Weingast B.R., Wittman D. The Oxford Handbook of Political Economy
Подождите немного. Документ загружается.
972 questions about a paradox
because the situation in which they would collide doesn’t arise. If it did arise, the
different implications would become apparent.
A related implication of cognitive limits is that computation takes time and effort.
This implies that we don’t really make complete plans or sets of decisions for the
future. Indeed, it is not cognitively possible. Once you grant that, each individual
knows that she will make decisions in the future. All decisions today are partial plans.
These remarks have an implication for the paradox of voting. Even if cycles exist,
they may not manifest themselves. Either they do not exist (or at least are not at
the top) for the given preference profile or an agenda-setting or electoral mechanism
yields an outcome without revealing the cycle.
This idea will bear on several of the points below.
3 Questions and Answers:Impossibility
Theorems and Social Choice Theory
.............................................................................
[1] If non-dictatorial social choice mechanisms are more prone to indecision (cycling)
and poor decision-making (chaos) than dictatorial systems, why are democracies so
commonplace these days?
One aspect of political decision-making that I didn’t consider in my approach to
social choice theory is that the political process includes communication, not just
decision-making. Information is critical. Hence the recent emphasis on deliberative
democracy is important. Yet, it seems clear that even after an intense deliberative
phase, irreconcilable differences will remain. Indeed, deliberation may make differ-
ences of opinion clearer.
Consider dictatorships. My work implies that dictators should be efficient at run-
ning the social choice mechanism, if largely for themselves. What this misses is that
dictatorships also tend to be poorer on communication.
Further, a real-life dictator has to involve others and take their interests into
account, so that social choice theory is not irrelevant even in so-called dictatorships.
Finally, I would not want to assume that the current prevalence of democracy is
permanent. Many democracies have ceased to be. Indeed, several, such as Germany
in 1933, have made a conscious choice to set aside democracy. Venezuela has made
a somewhat similar move, and other Latin American democracies are very shaky.
Even in the stablest democracies, the United States and several countries of Europe,
there is a mismatch between the issues in the political arena and the actual concerns
of the voters. The rejection of the European Union constitution in France and the
Netherlands, against the unanimous views of the political elites, is one example. In
the United States, observers have noted that the partisan divisions in elections and
kenneth j. arrow 973
in the legislatures are more extreme than the actual views of the electorates. One
symptom of the disconnect is low participation in voting.
[2] In light of the findings of social choice, a field launched by your work, what care
must we exercise in speaking about collectivities? What constraints must we observe in
our discourse about collective preferences, such as the “public interest?”
We must take care when considering collectivities, but I believe they can be mean-
ingful under certain circumstances. Take the notion that “something is in the United
States’ interest.” For example, it’s in the country’s interest to lower tariff barriers even
though some are worse off.Efficiency is a good place to start, but alone is not adequate
because of distributional arguments. For any policy, we should also ask something like
what happens for the seventy-fifth percentile worker. Hence I would like to see more
attentiontotheloserswhenweevaluatepolicy.
[3] When did the analogy between collective choice processes and individual multi-
criterion choice processes occur to you and how do you view that analogy today?
The analogy occurred to me right away. I saw that choice under uncertainty, with
comparisons of different states, was very much like social choice theory, though with
some differences. The analogy was even clearer before L. J. Savage made Bayesianism
respectable. In the Neyman–Pearson and Wald minimax approaches, the different
states were not really comparable. In the Savage reduction of statistical inference
to decision theory, there is interstate comparability of utility. The analogy was the
inspiration behind my work with Hervé Reyaud, Social Choice and Multicriteria
Decisionmaking (1986).
[4] What lessons should constitution-writers seek in the theory of social choice and
mechanism design?
I believe that constitutions should provide some way of ensuring decisiveness, even
at the expense of transitivity. From time to time society faces a compelling problem
that must be addressed. Sometimes, benign neglect is fine, but at other times it is not.
At these times, it is important to have a clear, impartial, and acceptable view of both
the status quo (what policies are currently in place, so that you know when you’re
changing things) and the range of possible alternative policies. The uncertainty over
the results of the 2000 election provides an example of when the United States needed
a decisive answer.
