Many economists think government failure derives from market failure. If
markets are the default way of organizing human activity, problems arise when these
processes fail to achieve Pareto optimality or gains from mutual cooperation.
Government then arises to provide public goods, correct for information
asymmetries, internalize externalities, regulate natural monopolies, and so on. But
giving this power to the state has a dark side: if the state apparatus goes wrong, we
have government failure. Nevertheless, the state is useful overall and in fact necessary,
dark side or not.
Nobody else thinks that way, which may explain (along with the plaid sport coat)
why economists often stand alone at cocktail parties. Defects in state action may be
analogous to market failure. But it is important to recognize that government failures
are prior to market failure. Further, the problems are not the consequence of hiring
bad people but are instead systematic problems in the way government works. As
Adam Smith put it, It is the system of government, the situation in which [people]
are placed, that I mean to censure, not the character of those who have acted in it.
They acted as their situation naturally directed, and they who have clamoured the
loudest against them would probably not have acted better themselves (The Wealth
of Nations , book IV, chap. 7, para. 193). Smith was talking about the
employees of the East India Company in this passage. But the insight is a general
one: the failure of a system of organization often arises from the incentives, the logic
of action, or the inconsistencies inherent in that system. The people who work in
that system probably act in much the same way that other people would act if they
found themselves in that system.
The public choice tradition of scholarship, operating along the hedge between
politics and economics, has recognized that government failure is a primitive, not a
derivative, feature of attempts to organize groups of people in a society. The results of
this body of scholarship might be summarized in terms of problems of government:
Aggregationthe Arrow Problem
Collective Actionthe Olson Problem
Delegation and Agencythe Niskanen/McNollgast Problem
Informationthe Hayek Problem
Institutionsthe North/Ostrom Problem
Limitationthe Buchanan Problem
Rent Seekingthe Tullock Problem
This is an enormous body of scholarship, some of it very influential (there are five
Nobel Prize winners on that list) in the study of law and government. The reason that
public choice is important in this context is that it examines governmentor, as early
versions of the approach had it, nonmarket institutionsas a legitimate object of study
for its own performance rather than as a derivative of or reaction to market processes.
Peter Schucks book Why Government Fails so Often is in many ways commendable.
In my opinion, he makes too few direct connections to the corpus of publicchoice
scholarship (as far as I can tell), but the outline of government failure he uses
gives us a book that is both highly readable and intellectually important.
The breadth of the book is impressive, and I can do no more here than offer a
terse digest of the argument. Schuck begins with the empirical fact that many people
(more in the United States than elsewhere) think government programs dont work
very well. He offers several explanation, including one we sometimes overlook: we ask
government to do things that cant be done, such as eliminate risk and provide
benefits at zero cost. One cant really blame the state for failing to do the impossible.
But then Schuck turns to his main claim: the U.S. federal government does in
fact perform poorly in a vast range of domestic programs (p. 4). His criteria are
empirical rather than normative: Given the goals of the program and its cost, is the
government able to achieve anything close to the goal for anything close to a cost
that can be justified by the benefits of the program? The answer is not just no;
Schuck claims the answer is almost never.
Schuck is no anarchist; he is firmly in the camp of analysts who believe that
the state has core functions that cannot be handled in any other kind of organization.
The problem is that even in these core functions many government programs are not
cost effective. Schuck bends over backward to grant claims about moral imperatives
and about differing perceptions of the normative importance of providing particular
programs. The problem is that even in the core functions that (almost) all of us would
grant are state territory, few programs satisfy even the most generous cost-effectiveness
standards.We spend too much for too little benefit, and too many of the resources are
diverted or wasted.
Schuck sees four costs in immanent government failure. The first is wasting
resources that have a social opportunity cost. It may be hard to measure the costs of
what Bastiat called the unseen, but the costs are real. Second, Americans who
depend, perhaps increasingly, on government programs suffer from both the
economic inefficiency and the simple human indignities of having to deal with the
bureaucratic labyrinth. Third is the cost of future economic growth as regulations and
restrictions damp down innovation and prosperity. Fourthbecause Schuck is not an
anarchiststate programs manifold failures threaten the legitimacy of government
as a tool for addressing social problems.
The reason this kind of objective cost analysis matters is that most people who
favor state action actually care about the population being served (or not served, as
the case may be). If the budget is larded up with pork and debt service, then programs
that benefit the poor will be starved of resources. Schucks thesis is not that the state is
inherently bad or hires bad people, but rather that the state often tends toward
inefficiency, waste, and a sequence of diversionary reforms. Thus, public dissatisfaction
with government programs is fully justified because government programs dont
perform very well.
One of the most provocative parts of the book is the section where Schuck
makes a very strongand perhaps controversialclaim for the primacy of costbenefit
analysis (pp. 5163). The alternative is to see government programs as moral imperatives:
There is a problem.We must do something. X is something. Therefore, we must
do X. X was a success, because we did something when there was a problem.
Schucks costbenefit approach poses two questions: Why would anyone think X
addresses the problem in the first place? And how would we know if X fixed the
problem? The political logic of we must do something armors the program against
evaluation or attempts to measure whether it succeeded.
