What Does It Mean To Impact Institutions?
Ferdinand Richardt, "Underneath Niagara Falls"

What Does It Mean To Impact Institutions?

Tiebout vs. Cascade Models of Institutional Change

The most plausible argument for mass immigration would be something like factor efficiency plus Tiebout competition. Labor mobility improves people’s lives in the short run by letting labor move to where it’s most productive, a straightforward implication of welfare economics. It also improves lives in the long run by letting people escape regions with bad policies and move to regions with good policies. Take citizens away from governments with bad policies, and give them to governments with good policies. In this way, bad policies are supposed to become less viable, and good policies to proliferate.

The most plausible argument against mass immigration has a much different model of institutional change in mind. If institutions arise from the people that form them, bad institutions can follow the people trying to escape from them. To allow this en masse is to allow those bad institutions to spread, which – in the long run – negates any factor efficiency gains. It is better, perhaps, to have exclusionary pockets of well-governed prosperity than the universal poverty that has characterized mankind for most of its history.

You’ll notice that the causal story in the second paragraph is a bit loose compared to the first. How exactly do institutions follow people? Most people who make that argument have in mind something overly concrete – something like “immigrants bring crime with them” (which makes it a question of labor economics), or “immigrants will just vote to get more welfare” (which makes it a question of political science). Unfortunately, most of the debate on immigration and institutions has happened on the level of questions like this, with ambiguous evidence being amassed on issues so concrete as to be only tangentially relevant to the broader question: how do institutions change?

In fact, the folk intuition on institutional change – what I’ll call a “cascade model” – is perhaps more warranted than its usually-loose formulation would suggest. Its implicit model of institutions as tentative and fragile solutions to deep social dilemmas comports better with what we know of human sociality than the optimistic “natural harmony” of the Tiebout model. The more visible manifestations of institutional breakdown – crime, or voting for welfare – are just two especially salient instances of a far more general phenomenon, with a variety of more or less severe manifestations.

Central Park

What Problems do Institutions Solve?

The Tiebout and the cascade conceptions of institutions both agree: institutions are about providing public goods. A public good is anything that provides more benefit to society than to the individual; anything that generates positive spillover benefits and therefore encourages free-riding.

The similarities stop there. Even this bare definition of a ‘public good’ belies a very different notion of the paradigmatic case.

Institutions in the Tiebout Model

The Tiebout model builds on the Pigou-Samuelson tradition, which thinks of public goods in opposition to private goods. Private goods can be reliably provided by a market. Public goods, on the other hand, will be underprovided relative to what people want because, without the ability to exclude non-contributors after it’s provided, people will tend to free-ride; to get the benefits of the public good without contributing to it. Everyone will wish there were more parks in the city, and be willing to pay some amount for it. But given the fact that it’s even better to let other people pay for the park, and that there’s no way to limit the park to payers,1 no one is motivated to actually contribute to building it.

The solution, then, is to form a government to force everyone to contribute. By hypothesis, everyone will be happy with this arrangement: they pay the value of the park to them in taxes, and – because they can count on everyone else paying as well – they get to enjoy a park in the city in return.

An institution, in the Tiebout-Pigou-Samuelson sense, is some mechanism in a community for (1) deciding on the level and mix of public goods, and (2) forcing people to contribute to them. In other words, an institution is defined by the answers to the questions: how high should taxes be, and what projects should they be spent on? Naturally, institutions are usually identified with government policy.

The example above is a sort of best-case scenario where everyone agrees on the value of a park. If some people don’t care about a park, they’re still forced to pay for it. If some people want bigger parks and are willing to pay more for them, there’s no way to satisfy that desire. Under these circumstances, how can we make sure to land on something like the “right” level and mix of public goods, or at the very least some mix that doesn’t dissatisfy everyone?

The Tiebout model is a response to a famous and depressing result that proved that voting, as we usually understand it, can never ensure that we don’t end up with a mix that pleases no one. Democracy, in other words, is no guarantee of good policy. But – Tiebout argued – as long as we let people move between regions with different institutions, with different mixes and levels of public goods, people will self-sort into communities that satisfy them. People willing to pay for parks get cities with parks and high taxes. People who’d rather have low taxes get cities with low taxes and no parks. Dictatorships and otherwise poor institutions that don’t provide a mix that somebody wants lose people until they reform or collapse.

With this mechanism, jurisdictional competition is supposed to solve the problem of optimal public goods provision in the same way that market competition solves the problem of optimal private goods provision. Exit, in other words – the ability to leave bad institutions – is more important than voice, the ability to vote for good institutions.

The implications for immigration are obvious. Immigrants “vote with their feet” for the level and mix in the new region, and against the level and mix in the old region. Not only does this provide the new region with another taxpayer and laborer, it also contributes in a small way to diminishing the viability of bad institutions on a global level. Not only is immigration good for institutions; without it, there’s no way to reliably get good institutions in the first place!

