Whether open borders would really “double world GDP” depends greatly on how it will affect institutions. So let me offer a brief introduction to institutional economics.

There is one problem, however, that must be dealt with first. Institutional economics has become identified with economists like Dani Rodrik and Daron Acemoglu, with papers like “The Colonial Origins of Comparative Development” and “Institutions Rule”  and with books like One Economics, Many Recipes: Globalization, Institutions, and Economic Growth and Why Nations Fail: The Origins of Power, Prosperity, and Poverty. All this is, as I see it, is fairly marginal relative to the real insights that institutional economics has to contribute, and the real task it has to do.

In the past couple of decades, there has been a boom in loose, hasty “institutional” narratives about comparative development, because “institutional” stories provide an interpretation of “total factor productivity”– the residual differences in international income that can’t be explained by factor endowments– that is (a) difficult to refute, because theoretically weak, and (b) somewhat politically correct. Cultural explanations of the wealth and poverty of nations are politically incorrect, though Deirdre McCloskey has ventured one. Racial/biological explanations are even more politically incorrect, though Greg Clark hints at one in A Farewell to Alms. But a book like Why Nations Fail stepped into a kind of vacuum, telling development economics what it wanted to hear. Its ascendancy is a disaster for the field. For lack of space, I’ll outsource my criticisms to Duncan Green and Jeff Sachs, satisfying myself with a few cryptic dicta. Property rights are not “inclusive economic institutions;” on the contrary, they exclude. They also permit “extraction,” e.g., coal mining, harvesting crops. “Inclusive economic institutions” ought to mean communism; to link democracy and capitalism by calling them both “inclusive” makes no sense. Non-democracies are no more “extractive” than democracies, in general. If anything, democracies take more resources from productive people by force, via tax and transfer systems. It’s not in the interest of democratic majorities to establish pro-growth institutions because, as Arrow’s impossibility theorem long since proved, it’s meaningless to speak of the interests of democratic majorities. It often is in the interest of autocrats to establish pro-growth institutions, so as to have more to extract: this is Mancur Olson’s “stationary bandit” argument. No matter how much we wish it were true, and no matter how much jargon we dress it up in, development is not a function of democracy. The thesis that growth depends on “inclusive economic institutions” which “inclusive political institutions” help to establish, while “nations fail” because “extractive political institutions” establish “extractive economic institutions,” has no merit.

(By the way, I’ve lamented the ascendancy of Acemoglu and Robinson before. See also Bryan Caplan’s link, and the comments below it, where I voiced more of my complaints. To my surprise, the Wikipedia page on Why Nations Fail actually links to both my post and Caplan’s! And here’s a post, “Comment on Smith’s Critique of Why Nations Fail,” with an extensive discussion following, in which I participated.)

I want to get back to basics. Here’s the core of institutional economics, which, by the way, has nothing to do with democracy. It starts, like most things in economics, with supply and demand:

Think of that chart as the central corridor of economics, of which every feature is a door into a room containing some subfield. Open the “D” door, and we enter consumer theory, with utility functions, budget constraints, own price and cross price and income elasticities of demand, etc. Open the “S” door, and we enter the theory of the firm, with long-run and short-run total and marginal and average cost, shutdown points and breakeven points, and behind that, isoquants and isocost lines; or, as an adjoining room behind the same door, we can enter industrial organization, with perfect competition and monopolistic competition and monopoly and oligopoly. Open the “P” door, and since prices are measured in money, we step into monetary economics, with price stickiness, and the equation of exchange, and various theories of the short-run trade-off between inflation and unemployment. Let “P” be wages and “Q” be employment and you’ve walked into labor economics. Let “P” be interest rate (or rate of return) and “Q” be quantity of loanable funds (or shares) and you’ve taken your first step in financial economics. Draw a new line below the equilibrium price and call it “world price” and you’ve got yourself an international trade model. Focus on “consumer surplus” and “producer surplus” and you’ve entered welfare economics. Push D down and to the right, or to the left, and call the gap between old D and new D a “tax,” or a “subsidy,” respectively, and you’ve entered public economics.

What about institutional economics? Where’s the door into that?