[5] Your general possibility theorem is, indeed, general (not in the sense of the use
in the title of the theorem, but in the sense of broadly applicable). Its application,
however, has tended to focus on democratic institutions in particular. One might
infer that the theorem is of central concern for democratic institutions while its
application to economic or social institutions has more of an “interesting intellec-
tual curiosity” flavor. Is this how you see the importance of the general possibility
theorem, of demonstrating a deep series of concerns about political institutions but
of logical relevance though lesser importance for other institutional settings? Or do
974 questions about a paradox
you see it as revealing something profound across a larger array of institutional
settings?
Let me concentrate on the question of whether the impossibility theorem is as
significant in interpreting economic outcomes as it is with politics. In the usual
formulations, economies tend toward an equilibrium among a variety of preferences
and technologies. Markets do not compel people toward uniformity of behavior.
Indeed, economics helps us understand how differences among people persist and
are compatible. Politics is different because of power, especially the coercion implied
in having one law for everyone.
The more specific question is, does the impossibility theorem apply to markets, the
general equilibrium model in particular? I think this is something of an open prob-
lem. General equilibrium defines one way of choosing among social states. As already
suggested, there are differences in the set-up between economic and political models.
Nonetheless, I believe an aspect of general equilibrium has the same impossibility
problem. Consider the question, is it better to adopt a new technology, specifically, a
technology that involves trade-offs, so that something gained, but something else is
lost? If we set up the problem right, then the impossibility theorem applies. But this
set-up is somewhat artificial, i.e. the context may not be so natural as it seems in pol-
itics. In contrast to the above assumption about technology, some new technologies
instead push out the frontier without involving a trade-off. In this circumstance, the
choice is between a set and a bigger set, so no paradox arises.
[6] Was social choice theory, as you originally conceived of it, intended to be positive
(i.e. predictive)?
Originally, I saw it as a normative program. Yet as I’ve lived with it, more and more
I think of it as descriptive. In Social Choice and Individual Values (1951), I cited
proposals for federal support of education, which had clear support of majorities.
Yet the issue of desegregation was used to defeat this.
Similarly, Frank Easterbrook (1982) and Maxwell Stearns (2000) have applied this
analysis to the Supreme Court. When you look at the reasons behind the decisions,
differentmajoritiesrelyondifferent and contradictory logic, even when individual
justices are consistent.
4 Questions and Answers:Economics
and Political Economy
.............................................................................
[1] Do you believe economic efficiency can be a meaningful and useful concept (and
whydoyoubelievewhatyoudo)?
kenneth j. arrow 975
Like pornography, I know it when I see it. Measures of economic efficiency suffer
from the social choice problem. Nevertheless, it is better to consider efficiency than
not. Take benefit–cost analysis as an example of applying the efficiency criterion. It
may not be utterly conclusive, but it provides an important piece of information for
public decision-making. For example, going back to the 1950s, benefit–cost analysis
has been used in water resource decisions. Its use has not prevented projects with
benefit–cost ratios of less than one. But those with ratios of .3 are much harder to do
even with political pressures.
[2]Inyour1963 American Economic Review article on moral hazard in health
care provision (Arrow 1963), you stressed the importance of trust—i.e. ongoing
relationships—as a means around our inability to contract fully for all possible out-
comes. Might trust play a similar role in a group’s allocative choices, such as those we
find in committees and legislatures? Might trust work in small groups but not in large
groups (à la Olson 1965)?
In group choice, trust relates to the deliberative phase, not the choice phase. For
example, one member of the group tells the others something, and they believe him
even though they have different interests. Trust implies that members count on the
others not to mislead them. This parallels the question of cooperation.
In my paper on the medical profession, I argued that the professions were designed
to signal trust. An interesting empirical question is whether this form of trust has
broken down—many of the factors I listed in that paper as engendering trust may no
longer hold. (1) The proportion of for-profit hospitals among all hospitals has risen
sharply. (2) Advertising of drugs and even of medical services has become prominent,
where it was previously virtually non-existent. (3) The rise of managed care means
that the institutions that deliver medical treatment no longer give primacy to the
individual physician.
[3] What does economics have to say about intrapersonal and interpersonal mecha-
nisms of preference formation?
The short answer is nothing, though we should say something. There has been some
formalization. Economics can learn.