Institutions in the Cascade Model

The cascade model, on the other hand, imagines public goods in a much more abstract sense. Rather than a concrete amenity, a public good is simply a generalization of a prisoner’s dilemma to a whole community rather than between two people. People have the option to cooperate (provide the public good, and be vulnerable to exploitation by free-riders) or defect (do not provide the public good, and try to free-ride on those who do). This includes the provision of concrete amenities as in the Tiebout model, but also structurally equivalent collective action problems and commons problems. In these cases, the relevant public good is the collective action and not depleting the commons, respectively.

Unlike in the Tiebout story, however, a community faced with the necessity of providing some public good cannot simply form a government, first of all because many of the relevant public goods are not goods at all in the sense that a government can raise taxes to pay for them. Politeness to strangers, for example, is a public good in this sense. But more importantly, it begs the question to think of a government as a solution to the free-rider problem of public goods provision, because a government is itself – like all large-scale organizations for collective action – a public good upon which people will be tempted to free-ride. If we define institutions as those things that solve public goods problems, the most basic institutions will not be governments, or even formal organizations at all. Instead, the most important institutions will be norms of public good provision; things like politeness, civic spirit, benevolence and solidarity; norms that answer the question: under what circumstances will someone contribute to a public good?

The free-rider problem means that public goods can never be provided by people in groups larger than about 5 acting purely in their own self-interest. Such a group cannot form a government, for the same reasons it cannot provide the public good in the first place. It will always be worth it for someone to free-ride in a way that imperils the viability of the group. Norms, however, are a way of harnessing humans’ innate preference for altruism – in practice, their tendency to care what other people think of them – to create conditions where a strategy like “I’ll contribute as long as everyone else does too” can survive and, under the right circumstances, even support a great deal of cooperation.

Through the process of socialization into the norms of a community and – importantly – by excluding non-conforming/non-contributing individuals, a community can achieve a “cascade” of cooperation, where conformity with common norms ensures that individuals’ ignoring opportunities for defection is rewarded by larger group payoffs. Small-scale public goods become viable. Tiebout-like governance structures become viable to coordinate larger-scale public goods. The results speak for themselves: outside of the eusocial insects (who, within a colony, are mostly clones of each other), human cooperation in large groups is unparalleled in the animal kingdom.

The Godfather

The Norms of “Good” and “Bad” Institutions

In the Tiebout model, a “good institution” is a government that provides the level and mix of public goods demanded by its citizens. In the cascade model, a “good institution” is a set of norms that promotes cooperation in increasingly larger groups.

The disconnect on the immigration question is apparent. Those with a Tiebout model of institutional change see individuals on a rather abstract level, as (1) a factor of production, which makes them valuable to a community as producers and taxpayers, and (2) a set of preferences over consumer goods and public goods, which makes them valuable in themselves from a utilitarian perspective. With action thus circumscribed to the expression of preferences, there is no room for strategic behavior or opportunism – or, more properly, arenas of strategic behavior, especially in the political realm, are ignored. A society of self-interested Tiebout individuals thus “hangs together” in a harmony of interests, provided competition prevails in both consumer goods and governance.

Those with a cascade model of institutional change, on the other hand, see individuals – in addition to the two things above – as embodying a set of norms, many of which consist of a set of necessarily-common values. Those preferences which for a utilitarian are simply given and intrinsically valuable are instead the result of a socialization process that coordinates the conditions under which individuals will contribute to public goods. Preferences for public goods are not simply givens; rather, as an element of social strategies, they are inseparable from the mechanisms used to provide the public goods.

Evidence does seem to suggest that diversity2 impedes public goods provision. This is a puzzling result from a Tiebout perspective. One needs a cascade-type model to see exactly why this would be the case. Consider two ways to organize the provision of large-scale (citywide or larger) public goods: a clientelist strategy, and a bureaucratic strategy.

The clientelist strategy in public good provision – which corresponds to what I’ve called a customary exchange strategy in private good provision, or to what North, Wallis, and Weingast call the “natural state” – is quite ancient, probably originating with the neolithic revolution. It’s also extremely persistent as a strategy, though the loyalties of the particular patrons and clients may always be in flux. It’s a way of scaling up a society (i.e. the scale of public goods provision) beyond the hunter-gatherer band, beyond the point where everyone can feasibly know everyone else. A clientelist society is organized into a hierarchy of overlapping groups with personal knowledge of each other. At the top, an individual or a family that knows all the clients that report to them. Each of these people is also a patron with further clients below them. Information flows up and down the hierarchy, at each step passed between people who know each other and have ongoing obligations to one another. This structure of obligations, as well as a thick set of mutual norms, ensures against free-riding and falsifying information. Benefits flow down the hierarchy in exchange for political support against rivals. There’s a personalistic quid pro quo quality to the provision of public goods: at each level, clients cooperate so long as the patron continues to provide benefits, and the patron provides benefits so long as the clients continue to cooperate.