To enter institutional economics, stop to notice a few of the assumptions that must lurk in the background for the chart to make any sense. In particular:

• Specialization. The model requires that some people have a surplus of the good to sell, while others have a need for the good to be met. Surpluses and unmet needs arise from, or at least are greatly multiplied by, specialization. Because I make X all the time, I’m very good at it; but for the same reason, I don’t have time to make any Y. So I supply X and demand Y. Thus supply and demand are born.

• But not too much specialization, because you need competition too. The basic supply-and-demand model assumes price-taking behavior on the part of both buyers and sellers. Only competition can ensure that. But now suppose that a society of 1,000 people divides its work into 1,000 tasks so that each person is a monopoly provider of the one thing they do. There’s no competition, and no reason for suppliers to be price takers. So markets of the kind shown in the chart, with their nice efficiency properties, occur only somewhere in the middle of a spectrum running from no specialization to perfect specialization.

• Property rights– including complicated contracts. Demanders must have some kind of property rights to the money they bring to market. They must have power to transfer these rights. Likewise, suppliers must have property rights in the goods they bring to market, and power to transfer them. “Property rights” can mean legal property rights, credibly backed up by a government, or customary rights, or private promises, or perhaps, occasionally, even the mere informal trading of favors. Often legal property rights and social trust are both necessary, as when contracts are signed, but cannot be made sufficiently detailed to cover all contingencies.

• Some method of executing transactions, and it shouldn’t be too costly. Institutional economists talk a lot about “transactions costs,” and that sounds really, really BORING. Sure, it does take a bit of time and effort to move money from hand to hand, ring up cash registers, sign checks, mail stuff, etc., but is that really important enough to merit our attention? To this objection, the first answer is that “transactions costs” include things like contracts and search and establishing trust and getting incentives right. Transactions costs in this sense can plausibly be estimated to comprise more than half of US GDP. Second, and even more importantly, transactions costs determine the size of the network of specialization and trade. The transactions that don’t happen are as important as the transactions that do. Everyday example: the typical reader of this post probably does his or her own dishes, even though the value of his or her time is a good deal higher than many people who live nearby. Why? Because it’s too complicated to find them, hire them, communicate the nature and scope of the job to them, move them physically to where the job needs to be done, coordinate their entry and exit, and structure their incentives properly. Heighten entrepreneurial cunning and social trust enough, and you’d press a button, and some underemployed person in your neighborhood would come and do your dishes for you at an easily affordable cost, while you devote yourself to business or pleasure.

So how does institutional economics relate to the supply-and-demand chart that is the heart of economics? Institutions determine what markets exist at all, and who participates in them. Institutions enable– or fail to enable– enough people to do business with each other that all sorts of specialized markets can come into existence, creating wonderful productivity increases and wonderful consumer and producer surplus. And institutions take up the slack when markets can’t operate because specialization has been pursued too far to be compatible with competition.

Even though Adam Smith’s trumpet triumphantly declares, in the title of Chapter 3 of The Wealth of Nations, that “the division of labor is limited by the extent of the market,” economists have a bias against believing it, because it creates fundamental problems for the theory of competitive markets, a bias well-expressed by Becker and Murphy’s paper “The Division of Labor, Coordination Costs, and Knowledge.” When I read this paper for the first time in grad school as part of my dissertation research, I was very annoyed with it, because it missed the point in such a lucid and convincing way. Today, I love the paper for the same reason: it expresses a fallacy that pervades the economics profession, in a form that’s clear enough to be straightforwardly defeasible.

Becker and Murphy start by assuming that society’s tasks are infinitely divisible, that productivity in any given task depends on specific human capital, and that specialization raises productivity by allowing workers to focus their specific human capital acquisition efforts on one narrow task and therefore to acquire more of it. So far, the model points to perfect specialization, i.e., a society in which everyone is the unique specialist in one very narrow task. But then they argue:

Conflict among members grows with the size of a team because members have greater incentives to shirk when they get a smaller share of the output (see, e.g., Holmstrom (1982)). Moreover, efforts to extract rents by “holding-up” other members also grows as the number of members performing complementary tasks increases (see Chari and Jones (1991)). Further, the chances of a breakdown in production due to poor coordination of the tasks and functions performed by different members, or to communication of misleading information among members, also tends to expand as the number of separate specialists grows… Principal-agent conflicts, hold-up problems, and breakdowns in supply and communication all tend to grow as the degree of specialization increases.