A considerable literature exists on interpersonal effects, such as peer effects. For
example, Glaeser, Sacerdote, and Scheinkman (1996) have written about peer ef-
fects in criminality. Our usual explanatory variables, such as poverty or education,
explain very little of the variability in crime across metropolitan areas and even
less of that across nations. Glaeser et al. argue that crime is a game with multi-
ple equilibria, a form of coordination on preferences, not behavior. People have a
desire to conform. There are two equilibria: one in which everyone is a criminal,
the other in which no one is. This model appears to explain a lot of variance. Yet
these and related models are very simple; the cost of departing from the local norm
is a concave quadratic function. Yet they have considerable statistical explanatory
power.
976 questions about a paradox
I’m persuaded that there’s something important about peer effects, yet important
difficulties remain. The simplest version of the peer effect hypothesis—“I want to be
like others”—is too naive. A useful formulation has yet to be found. Moreover, how
important is context? Clearly peer effects are not universal. A parallel set of models
have been studied to deal with the diffusion of innovations: the models frequently
cause an innovation to be adopted if it is being adopted by others without regard to
whether the innovation is good or bad. In a similar fashion, the adoption of a locally
current norm may depend on its interaction with other elements of the individual’s
preferences and opportunities.
[4] Does economics inevitably require a political foundation? Specifically, the econ-
omy depends on property rights system, legal regimes, protection against expropria-
tion, a stable macroeconomy, and a host of other regulatory policies. All these decisions
have a direct effect on the economy and economic development; and all are endoge-
nous to the political system. Does this imply that a mature theory of economics and
economic performance requires that economics gain a political foundation?
The premiss in this question is obviously correct. But we should also mention the
legal and social foundations of economics along with the political ones. Each of these
four realms—economic, political, legal, and social—has a different set of institutions
that interact, but they are not the same. Just as economic problems can affect the
political agenda, so too can social problems affect economics and politics. Under
social foundations, I include groups and their effects, including social norms; and
generally, how people behave when they interact.
This issue arises in the study of economic development, where a major question is,
do institutions matter? And, in a collateral version of the question, maybe institutions
matter, but they are endogenous. Describing institutions is not very easy. Commodi-
ties are easier to describe, but are still hard. Institutions are more like services, where
so many difficult questions arise. For example, is a physician-hour the same now as
fifty years ago?
Moreover, the study of institutions faces significant problems of measurement and
empirical testing. For example, we expect the effects of many institutions to have very
long lags. This implies that it is hard to use cross-country data to study institutions.
The latter have yielded a very intensive study of the last thirty years of economic ex-
perience, but data limitations mean we cannot study longer periods in the same way.
The effect of the economy on the political system is equally worth talking about.
For example, I went to the former Soviet Union in 1991. One of the questions we faced
was, should we recommend that they adopt contract law and accounting systems?
I wondered, do we have to recommend that? It seems so straightforward. And yet,
the Soviets did not have either of these things. On the empirical side, the Russians
subsequently did begin to provide these while the Chinese have not. Yet which is
doing better—and why?
Returning to the question: the trouble is, we have a relatively strong theory of
economics, but our theories of politics are weaker, and so adding a theory of politics
to economics under the current circumstances only weakens it. Similarly, in principle
kenneth j. arrow 977
we should also add to economics very important social components. But theory in
that area is even worse than in politics.
Next let me raise a point about the influence of psychology on economics, which
has been quite substantial in recent years. In many ways, research in psychology is
more rigorous than in politics and sociology. So, even though psychology is not as
important for economics as politics and social interaction, economists have found
the theory and methodology much more useful.
[5] Do you have any thoughts on the striking parallels and contrasts between the
1950–1970s economics and social choice/theoretical political economy? In general
equilibrium theory you and others showed that, under general conditions without any
reference to characteristics of specific markets, a general equilibrium exists. The two
fundamental theorems of welfare economics also contributed to the view that economic
systems have nice properties.
Things worked out differently in theoretical political economy. First, your impossi-
bility theorem suggested that at a general level systems with rational individuals were
not likely to exhibit social rationality. Second, in parallel with general equilibrium,
students of majority rule sought in the 1960sand1970s to study the equilibrium
properties of majority rule and ended up proving the opposite of general equilibrium
in markets—that in general majority rule lacks any equilibrium.
