H/T “No Hope for the Human Race”
Why did so many economists fail to predict the financial crisis? Why is there such a vicious debate over something so seemingly simple as to whether fiscal stimulus is good or bad? Why do some economists seem to get it so wrong? These are the family of questions that a lot of people have been asking for a long time, especially following major economic catastrophes. A failure of precision has taken a toll on the economist’s credibility. I don’t think this is fair. I think there’s a lot of confusion on what an economist does, and a lot of the ambiguity is due to how economists think of themselves.
To me, asking an economist why the profession has failed to provide a suitable answer to cyclical fluctuations is akin to asking a political scientist why she hasn’t found a solution to war. These are complex problems, with no straightforward solutions. We, of course, have some general ideas on what the problem more-or-less is, but the process by which these problems arise and how “solutions” will affect this process is still shrouded in quite a bit of uncertainty. One is tempted to ask why, if we can build something like a nuclear reactor, are these “basic” economic questions still largely unanswered?
One well-worn route is that of distinguishing between the natural and the human sciences. Just like with every other issue, every economist holds a slightly different view on this distinction (some don’t see it at all). But, the basic idea is that whereas in the natural sciences we deal with predictable events, the social sciences — which revolve around humankind — are significantly different. The belief is, more-or-less, that in the natural sciences the basic units (say, the atom) have a constrained set of possible actions which are timeless, which means that these relationships don’t change over time. With humans this isn’t true, because the set of possible behaviors is endogenously constrained, meaning that it’s a function of human ingenuity and creativity. How can we predict how an individual will act if the range of possible actions is decided directly by the individual herself? The common summing up of the position is that “humans aren’t automata.”
I like the distinction, but since I’m mostly unfamiliar with the natural sciences I’m not comfortable with it. But, I think there’s at least some important element of truth in it. Historically, econometricians have struggled to build a permanent testable model of the economy, because they have failed to model the underlying “structure.” These have been problems since the birth of econometrics, but it culminated in the 1970s with Robert Lucas’ now well-known “Lucas critique.” The problem with basic economic policy on econometric analysis is that because econometricians lack a good, basic model of the structure, once policy changes the structure will adapt and the outcomes will change, and past data cannot offer us any clues on the direction of change. This is similar to the above distinction between the natural and social sciences in that it shares some vague idea of non-predictability. More importantly, it hints at the endogeneity of the economic process: structures adapt on their own, and if we don’t know how structures adapt, then our ability to predict changes will be much more limited.
I prefer a related explanation for the non-predictability of the human sciences: complexity. This isn’t a distinction between natural and social sciences, but between simple and complex phenomena. It’s a problem that besets the natural sciences, as well. The idea is that events like business cycles are influenced by a large number of variables, whether these variables are subject to endogenous change or not. The more complex something is, the more difficult it is for us to completely understand it. At some point, the best we can do is to point out typical features and build theories of relatively general character. If we make predictions at all, it can only be about these typical features, because more specific phenomenon require us to understand a larger proportion of the complex event (and, again, as the event becomes more complex, the less of it we’ll understand).
Economists have made a lot of headway in understanding the more general aspects of the economy. We have a pretty good understanding of how humans coordinate and why ordered events emerge at all out of the chaos of undirected human activity. We now know much more about institutions and their role in constraining the range of human choice. There is a healthy debate on the business cycle, but more than ever there is some consensus on the basic characteristics (credit booms, risk adjustments, et cetera). But, all of this knowledge is still more-or-less useless in the realm of exact prediction, because we have no idea how the smaller, but more numerous parts of these larger events operate — and even when we have an idea, it’s much more difficult to assign probabilities to different outcomes amongst these smaller parts. What’s more, we often don’t know what these smaller parts even consist of!
There’s a tendency to think of the economist as an engineer. An engineer builds a system and can solve its problems, most of the time. Why can’t the economist solve the problems of the human-built system that is the economy? But, the human economy is more complex than some people realize. It didn’t come about from the head of a single person, but is an ordered, structured process designed, over time, by all the individual agents that partake in exchange. What we know about it only scratches the surface. And, like with all complex phenomena, there is a paradox to the process of learning. Usually, the more we learn about something we realize that there’s more to what we don’t know than we previously thought.
The lay public may think of economists as engineers because they are interested in seeing major social problems solved. Business cycles hurt the majority of people. We are worse off because of them and it would be nice to get rid of them. Or, we’d like to find ways of maximizing economic growth and benefiting all in the best possible way. But, it’s not just the lay public’s fault. Economists are responsible for selling an image, and for as long as economics has been a field, many scientists of this profession have sold themselves as engineers. Beyond the merits of fiscal stimulus, one of the deepest rivalries within the profession is between those who think they can help control the system and those who believe we don’t know it well enough to intervene. The image of the science has been built around the former, but it should become increasingly clearer that the latter are in the right.
