By Michael Rozeff
Ludwig von Mises Institue
September 6, 2006


Joseph Stiglitz shared the Nobel Prize in 2001 partly on the
basis of an important paper of his (with Greenwald):
"Externalities in Economies with Imperfect Information and
Incomplete Markets." In that paper he says: "There exist
government interventions (e.g., taxes and subsidies) that can
make everyone better off."

Stiglitz is a prolific, outspoken, and outstanding spokesman
for the pro-government school. Stiglitz sees market
imperfections that are remediable by government everywhere he
looks; and this paper is supposed to provide the intellectual
and analytical foundation for government intervention. I will
argue that this important and oft-cited paper completely fails
to prove the potential worth of government interventions.
Murray Rothbard's views directly oppose those of Stiglitz. In
his paper "Toward a Reconstruction of Utility Theory and
Welfare Economics," Rothbard writes: "We conclude therefore
that no government interference with exchanges can ever
increase social utility," (his emphasis).

Whose view is correct is tremendously important as it touches
upon the entire realm of political organization. If Stiglitz's
pro-state view is correct, it provides justification for the
interventionist state. If Rothbard's anti-state view is
correct, then it removes the justification for all government
activities, including those supposedly justified by
externalities. Since the set of externalities can be made
almost indefinitely large by liberal interpretation, it is
important to assess Stiglitz's arguments that the state should
or can raise social utility by operating on externalities.
I will examine Stiglitz's views primarily on his home ground,
that is, with reference to his theorizing or his model. This
shall be done in a way that does not burden the reader with
his mathematics yet is completely faithful to his equations.
At the end of the argument, the reader will understand exactly
why we can reject his conclusion.

The Stiglitz Model

The Stiglitz model has no mathematical errors in it that I
know of. The equations follow one another correctly. The
implications are correctly deduced from the premises. The
problems with the model are deeper than mathematical. The
fundamental problems with the model involve its treatment of
government and the treatment of externalities. Stiglitz grafts
these onto the standard neoclassical model of exchange and
production. The grafts, however, are ad hoc, incomplete, and
inconsistent with what they are grafted onto. The resulting
theory provides no support for government.

Stiglitz starts with a standard timeless neoclassical model of
general equilibrium under certainty, and part of the paper
extends it to uncertainty. The paper's innovations include (1)
a simple method by which to tell whether in the model there is
a set of taxes, subsidies, and lump-sum transfers that
indicate a positive net "government" revenue while leaving
household utilities unchanged, and (2) applications of the
model to several situations that had previously been treated
in disparate fashion. These innovations, which I do not
criticize, are responsible for the paper's fame.

Model Basics

There are two sectors in the basic model, households and
firms. In standard neoclassical fashion, the households
maximize utility and the firms maximize profits. Firms are
"black boxes" that process inputs into outputs according to
some fixed technology, so they do nothing interesting in the
model. The paper's purpose is to explore the effects of
externalities on the Pareto optimality of household and firm
optimizations. The paper extends the neoclassical model. We
could dismiss the whole exercise as flawed because of its
neoclassical unreality, but, as I have said, I propose to
entertain the neoclassical model core and show that even if we
accept it, Stiglitz's extensions are unacceptable.

Stiglitz (and other neoclassical economists) model utility as
depending on two sorts of items. One is goods that are
produced, traded, and have prices. (Later in the model, the
"government" will be alleged to do particular things regarding
these goods such as tax them.) The second is items (called
externalities) that are not traded and do not have prices in
the model. They do not appear in the budget constraints of the
households (or firms) as costing anything, but the household
utilities are affected by them. Because they do not appear in
budgets and cannot be traded but do affect utility, household
optimizations leave possible utility gains on the table. The
equilibrium is therefore not Pareto optimal when these
externalities are present.

Government in the model is an entity that leaves each
household's utility unchanged while maximizing its excess of
tax revenues over lump sum transfers to households. This is
its behavior function. The only way that it can have such an
excess is when there are externalities. When such an excess
occurs, it amounts to money or utility left on the table by
households and firms. Such sub-optimization has to occur by
the assumption that neither is able to trade certain goods
that affect profits and utility. The implications follow
implacably from the assumptions.

Four Main Criticisms

Now, what are the fundamental criticisms of this model? There
are four big questions.

In the model, the households and the firms optimize, yet
somehow they are still not as well off as they can be. They
leave gains on the table that are sopped up by government.
Why don't households and firms fully optimize, that is, take
into account the externalities?

In the model, why is government able to recognize the
externalities and compensate for them by a series of taxes,
subsidies, and transfers while households and firms cannot
and do not?

In the model, what sort of entity is this government and how
does it relate to the entities that it is taxing?

What does the entity called government have to know in order
to effect the taxes, subsidies, and transfers?

I consider each of these questions in turn.