They go on to assume that average coordination costs have the functional form C=C(n), dC/dn>0, that is, average coordination costs increase with the size of the “team.” That’s why they think that “sometimes the division of labor is limited by the extent of the market, but more frequently in the modern world it is limited by other forces.” To support this claim empirically, they go on to give many examples of markets where there seem to be quite a few fairly substitutable specialists. The facts they cite would constitute evidence against the theory that “the division of labor is limited by the extent of the market,” if that theory implied that there should be only one monopolistic specialist in most tasks. But it does not, and the literature they cite in support of the concept of “coordination costs” implicitly contains the reasons why the extent of the market can limit the division of labor well before the technical limits of useful specialization have been reached. Most “coordination costs” are related to missing markets, and would be mitigated if the market were large enough.

Start with the “team production” problem of Holmstrom (1982). The output of a team is usually such that the sum of the marginal products of its members is greater than the value of what the team produces. But that’s a complex and jargon-ridden way of saying it. Let me try again. There is a popular saying that “the whole is greater than the sum of the parts.” Now, if that’s true, what happens if you take one part away? The whole is no longer a whole, and its value is greatly reduced. That reduction in value is the “marginal product” of the part, in the sense that it is what that part adds to the value of the whole, provided all the other parts are in place. But, precisely because “the whole is greater than the sum of the parts,” every part has a large “marginal product.” The traditional theory of the firm assumes that each factor is paid its marginal product. If “the whole is greater than the sum of the parts,” then the sum of the marginal products of the parts is greater than the whole, so each factor can’t be paid its marginal product within the team’s budget constraint. And that creates “hold-up problems,” with each team member being in a position to blackmail the rest for a large share of the team product. Such hold-up problems can cause the team’s cooperation to break down, even though its potential product is more than enough to pay team members their opportunity costs.

But now suppose that the market is large enough that every team member is readily replaceable. If you want to interrupt me here, and object that even if outsiders were ex ante replaceable, once the team has begun work, its members gain lots of project-specific knowledge that makes them irreplaceable, then you haven’t meditated deeply enough on the condition that “the market is large enough.” In a large enough market, there will be many other teams working on very similar projects, at a similar stage, so team members remain replaceable throughout the project. In that case, the hold-up problem is easily solved. If any team member demands an undue share, just fire him and poach a similar person from some other team. If the market isn’t quite that large, maybe there’s a moderately similar team member who can be recruited if someone demands too large a share, and that at least mitigates the problem. The larger the market, the less the “coordination costs” associated with moral hazard in teams, relationship-specific or asset-specific or project-specific knowledge and investments, opportunism, hold-up problems, etc. A similar logic applies to production breakdowns due to logistical problems or miscommunication: the larger the market, the easier it will be to buy key spare parts when there’s a bottleneck.

And so, in my dissertation, I proposed the following modification of Becker and Murphy. Rather than C(n), average coordination costs are C(n,N), where n is the size of the team, and N is the size of the market. C(n,N) is an increasing function of n, but a decreasing function of N. Division of labor is limited by the extent of the market, but there isn’t perfect specialization, precisely because as the economy approaches that limit, coordination costs become too high. People under-specialize to avoid getting trapped in monopoly-monopsony relationships where they’re vulnerable to opportunism, hold-ups, coordination failures, etc. To the extent that monopoly-monopsony relationships are inevitable, that’s what firms, those “islands of planning in a market sea,” are for. Where market relationships would be dysfunctional, power relationships arise instead. In a world of perfect information and costless and complete contracts, there would be no need for “firms” or “jobs”: we’d all be free agents, trading goods and services. But because a boss and employee need to do a lot of relationship-specific investment, and the boss has better information about the value of various things the employee could do, the economy is mostly organized as firms and jobs, instead.

The role of “good institutions,” then– at least, as I read it: but the issue is very complex– is not so much to give people political representation and redistribute income, or even to provide public goods, tax or prohibit activities with negative externalities, subsidize activities with positive externalities, and ensure that industries remain competitive. I don’t think it’s even primarily to enforce property rights in an everyday sense– plenty of regimes can do a decent job of preventing shoplifting and trespassing– but rather, to do arcane things like protect minority shareholders, and put Martha Stewart sent to jail on obscure charges that most people can’t understand. Why does it matter to protect minority shareholders? Because then people will hand over their hard-earned money to companies they’ve barely heard of, which is pretty amazing if you think about it. And that allows firms to raise cash for new projects, and for people like Mark Zuckerberg who have done something really useful to become billionaires through an IPO, mitigating the burden of nondiversifiable entrepreneurial risk.

