actors by social scientists.
Markets as social mechanisms 131

Put aside for the moment the very different conditions and behavior of
CEOs of large corporations,11 and consider here Charles Smith™s impor-
tant work on auctions (1989). Auctions are an important kind of market,
not least because they appear to illustrate the general equilibrium model.
Smith offers a rough classi¬cation of kinds of markets. Auctions must
be distinguished from “¬xed-price” exchanges in which buyers confront
prices which have been established and are “¬xed” in the sense that they
are stable over substantial periods of time. This is certainly the most typ-
ical sort of market. Finally, there are “private treaty” forms of market
exchange in which buyer and seller “actively negotiate the price between
them” (Smith, 1989: 15). Generalized in terms of supply and demand
curves (appendix D), the private treaty form is the conceptual ideal for
neo-classical theory. Where buyers and sellers “actively negotiate the
price between them,” informational problems may be solved, and if a
deal can be made, since the actors are rational, no one gets cheated. We
consider below “¬xed-price” markets and argue that there are several
ways that the prices confronting buyers are ¬xed.
Not only are ¬xed-price markets, private treaty exchanges and auctions
very different kinds of markets, but there are also several kinds of auc-
tions. Critically, there are important differences in the social mechanisms
of these variant forms “ differences which need to be established empiri-
cally. Smith groups auctions into three subtypes: “commodity/exchange,”
“collectable/dealer,” and “art/one-of-a-kind.” Among the features which
make for differences between these and other kinds of markets are the
importance in determining price of a wide variety of factors; the impor-
tance of historically stable practices and changes in these; the “rules”
which are constitutive of the process; collective consensus regarding value
or the absence of such consensus; uncertainties regarding costs; differ-
ences in individual taste and judgement; and even “the will to possess.”
Consider, for example, two instances of bizarre price ¬‚uctuations.
In 1985, the all-time record for a thoroughbred yearling was set at
$13,100,000 for one horse while in the next two years the highest price
paid was $3,600,000 and $3,700,000, respectively. In the recent past, the
Dow Jones average went from 800 to 2,700 in about ¬ve years, only to
drop 1,000 points within a week. In neither case had these changes much
to do with changes in variables analyzed by the neo-classical model. The
¬rst is explained in terms of “the will to possess,” and the head-to-head
competition between a handful of buyers, the sheikhs from Dubai and
Robert Sangster (Smith, 1989: 192). By 1986, they were no longer in
11 The literature is large. But see, for example, Chandler, 1962; Berle and Means, 1968;
Galbraith, 1968; Barnet and Muller, 1974; Lazonik, 1991; Dugger, 1992; Geneen, 1984;
Bakan, 2004. See also the discussion of “imperfect competition,” below.
132 A Realist Philosophy of Social Science

The stock market ¬‚uctuations had no such competition, and more
importantly, not much to do with changes in the shape of the economy
or the “real” value of stocks. Rather, the introduction of new computer-
based trading programs that made use of ¬nancial futures and option
contracts enormously complicated trading practices and opportunities
such that key “decisions” were made, in fact, by computers programmed
to respond to speci¬c indicators, such as capital ¬‚ows. None of this
would have been possible, even given the technology, if the rules gov-
erning the stock market did not establish the rights required for trading
¬nancial futures, option contracts and “derivatives.” We will return to
this extremely important point.12
The point to be emphasized here, however, is that in each of these
variant forms of markets, there are speci¬c social relations, rules and
practices, mostly unarticulated, which constitute and legitimate the con-
ditions of the exchange, and which, accordingly, enable and constrain the
participants and, as such, the results. While, as noted, the conditions set
out by neo-classical theory are sometimes suf¬ciently close to concrete
reality to be illuminating, this is not the case in general.

De¬ning a market
It is profoundly paradoxical that while it is the central concept of modern
economics, the concept of a market is either taken for granted, unana-
lyzed, or more likely, de¬ned, tautologically, in terms of the theory such
that, if the conditions set out by theory are not satis¬ed, then there is
no market! As Dyke noted, if, for example, “the market must ful¬ll the
condition of the independence of irrelevant alternatives, then a market
hardly ever exists” (1981: 116) (see appendix D). But plainly, explaining
markets requires that our theory confront actual markets. Although the
concept of a market is a high abstraction, we need a workable de¬nition.
There is a widely available intuitive sense of a market: a place where
there are many sellers of various goods, either of one sort or of many dif-
ferent sorts, and there are many shoppers. One thinks of Covent Garden
in London, the Mercado Centrales in Managua, even the Swap Meet in
Aiea, O™ahu. Markets of this sort have existed for a very long time, and
almost everywhere. Sadly, this image is anything but helpful when one
talks about industrial markets, labor markets, the stock exchange, or even
Keeneland Thoroughbred horse auctions.
A standard sense (to be rejected here) de¬nes markets as:

12 For detailed description of the social process which produced equity option markets and
sponsored word / phrase Internet search engine markets, see Smith (forthcoming).
Markets as social mechanisms 133

A social institution in which people freely exchange commodities (goods,
resources, services) generally through the medium of money.13
The de¬nition trades on the intuitive image, but there are two problems.
One regards the role of money, the other the idea that “people freely
exchange commodities.” What can “free” mean here?

“Free exchange”
The weakest (and most widely held sense) would seem to be that there
is no coercion in the sense of legitimate or illegitimate threats of force.
Giving up one™s wallet to a gunman is not a market exchange. But what of
state-enforced requirements on a minimum wage, an eight-hour day or
regulatory agencies and anti-trust laws? These are surely legally enforce-
able constraints on exchange. The ideological point is clear enough: The
myth would have it that there is a clear separation of state and econ-
omy. Markets are presumed to be autonomous institutions, ideally uncon-
strained by state action. But this idea leads to total conceptual confusion
since in all modern societies, the state is essential to the very constitution
of markets.
To take one obvious but much overlooked example, property rights
are surely critical as regards exchange. Indeed, thoroughly undermining
the image of a market with which we began, Coase (1995) argues that
rights to perform certain actions are what is traded. It is hard to overstate
the importance of this in a world where actions by economic actors can
have monumental consequences as regards health and safety and the
environment, and where ¬nancial markets have taken on extraordinary
forms. He concludes: “As a result, the legal system will have a profound
effect on the working of the economic system and may in certain respects
be said to control it” (1995: 11).
The question, then, is not whether the state must act in constituting
markets; the question rather is, what is the character and what are the
consequences of widely varying forms of that constitution, of who bene¬ts
and who (and what) does not? For many people today a “free market” is a
market constituted so that entrepreneurial actors are not hindered by laws
or regulations aimed to protect employees, consumers, the environment,
or public goods not provided by the market. One might argue that such
a market is desirable, perhaps because it is ef¬cient. But indeed, even if
this were true “ and it is not dif¬cult to show that the idea is fatally ¬‚awed
(appendix D) “ there are a host of hard-won legal constraints, for example
13 Outhwaite and Bottomore, 1992: 359. The text is paraphrased from the authoritative
and useful Blackwell Dictionary of Twentieth-Century Social Thought.
134 A Realist Philosophy of Social Science

on child labor and the length of the working day, which no reasonable
person would want to see repealed.


. 4
( 7)