Why don't households and firms optimize fully? Stiglitz posits
that they optimize utility, which is standard neoclassical
economic theorizing. Assume that there are externalities. A
genuine theory would explain how the externalities arise even
while households and firms optimize. It would introduce a
theory of costs or some other factors that cause
externalities. This is essential if it is to be argued that
government can recognize and overcome these externalities.

Stiglitz provides no such theory. He simply inserts the
externalities into utility and production functions without
prices and then allows households and firms to optimize. This
procedure can be done mathematically, which is what Stiglitz
does, but it is entirely ad hoc. It lacks theoretical
justification because the externalities in the model do not
themselves arise in the context of other optimizing actions of
the economic actors.

I do not say that there should never be ad hoc theorizing.
There is a cost and benefit to it, and there will be times
when it is worth doing. But in a case where the central
message is to promote government and downplay markets, it just
will not do to introduce the reason for government in an ad
hoc manner.

There is a further problem with introducing externalities in
an ad hoc fashion, namely, model inconsistency. The Stiglitz
model is an equilibrium model, which means that everyone has
already done the best they can possibly do. From that
perspective, private parties already have done the best they
can do in order to internalize the externalities (create
contracts, exchanges, and prices for them). It is then
inconsistent with the spirit of optimizing models and
equilibrium models to introduce ad hoc externalities. Where do
they come from, and why aren't they already priced out? The
Stiglitz model contains at its core an ad hoc element that
drives all the results and that is inconsistent with the
optimizing and equilibrium assumptions of the model. It
assumes that actors who otherwise are absolutely punctilious
about optimizing utility and profits ignore the potentialities
of taking externalities into account. The model has
inconsistent assumptions.

Why is government capable of optimally rectifying
externalities? Households and firms ignore the externalities,
but the optimizing government does not. Why is government so
perspicacious? In fact, to Stiglitz his key equation "provides
a simple set of necessary conditions characterizing the
optimal level of taxes in the presence of externalities." It
is true that this equation makes sense within the model — once
one grants the existence of externalities and the existence of
an entity called government with the properties that the model
endows it with. It says that the government should push the
tax to the point where its marginal cost (in terms of
consumption and utility losses) equals its marginal gain (in
terms of lowering a negative externality.) But why does the
government optimize a measure of profits by taking into
account the externalities and the other sectors do not? The
theory provides no explanation.

Stiglitz introduces this optimization as a mathematical
convenience, employing it as a simple way to test for Pareto
optimality. But he also regards government as a separate
entity that interacts with households and firms which are the
other entities of the model. If government is a real entity as
he supposes, how and why does it have an optimizing capability
that households and firms lack? Stiglitz never tells us. He
places another basic inconsistency at the heart of the model.
What actually is government in the model? This issue is very
important because there is absolutely no doubt that Stiglitz
considers the government in his model to be a basis for
government policies and government interventions to improve
upon market economies, not only in this paper but in many
subsequent writings of his.

Logically, within the model's confines, the government is
either a creature of the other economic actors (households and
firms) or it is not. If it is an institution created and run
by the household and firm sectors to rationalize
externalities, then (assuming that it can and will do this and
that it is the optimal means of doing this), we actually are
not dealing with government at all in the usual sense of the
word. We are dealing with a voluntary means of negotiating
exchanges, a kind of a market, and we are not dealing with
coerced taxes. There is in this case no warrant whatever for
speaking of the model's government as representative of the state.

Real government is not an ongoing type of firm created and
managed by the private sector to iron out certain problems. If
it were, it would not be marked by the extensive power that it
has to impose measures. It would not be marked by life and
death debates over its every action. It would not be marked by
divisions between one part of it and another and between it
and society. It would not be marked by what seems to be its
creation of conflicts and externalities that rive society. It
would hardly be the destroyer of money or of wealth or the
inefficient manager of every activity that it touches. On the
other hand, if the government is the coercive institution in
the model that we know it to be in reality, then how is this
to be explained? What actors in the model create the
government? What actors run the government? What are their
optimizing behaviors in running it? What are their costs and
how are they spread over the other actors?

The fact is that Stiglitz never describes what this government
is or how it comes about. He never even describes its basic
properties. Government in the model is basically ad hoc. It is
motivated to produce taxes, subsidies and transfers to achieve
optimality in a manner that is nowhere described. It simply
does it. There is no warrant for making this assumption and
there is no warrant for taking it to be characteristic of
actual government. Stiglitz cannot sensibly talk about what
government should or should not do without a theory of government.

What does government have to know in the model in order to
accomplish its task? The list is staggering:

For each and every household, the marginal effect that a tax
has on each and every good that that household consumes;
The rate of change of each firm's profits with a change in
each externality (holding all else constant) and the rate of
change of each firm externality with a change in tax, all
else held constant;

The rate of change of each household's spending with a
change in each externality (all else constant) and the rate
of change of each household externality with a change in tax
(all else constant).