How important are institutions for comparative development? That’s too difficult a question for me to answer with much confidence, but tentatively: I think culture does a lot of the work that is sometimes attributed to “institutions,” but culture and institutions together probably are the main explanations of differences in “total factor productivity.” To see what I mean by “culture,” consider the question: why do US universities really create and disseminate knowledge, when it’s ridiculously difficult to monitor what a professor actually does in the classroom? Because professors are inducted into a culture of learning, to the point where we really care about knowledge and truth and rigor, etc., not (only) for the sake of payday or tenure or even prestige, but for its own sake. We give Ds to terrible students, even if it gets us bad teacher evals, because it’s wrong to tell the world that someone understands economics, who doesn’t. And I’m sure similar forms of professionalism apply in many fields. I’m skeptical, on empirical grounds, of Weber’s “Protestant work ethic” as a theory of comparative development, but I think that kind of explanation, concerning how culture, history, and religion shape individual behavior and values in a way that isn’t reducible to homo economicus, is probably important to the wealth and poverty of nations. In cutting-edge, innovative industries, culture usually has to do the initial work in solving team production problems. Rewards are wildly incommensurate with effort and risk, and perverse incentives are all over the place, but people do the right thing because they love cars, or computers, or whatever. Later, institutions codify and enforce the modes of cooperation that culture discovered, though even in mature industries, norms and values and intrinsic rewards are crucial to sustaining high performance.

Given my perspective on institutional economics (including culture), how would open borders affect the institutions of rich countries? Would they “kill the goose that lays the golden eggs?” Or would poor people enjoy the productivity-enhancing power of sophisticated, wealth-fostering institutions (and cultures), without damaging them?

There’s a farm I know in Maine which sells veggies by an honor system. The veggies are laid out on tables in a shed, with prices, and it’s up to you to put money in their cash box. No one is watching. It’s a very convenient and affordable way to buy fresh vegetables. This business model might collapse under open borders, as a critical mass of immigrants prove less honest than Americans, and shoplifters make honor-system sales unsustainable. Or, it might not. It wouldn’t surprise me much if even more honor-system commerce took place under open borders. But I’d expect less of it. I’d also expect to see more littering and open defecation under open borders. I think the kinds of transactions that are as vulnerable as this to a deterioration in mass behavior are of minor importance to the economy, but I’m not sure.

In big business and high finance, by contrast, I doubt open borders would do any serious damage to culture and institutions, as long as open borders doesn’t mean open voting. These are elite institutions with plenty of gatekeepers, and no one rises high enough to make the rules without being thoroughly shaped by the rules first. Even the average native-born citizen knows little about them and has no power to influence them, except very indirectly at the ballot box. Big business and high finance have a much larger effect on GDP than rural veggie stands, though their relative importance for quality of life is perhaps a good deal less than for GDP.

In general, I think most civil society organizations and trust-based commercial relationships wouldn’t be harmed by open borders because one isn’t dealing with random strangers, but with people self-selected and/or screened in various ways. Universities, for example, only accept students, and hire faculty, who meet their standards. The fact that someone is physically on US soil doesn’t compel a university to do business with them in any way. The exception here is that anti-discrimination law sometimes does compel private and public organizations to deal with people they would prefer to screen out (especially under the deplorable “disparate impact” doctrine which forbids people and companies to statistically discriminate on grounds that are correlated with race, and that courts or bureaucrats should arbitrarily decide are inadmissible).

My main fear for institutions under open borders is that law and public opinion would fail to recognize that private discrimination against immigrants is morally fine and should be legal. If firms, public bureaucracies, and civil society organizations, whose members and leaders obviously have the best knowledge about those organizations’ goals, needs, and workings, were arbitrarily forced to include people they didn’t want by bureaucrats and judges, then they could sustain serious damage, and the killing the goose argument might come into its own.

Open borders might also have a beneficial impact on institutions through international Tiebout competition.