People expect too much from economists, and maybe economists expect too much from themselves. We cannot predict major events and we cannot know how different interventions will impact the multitude of agents and variables that decide major economic outcomes. Even the best predictions prior to the most recent crisis were vague and general: they were predicting the “typical features.” We are not doctors, engineers, or car mechanics. We can only hope to better understand social phenomena. In this sense, maybe people would have a better picture of what it is we do if they assumed that economics is mostly a retrospective science.
The second shortcoming of rationalism was its neglect of the problem of erroneous thinking. Most of the rationalist philosophers failed to see that even honest men, sincerely devoted to the search for truth, could err. This prepossession prevented them from doing justice to the ideologies and the metaphysical doctrines of the past. A doctrine of which they disapproved could in their opinion have been promoted only by purposeful deceit. Many of them dismissed all religions as the product of the intentional fraud of wicked impostors.
— Ludwig von Mises, Theory and History (Haven: Yale University Press, 1957), p. 270.
One of my favorite Credence Clearwater Revival songs,
Sometimes the state is the least costly means of solving a public goods problem, but there’s good reason to believe that the size of effective states should be limited. Not for normative reasons, but for positive reasons of efficiency. In Calculus of Consent, Buchanan and Tullock introduce two cost curves. One of them shows the relationship between the external costs of collective action and the proportion of the population that has to consent to the decision, with zero external costs achieved at unanimity (if everyone agrees to a decision, nobody loses). At the same time, they consider the costs of reaching a decision based on the decision-making rule, where the closer to unanimity that a decision-making rule gets the costlier it is. This is because there are costs to bargaining with people, and there are costs to learning information that can put you in a bargaining position, et cetera. One direct conclusion is that the optimal decision-making rule is where costs are minimized, which is oftentimes not at unanimity (meaning some external costs to government are oftentimes worth bearing). Another indirect conclusion, that the authors suggest later, is that the costs to decision-making can be lowered if the voting population is reduced (if the state is limited to smaller populations).
There are at least two other reasons I can think of why this last proposition may be true.
The reason we may want bigger states is to overcome economies of scale problems with certain public goods. Economies of scale is when marginal costs fall with growing output, implying high fixed costs. In private industry, large economies of scale may serve as a barrier to entry, since firms may need to have significant budgets to overcome fixed costs and the possibility of temporary losses until production is increased.However, there are costs to large states.
As the externality becomes larger the more information is necessary to solve it. As the amount of people affected by the externality grows, the greater are the costs imposed on the state to extract information from each individual — in other words, the marginal cost curve to information collection is upward sloping. But, it’s not only the information the state can gather that’s relevant; also important is the information that has a low probability of ever being known. We can call this radical ignorance. We know that we don’t know everything, and to some extent it’s worth paying to be informed on random knowledge. This is why we read the newspaper even if we have no idea what that newspaper will publish. But, until we discover it, the characteristics of these data are simply unknown to us; that is, while we recognize that there is information we don’t know, we also don’t know what this information entails exactly. The larger the scope of decision-making and the more people it affects, the higher the probability that we don’t know things that may actually be relevant, therefore the higher the probability that our decision will be, in some way, wrong. This affects the state as much as it does the individual in the private sphere.
The implication is that informational costs — both the costs of acquiring and the costs of practically non-acquirable information — can impose theoretical size constraints on the state, despite whatever advantages due to economies of scale. This means that there are some public good problems that could, if we drop certain conditions, be efficiently solved through collective action, but because of informational costs are no longer worth solving. It may be the case that it’s less costly to let them continue.
Related to this discussion, although only tangentially, is that the larger the state grows the less efficient it becomes at solving local, or smaller-scale, externalities. One reason this may be true is because as the amount of citizens grows the less able the government is to distinguish between relevant population groups. Also, informational costs make it less likely that a larger state will allot the right amount of funding to each group affected by some externality. This is why some countries practice decentralization, or Federalism, where local governments take the place of a central authority in local decision-making. It does have to do with local representation and avoiding problems of being ignored by a larger, centralized state, but these concerns are directly related to the economic problems just referred to.
This whole discussion is all good evidence in favor of the proposition that smaller, localized governments are superior to larger, centralized governments. By local I don’t mean, necessarily, state-level governments. It could be city-level government, or even smaller than that. This is ultimately an empirical question. What this may suggest is that “anarchy,” ultimately, may just be institutional evolution towards smaller states. Economies of scale problems, if they exist, can be solved by imposing stronger constraints on any larger governments that may arise, prohibiting them from interfering in local affairs, but providing them the ability to solve larger scale public good problems (e.g. national defense, if national defense really does fit the conditions). This is, in fact, what we saw in the United States during the transition from the English monarchy (although there were already local democratic governments in many of the colonies) to loose confederacy to Federalist state. Although, in this real world example, the constraints aren’t strong enough.