If we again stay within the confines of the model itself,
there is nothing in it that tells us how the actors who govern
are supposed to acquire this information or what the
acquisition costs are. If we think of the model as being
re-optimized as time passes, then there is variation in the
parameters and the governing actors would have to acquire even
more information.

In or out of the model, the information requirements present
an impossible task. Where does the knowledge of the required
rates of change reside? Preferences and production
technologies are not self-evident. They are internal to human
beings. It is not clear that humans either know them or can
communicate them, certainly not in the detail required by a
social planner. They possibly learn them over time and they
surely alter over time. Under these conditions, the economic
actors affected can surely not rely on a set of social
planners or engineers to tell them what they, the actors,
prefer or how they plan to produce a good, or even what goods
they plan to produce. If a government is introduced into the
model as a device to tax, subsidize, and make transfers, it is
entirely an ad hoc device. There is no economic theory of how
it can accomplish these tasks optimally, or how the actors
expect it to do so, or empower it to do so.

In sum, in terms of real economic theory the Stiglitz model is
about as empty on the issue of externalities and government as
it can possibly be. It is simply a mathematical representation
of ad hoc assumptions. Not only that, these assumptions
conflict with the optimizing and equilibrating that goes on in
the rest of the model.

Other Problems

I will now mention a few other serious problems that the
Stiglitz model ignores, although my intent is not to provide a
complete list.

There is not in reality a fixed set of goods or production
technologies. The problem faced by a Stiglitz government is a
dynamic one. But it can't even identify the sets of items that
the household and firm sectors are optimizing over at an
instant in time.

In or out of the model, it is not at all clear that
government, however constituted, is somehow better suited to
handling externalities than free markets. The real economic
questions concerning externalities are why they arise, how
they relate to property rights, what to do about them, or when
it pays for interested parties to do something about them. In
a sense, they are no different than any other kind of
inefficiency. Therefore, the presumption is that free
exchanges can address the perceived costs and benefits
associated with them whereas government is ill-suited to the
task. Stiglitz, by introducing externalities in an ad hoc way,
simply assumes that free markets have already failed. This is
why the Stiglitz government appears as the promoter of
economic efficiency.

In the Stiglitz world, some goods are taxed, others
subsidized, everyone adjusts, and then some find themselves
better off as a consequence, and no one is worse off. This is
done continuously as the economy evolves. As prices alter, the
pertinent taxes and subsidies alter too. If it is not done
continuously, quite possibly an imposed tax will lower welfare
when tastes, conditions, and prices shift. So, logically, to
assure betterment, it will have to be done continuously. This
means an infinite cost of transacting.

Hans-Hermann Hoppe has correctly pointed out that so-called
public goods can change their character and become private
goods (and vice versa) depending on the state of mind of the
actor. This adds to the burden of the government's
already-impossible information requirements.

The model ignores the fact that future opportunities are
altered when the government intervenes. With government in the
picture preempting action, no private sector actor has an
incentive to overcome the reasons for the initial inefficiency
by internalizing the externalities. The means have been
pre-chosen as the tax, subsidy, and transfer schemes.

Therefore innovation is stifled. If a lower cost or superior
method could have been found, it will not be if the solution
is government.

In an economy with Stiglitz government interference, whenever
any new product is invented, there is uncertainty about its
relations with other products and processes. The producer has
no way of telling what externalities the government will deem
important enough to intervene in the markets. The presence of
the state will chill many markets and potential markets. The
chances are that future production, trade, and original
appropriation will all be deterred.

Of course, Stiglitz entirely ignores any malicious acts by
government. He ignores any power dynamics. If government is
capable of intervening in markets across the entire economy on
the basis of externalities, the chances of gains in power are
greatly increased.

Stiglitz ignores any error-correction mechanism when
government makes mistakes in its assessments, that is,
government failure. If government were to be the institution
to handle externalities, we can be quite sure that its
incentive structure would make it inferior to free markets in
correcting mistakes.

Conclusion

The book that deserved the prize

Stiglitz believes that his theory (model) demonstrates an
economic role for government in principle. Upon this basis he
has for years promoted government solutions and denigrated
free markets. Stiglitz believes only in mathematical
neoclassical models. Even if we grant him his preference, he
has still completely and entirely failed to prove his point
that government has a potential role in alleviating
externality losses. He has given us an inadequate and
incomplete theory with inconsistent and ad hoc assumptions.
His equations contain no theory of government whatever and for
that reason alone they cannot possibly provide us with
real-world prescriptions.

We can have no confidence that the Stiglitz model captures the
essential aspects of real world economizing that it purports
to. We therefore can have no confidence in any belief that
rests upon this or similar theories that government has a
proper role to play in increasing economic efficiency or
social welfare by use of taxes, subsidies, and transfer payments.

Michael S. Rozeff is the Louis M. Jacobs Professor of Finance
at the University at Buffalo.