Theory demonstrates real differences between politics and economics. The theory
of economics has nice theorems with good properties. The competitive context, for
example, produces automatic pressures for efficiency. But as soon as you have large
aggregations of power—whether in economics or politics—analysis becomes diffi-
cult. Moreover, assumptions such as self-interest play a bigger role; and departures
from self-interest become more important.
In politics, the absence of the existence of a political equilibrium makes theorizing
very inconvenient. The use of single-dimensional models to study politics is useful,
but problematic.
5 Questions and Answers:
Methodological and Miscellaneous
.............................................................................
[1] What is the big question in political economy that you would advise young re-
searchers to pursue?
There are lots of questions to which we need answers. How do individuals in the
political system coalesce to form groups, such as pressure groups? How do these
groups form? Second, once formed, how do they work within the system to have
978 questionsaboutaparadox
influence? A specific instance of this has always concerned me: why is the political
influence of farmers inversely proportion to their numbers?
[2] What are the prospects and limitations for a more behavioral (or boundedly
rational) approach to the study of political economy?
We now have real concrete work on bounded rationality. But there is a trouble with
theories of bounded rationality: they are very difficult to use. Moreover, some of the
behavioral anomalies arise in experiments, but the situations in which they arise don’t
occur in real life. Consider, for example, anchoring. An individual is asked to estimate
a magnitude after being given a clearly irrelevant observation; the observation has a
distinct effect. But this is an experimental paradigm, not one likely to occur naturally.
This means that some of the problems emphasized in the literature are not surprising.
But other behavioral problems and anomalies are obviously important. Consider
the issue of framing, especially important in politics. But how does it work? We have
no theory. An obvious place to start: does the frame that is ultimately chosen reflect
mere chance? Are some people much better at creating frames? Might an evolutionary
theory of survival of frames be appropriate?
[3] With the publication and wide dissemination of results, such as your general
possibility theorem, many scholars in economics and political science opine that co-
operative game theory, and particularly social choice theory, is now a “dead field.”
The implication is that we will learn more by focusing our theoretical energies on
non-cooperative game theory, particularly in its newer behavioral and computational
guises. Do you agree with this characterization of cooperative game theory and, in
particular, of social choice theory?
Non-cooperative game theory is individually based and elegant. Nevertheless, it does
not provide an adequate or full theory. Consider, for example, the experimental evi-
dence that people cooperate far more than predicted by prisoner’s dilemma models.
As soon as the prisoner’s dilemma is repeated in an experimental setting, even a little,
people tend to cooperate. People can see collective rationality even though the only
non-cooperative equilibrium in a finitely repeated prisoner’s dilemma game is the
one-shot defect-defect equilibrium. We also see this type of cooperation in real life.
We don’t yet have a good theory of this type of cooperation.
Take the Hobbes/Coasian view: people will bargain to a more efficient agreement.
Many theoretical impediments arise to efficient bargaining: holdouts and an empty
core. I find all sorts of reasons that collective rationality is not compelling. Yet peo-
ple do seem to cooperate. For example, competitors in an industry form research
alliances even when they compete intensely. As Brandenburger and Nalebuff (1998)
emphasize, you get competition and cooperation at the same time. We have no good
formalization of this process. Yet it is clearly important. Cooperation and conflict in
the political world should be the same.
To summarize, it’s clear that the cooperative phenomenon is there. We can’t aban-
don the problem of cooperative theory even though we have no good theory.
kenneth j. arrow 979
[4] Do you see any significant costs in the mathematization of political economy that
occurred over the course of your career?
Resources are scarce, so devoting attention to mathematization means less time de-
voted to something else, like data collection. We can never really know what these
costs are. Hence I feel we should let people make their own decisions in a decentral-
ized way. Everything has costs, but we have no way of making an ex ante prescription.
[5] Howcanwebemoreeffective in bringing the implications of our research to the
attention of policy-makers to help them make better-informed policy choices?
I’m the last one to answer that question! I just finished making a policy recommenda-
tion about the financing of new malaria-reducing drugs. So far, it’s been ineffective.
The report took two years of hard work, and it’s not obvious that anything will
happen.
6 Concluding Remark
.............................................................................
I may spend the rest of my life working on these problems.
References
Arrow,K.J.1951. Social Choice and Individual Values. New Haven, Conn.: Yale University Press
(2nd edn. 1963).
