Public Choice originated as a heterodox corrective to a misguided focus on “market failure,” a central concept in the welfare economics literature of the 1950s. The orthodox work compared real-world markets with the “ideal” allocation of resources that would be selected by an omniscient, benevolent despot.
Opponents argued that government also faces the “knowledge problem,” and state employees are not immune to incentives. Real governments are neither omniscient nor benevolent, and so state action is not always better than markets. Sometimes, relying on government may make things much worse. In fact, it has been estimated that the total “killing by the state,” or total death toll from democide, exceeds 250 million since 1800.
Still, unless you are an anarchist, you accept that some government is necessary. But how much? Doing what? James Madison, primary author of the US Constitution, argued that (1) if people were angels no government would be necessary, and (2) if people were governed by angels then no constitutions would be necessary. But government is necessary, and constitutions are necessary, so market failure has reasonably to be compared to its doppelganger, “government failure.”
In deciding whether to use markets or the state, the real question is comparative: Which imperfect system solves the problem of ordering society relatively better, in a particular context, addressing a specific problem? In analytic terms, more mundane government failures can be divided into two types: failures of substance and failures of procedure.
1. Substance
Since (almost) any political order is better than the Hobbesian state of nature, failed states are the clearest examples of substantive government failure. The failure to keep order, to maintain property rights (including collective control to solve commons problems), to maintain a reasonably efficient and fair judicial system, and to maintain both the value of the currency and the ability to borrow by avoiding excessive debt, are all examples of substantive government failure. By these standards, many US municipalities are astonishing failures, and have been for years.
At the national level, there is a list of substantive government failure in the US over the past three decades:
the regulation of the financial system
the availability of a reliable, independent (not politicized) system for adjudicating disputes
the stability of the currency
the ability to control the expansion of debt, to secure modest interest rates on borrowing
the incapacity to provide the basic state function of controlling the border
The presence of these substantive government failures need not be taken as a need for expansion of the market sector. They are instead per se failures of basic government functions. Nonetheless, this panoply of failure does raise the question of whether the scope of state action should be restricted to the functions listed above, with an eye toward improving the function of the state by restricting its charter; This approach is characteristic of “state capacity libertarianism.”
2. Procedure
The fundamental political problem is the choice among Pareto optima. The reason is that any decision that is Pareto-improving will — in principle — be chosen by unanimous consent. But there are often two, or more, alternatives to the status quo, both of which are Pareto improvements, but which differ in those who would benefit from the change. Or, what amounts to the same thing, the existing situation is a Pareto optimum, and an alternative Pareto optimum has been proposed. By definition, the alternative will make some people better off, but some people will be worse off.
Which of the alternatives is “better,” in the sense that it would be chosen by an omniscient, benevolent despot? You might think that’s a ridiculous standard, but remember that is how “market failure” is measured. William Keech and I argued that government must somehow choose the “best” Pareto optimum, meaning that the utilitarian problem posed by the Kaldor-Hicks- Scitovsky paradigm will yield the maximum gains to the gainers, which must be larger than the losses to the losers.
One caveat: It is possible to argue that the entire welfare economics approach is nonsense, because no one could identify the “best of all possible worlds” that way. Fair enough. But we conducted our analysis by analogy to the market failure paradigm, where the optimal outcome is known. Even then, conceding the dubious claim about knowing the correct solution, government fails to choose the optimal Pareto optimum, because government must use a procedure to decide. And all procedures are flawed.
A simple example might be a decision to build a dam for flood control over a large region. The dam will be financed out of general tax revenues (each of the N citizens pays 1/N of the cost). The status quo, A, is the current world, with no dam. The alternative is B, where the dam is built. Which is better?
Many people will benefit if the dam is built, but some people will be harmed, because they have to leave the farms, homes, and villages where their families have lived for generations. A (no dam) is a Pareto optimum, and B (build dam) is a Pareto optimum. Which one is better?
Procedurally, governments have two main ways of answering the A vs. B question: democracy or technocracy. Will either deliver the “best” solution?
Suppose there are five citizens making the decision. Three of them would like for the dam to be built, but–given the tax costs of building — their preference is only slightly in favor of B. Two citizens, who would lose their homes, oppose the dam fiercely.
Process 1: Majority Rule
If we could account for the values the citizens place on the dam, three slightly in favor and two fiercely opposed, we would not build the dam. But that is not how democracy works. Instead, we take a vote, using majority rule, outcome B wins 3-2, and the dam is built.
B is Pareto inferior to A, however, because the two citizens opposed to the dam suffer harms that exceed the slight benefits to those who favored the dam. Using majority rule, or any voting mechanism, is a procedural government failure, because it almost certainly fails to choose the Pareto optimal outcome.
Process 2: Cost-Benefit Analysis
Experts should be able to solve this problem; all they need is accurate information about how each citizen values alternatives A and B, and they can “add up” the utilities, just like Kaldor, Hicks, and Scitovsky claimed. But there is a knowledge problem: when the bureaucrats ask the citizens which alternative they prefer, there is an incentive to exaggerate the “value” of whichever outcome is preferred. If I prefer that the dam be built, I will claim that I value it very highly, since my reported value has no effect on the cost I pay. Again, B is built, even though A “would be better” if there were an omniscient, benevolent dictator.
And that was our conclusion: the problem is not that markets fail, but rather that there is no omniscient, benevolent dictator. Keech and I considered a total of five possible procedures for choosing among Pareto optima, and showed that each of them fails to reliably identify the optimal Pareto optimum.
The actual argument is rather technical, and excessively detailed for this forum. I wanted to look back on our effort from ten years ago, to remind those who (correctly) see problems in market processes and immediately advocate for state action to “fix it.”
Perhaps the easiest way to explain government failure in one lesson is to remember that there is no such thing as “the state.” Instead, essential decisions about resource use will be made by political actors. This suggests what I have called “The Munger Test.” If someone says, “I believe that government should make decisions about what information is true in an emergency, and what should be censored!”, then you should make a simple suggestion: Take out the word “government,” and replace it with “Trump” (or “Biden,” I’m not making a partisan point).
See if the person still believes their argument, with that amendment. They probably won’t. That’s the “one lesson”: government is made up of people, using a process of discovery — voting or bureaucracy — that fails compared to an imaginary standard of omniscience and benevolence. In truth, results are rarely so bad in private life that government meddling can’t make things much worse.