Schuck makes fourteen general claims about costbenefit analysis; I have space
to discuss just two of them. First, number 8: Ending a failed policy is a kind of policy
success (p. 59). Of course, a program creates beneficiaries who will suffer when the
program is ended, and cost savings are spread out over millions of taxpayers. Schuck is
arguing for a language of politics where we give politicians credit for refusing to do
the wrong thing. That may mean the right thing is to end a wasteful program.
The other claim I thought especially insightful is number 13: Policy assessments
must take most of the existing social and institutional context as given (p. 62).
This point is so obvious that it completely escapes policy makers, either because they
find it useful to ignore or because they really think they live in an alternate reality.
Sports stadiums will revitalize downtowns, having enormous multiplier effects. Enormous
incentives for development will pay for themselves in increased taxes after a
downtown area is completely transformed overnight. And health-care reforms will
more than pay for themselves, sharply reducing cost while increasing access and
improving the quality of care. And lets make all the roads run downhill in both
directions while were at it.
It may not be obvious why this point is so important, but those who understand
forecasting recognize that it would be a game changer if agencies were obliged to
take cognizance of it. For long-term forecasts, small changes in assumptions in the
out years (meaning, after the next election) can dramatically affect the cost
effectiveness of the program being advocated. It is common to assume that taxes will
be repealed, deficits will disappear, and unemployment will remain lower than 6 percent
forever. The last example was actually the basis for the Congressional Budget Office
[CBO] forecast of the economy in 2012. Forever. Nice.
But we cant blame the agencies for using faulty data in this case. It is Congress
that requires that agencies make forecasts of cost and benefits, but then it is Congress
that fills in the crucial assumptions that determine the outcome of the costbenefit
analysis. The CBO is obliged to accept and embody in its forecasts the assumptions
that are written into the law that the CBO is supposed to be evaluating. If we assume
that gross domestic product will grow at 4 percent in real terms, unemployment will
fall to 6 percent and stay there, and productivity will grow at 3 percent annually, the
results of our cost-efficiency analysis will be a fantasy, serving nothing but
Congresss desire to bribe taxpayers with their own money.
This brings us around to voters. If agencies are forced to use rules and assumptions
that disguise costs or exaggerate benefits, then why not blame Congress? Why
dont voters demand better representation?
Schucks answer is right but depressing. Congress makes those rules, too. We
dont get to vote for Congress as an institution. We get to vote for our member
of Congress, who runs in a district designed to ensure continued reelection, in a
campaign finance system designed to thwart competition, and in an electoral system
that raises the costs of seeking office in the first place. In his discussion of the Voting
Rights Act of 1965 (pp. 34550), Schuck grants that the suppression of minorities
rights to vote or to seek office required federal action as a remedy. But a consequence,
perhaps unintended at first but now a part of the electoral landscape, was the
enshrinement of strategically drawn single-member districts that make substantive
minority representation much more likely. Of course, if the state requires that black
voters be gathered into a single district, the surrounding three or four districts
become less diverse and (in recent American politics) more Republican. This isnt
the only cause of polarizationthe Senate is venomously polarized, and states arent
gerrymanderedbut it is an example of the way that a program with a plausible and
even laudable purpose can be changed over time to create and preserve a catastrophic
failure of government action.
One churlish quibble, which must also serve as an excuse: Schuck does something
that is inexcusable for a book of such immense scholarship and broad appeal.
The background work in a variety of literatures that he synthesizes is enormous. But
then he makes it hard for the reader to find any of this work efficiently. There are two
ways, and (unless the book is fully electronic) only two correct ways, of organizing
references. The first is to have footnotes at the bottom of the page providing references
for sources cited in the text, with a list of works cited at the end of the book,
then to include an index of topics and important names included in the book. This
means that the footnotes arent too long and that the reader can use the index to find
a topic, use the footnotes on that same page to find the references, and then look
up the references.
The second is to use endnotes, as Schuck does, and then to include an index in
which a reader can look up a topic, find the place where it is discussed in the text, and
find the references in the endnotes. More importantly, a reader can discover what a
particular scholar has written on topics referenced in the book.
Schuck uses endnotes and then does not include the 1,200 endnotes in the
index. Worse, there is no Works Cited list. The reader cannot find an author or
topic without reading all of the endnotes. I understand that Schuck, as a law professor,
judges his scholarship by his endnotes. But its not the size of the list, but how
you help the reader use it that should matter. This is a querulous point, of course. The
only reason I raise it is that so much interesting and useful material is presented here
but then is placed behind a paywall of deadweight inconvenience.
Taking stock of the book as a whole, I find that the end is a bit of a false note.
Schuck goes to considerable length in the final chapters to argue that big government
is here to stay. But he also argues that the relationship between governments
growing ambition and its endemic failure is rooted in an inescapable, structural
condition: officials meager tools and limited understanding of the opaque, complex
social world they aim to manipulate (p. 412). Schuck is a libertarian in denial,
trapped in a worldview that his own analysis should help him escape. Or perhaps I just
underestimate the power of the Dark Side.