1963. Uncertainty and the welfare economics of medical care. American Economic Review,
53: 941–73.
forthcoming. The subject-matter of social choice theory. In The Handbook of Social
Choice and Welfare,vol.ii,ed.K.J.Arrow,A.K.Sen,andK.Suzumura.Amsterdam:Elsevier.
and Reynaud,H.1986. Social Choice and Multicriteria Decisionmaking.Cambridge,
Mass.: MIT Press.
Brandenburger,A.M.,andNalebuff,B.J.1998. Co-opetition.NewYork:Doubleday.
Easterbrook,F.1982. Ways of criticizing the court. Harvard Law Review, 95: 802–32.
Glaeser, E. L., Sacerdote,B.,andScheinkman,J.1996. Crime and social interactions.
Quarterly Journal of Economics, 111: 507–48.
Olson,M.1965. Logic of Collective Action. Cambridge, Mass.: Harvard University Press.
Stearns,M.2000. Constitutional Process: A Social Choice Analysis of Supreme Court Decision-
making. Ann Arbor: University of Michigan Press.
chapter 55
..............................................................
POLITICS AND
SCIENTIFIC
ENQUIRY
RETROSPECTIVE ON A
HALF-CENTURY
..............................................................
james m. buchanan
1 Introduction
.............................................................................
As the new century moves forward, the ranks of those who can personally re-
call the state of play in political analysis at the middle of the last century are
rapidly thinning. And without personal recollection, which must itself sometimes
be forcefully imposed on what seems best forgotten, it remains impossible to ap-
preciate the depth of confusion that described the then-prevailing “science” of
politics. To modern scholars who must work within the constraints of published
works, the glaring failure of social scientists to dispel the fog, or even to recog-
nize that the fog exists, seems both inexplicable and inexcusable. They should
recognize, however, that while with modern tools of analysis the fog may seem
to lift, thereby offering more satisfying explanatory results, many of the sources
of the basic confusion have not been fully displaced. Potential value has been
extracted from the “science of politics” as it has developed over a half-century, but
this “science” itself has generated its own critics, perhaps an indirect measure of
some success.
james m. buchanan 981
I entitled an early paper “An economist’s approach to ‘scientific politics’ ”
(Buchanan 1968). The two parts of this title, properly understood, can be used to
describe, in summary, what has happened to scholarly discourse on politics over
the last half of the twentieth century. The study of politics became “scientific” in
the authentic meaning of this term. At the same time, the economists’ approach—
incorporating methodological individualism and rational choice—came to dominate
explanatory enquiry.
Science is systematic enquiry into that which is. Two thresholds must be crossed
before any such enquiry can make progress: first, agreement must emerge on the
object of enquiry, on the categorical definition of “that which is;” second, “that
which is” must be separated from “that which ought to be.” Interrelated failures
along both these dimensions of enquiry have continued to be major barriers to
understanding.
2 What IstheState?
.............................................................................
In the paper referenced above, I recounted the parable of the blind men and the
elephant. If they agree that their task is to come to a composite understanding
of an elephant, these blind persons (men and women) can, by using their active
senses of touch, smell, and hearing and by comparing notes, reach an adequate and
comprehensive result. Suppose, however, that some members of the group think that
they are aiming at understanding the structure of an ostrich. In this case, the result
will be incoherent.
Continuing confusion as to what politics is, as a potential subject for systematic
enquiry, has plagued discussion through the ages, and, by the nature of the case,
must continue to make analysis difficult. To the economist, and perhaps to most
modern social scientists, the initial concentration of attention must be placed on
the behavior of the persons who make political-collective choices for the organized
polity. Even such an innocuous starting point bypasses, however, the whole tradition
of enquiry that commences with the state, or collectivity, as such, treated as an organic
and monolithic unity independent of those persons who occupy agency or subject
roles.
This organic model, or rather conceptualization, cannot be so readily dismissed as
economists might be wont to do. “The state,” as an acting unit, does exist, if for no
other reason than that it is treated as such in common parlance. And, by necessity,
the isolated individual, as subject, confronts this entity in its various manifestations.
Regardless of how its internal decision processes are organized and the results are
generated, the collectivity, as such, becomes a reality beyond the behavioral control
of the individual, who is reduced to parametric reaction only.
It is plausible, therefore, for the putative scientist to commence enquiry with the
collective unit, as such, and to examine how states have acted throughout history;