Suppose I want to host a party, but I don’t have the money to pay for everyone’s debauchery. I decide to make the party “BYOB,” or bring your own beer/bottle. Ideally, each person will buy and bring exactly the amount of alcohol that he wants to drink at the party. Unfortunately, in practice this rarely works. Some people bring alcohol and others, knowing that their friends are coming with the goods, will not, meaning that some ultimately end up subsidizing others. Enforcement is difficult, because it doesn’t pay to have much security. People will be too busy having fun, and the free rider will be able to snatch a few brews here-or-there. One solution is to charge at the door. There is still the problem of heterogeneous preferences for alcohol (meaning some who pay the fee will consume less than what the fee is worth, and others will consume more), but this rule tends to work better than BYOB, because at least everyone pays something.
This situation is somewhat analogous to tax collection. Assume, for the sake of argument, that there are certain public goods which can be provided for by the state with comparative efficiency. “Comparative efficiency” here means that the solution is less costly than leaving the public goods problem persist or the alternative private solution. If the state supplies these particular public goods more efficiently than the alternatives, there’s some reason to believe, as Buchanan and Tullock argue, that “rational” beings will choose this route — so, we can dispense with problems of coercion at this level. However, should the state fund its investment through voluntary or coerced tax payments?
If taxation is voluntary, meaning that the government essentially becomes a large charity, we have the same problems that “BYOB” parties do. Under the expectation that others will pay, some will choose not to pay. If the government can still acquire the funding for the provision of public goods, then we have a problem of free riding. Those who don’t pay for the service still receive the benefit, at the expense of those who had to unwillingly subsidize them. But, if the tax is forced on to everyone, the government can guarantee that all members of society will pay at least some of the cost of the service, reducing the externality by at least some amount.
Maybe so far everything said is obvious, but I think this is more significant than some people realize. We can frame free riding as theft or an infringement on property rights, which we often do when we discuss externalized costs. If the state is the least costly solution for the provision of some public good we already agree that this free riding problem can’t be solved privately — that is, there is no Cosean solution. The state’s coerced taxation scheme is the most efficient method of externality reduction. What this implies to me is that coerced taxation, under these specific conditions, can reduce the total amount of property right infringement. Put another way, tax coercion can help decrease aggregate coercion exhibited by society.
A lot of people don’t like what this thought experiment implies. I can certainly sympathize. But, as I tried to argue with Roderick Long, what you have to argue against isn’t the use of coercion by the state, but the possible justifications for the use of coercion. I don’t mean only on the theoretical level. It’s easy to build a theory with some policy recommendation, even a theory that could fit reality, but the other half of the story is what is empirically valid. If there are no real world public goods that fit the description necessary to justify collective action, then we can conclude that in the real world (as it presently exists) taxation adds to aggregate coercion. But, this is a much more difficult argument to make, which may be the reason why so few people have made it.
Man has only one tool to fight error: reason.
— Ludwig von Mises
1. Robert Fogel, well known for his work in economic history, including the economics of American slavery, has passed away at the age of 86.
2. Greece has been downgraded to an emerging market.
3. Kenya, a country facing a growing separatist movement, has cracked down on an attempt to introduce a local currency to facilitate exchange in a slum of Mombasa.
4. Stephen Corry reviews and critiques Pinker’s The Better Angels of Our Nature. Corry challenges Pinker’s data and interpretation. My reservation with Pinker’s book is somewhat different and boils down to “correlation is not causation.” (Edit: Here you can find various related links, including a full version of the Truthout piece.)
5. George Selgin on deflation,
6. Daniel Kuehn discusses a recent revelation of mine: dropping the term libertarianism. I was going to write a long post on this, but I’ll wait until I develop my thoughts more. I think libertarianism is still useful in characterizing my “policy recommendations,” but I’m just not sure how useful it is as a name for a specific political philosophy. I’m a liberal, because I believe in an inclusive society (i.e. pluralism), which means that I believe that the more people are included in the debate on the direction of their future the better. I contrast this with how I interpret one facet of conservatism, which is to argue that their beliefs — however justified (say, morally) — should be those implemented, and that debate is a superfluous (and even dangerous) aspect of political change. This is related to the broader issue of the limits to our knowledge, including the fact that nobody knows for sure what the best set of institutions is. The idea behind liberalism, or at least my interpretation of it, is to widen the debate and to introduce competition as a means of exploring the different directions of change people want to follow.