From the perspective of modern Western institutions, the clientelist strategy stinks of corruption. Bureaucratic polities such as our own, aided by an extensive money economy, coordinate public goods provision impersonally, using a strategy like: cooperate unconditionally, and delegate the punishment of free-riders to a professionalized bureaucracy that administers a monopoly on force according to prespecified public rules. This is, in a very general sense, what we think of as “good governance”. Without having to worry about ingratiating one’s self to a patron or keeping one’s clients happy, and within the structure of broadly liberal-democratic rules, people direct their efforts toward positive-sum production rather than the zero- or negative-sum rent-seeking that dissipates much of the surplus of a clientelist society. Such a society provides enough predictability and equality of opportunity to support not only large-scale markets and astounding economic dynamism, but also an unprecedentedly large sphere for private action where nonconformity is possible, because conformity is no longer doing the work of signaling willingness to cooperate.

Compared to a bureaucratic society, however, a clientelist strategy exposes people to much less risk of defection from others. By restricting dealings to people you personally know (often family), and by enmeshing yourself into a network of ongoing mutual obligations with those people, you can be reasonably certain that the people you deal with won’t act in ways contrary to your interests. Public goods and governance services do get provided, though much of the surplus gets diverted into various wells of obligation in the patronage network. This being a valuable target for free-riders outside the network, much effort is spent on excluding people of unknown provenance from the patronage structure.

By contrast, a bureaucratic strategy exposes people to much greater risk of defection. One must depend on the bureaucracy (1) to punish free-riders, potentially a long and uncertain process compared to direct social pressure, and (2) to adhere to its own rules in administering its force, which will not always be in its own interests. If exchange partners do not generally behave in a trustworthy manner even in situations where they could get away with cheating (i.e. if they defect, or free-ride on the general sense of trust), or if producers cannot trust bureaucrats to administer justice according to the prespecified rules, a bureaucratic strategy cannot sustain large-scale cooperation, and reverts to a clientelist strategy.

For this reason, governance by clientelism and governance by bureaucracy are often associated with “low trust” and “high trust” societies, respectively. Patronage, for all its inconveniences, is an effective and durable way to organize public goods production in a society of individuals inclined to free-ride. A bureaucratic strategy, on the other hand, only works when the presumption of trustworthiness, even in the face of profitable opportunities for defection, is in fact borne out to a sufficient degree. It requires a particular heedlessly cooperative ethos. In the absence of that ethos, cascading defection collapses the polity back into a clientelist system.

Urban Decay

Examples of Cascading Defection

The image of a cascade does not necessarily imply cataclysmic collapse. Particular norms can collapse independently of others at times, some more significantly than others. It refers, rather, to the self-reinforcing aspect of introducing defection into a high-trust society. Beyond some threshold of defection, cooperators themselves are forced either to resort to repressive measures if possible, revert to precautionary strategies like clientelism if not, or resign to nonviability.

Increased crime would be one example of defection causing a cascade: a relatively crime-free and lightly policed society is an excellent opportunity for the less scrupulous. If we take “not needing to police heavily” as the relevant public good, and voluntarily refraining from crime as the production of that public good, an increase in crime (=defection/free-riding) makes the light-policing equilibrium nonviable. The result is urban flight, and for those not able to leave, an overbearing police state.

“Voting for more welfare” is another example, employing a clientelist strategy in a bureaucratic society. Many left-parties across the Western world – and, notably, in Turkey – seem more than willing to act as the patrons of immigrant clients accustomed to such a system, to the detriment of the impersonal cooperation necessary within a bureaucratically administered liberal state.

But these two salient examples, which still remain more hypothetical than actual in most places, are hardly exhaustive of the many ways large and small that cooperative norms can be disrupted. This explains the disconnect between urban crime statistics, which have mostly been falling since the 1990s, and perceptions of danger, which seem to have risen during that same time period. This is likely no delusion. People are picking up on the normative breakdown within increasingly diverse dysfunctional urban centers in various ways large and small, even as the largest symptom – crime – fell for other reasons. Locking cars, “helicopter parenting”, increased reliance on public services rather than neighbors, lower civic engagement – all of these are precautionary strategies with the effect of minimizing vulnerability to some amount of defection that a bureaucratic society wasn’t built to handle.

It would be a mistake, of course, to lay the blame for all of this at the feet of immigration. Immigration is neither necessary nor sufficient for these problems to arise. Even so, and especially coming from clientelist societies, it can exacerbate existing problems and create new ones; problems not necessarily so obvious to a labor economist as crime, but problems with a definite impact on quality of life nonetheless.3

More broadly, the cascade model also suggests a reply to the challenge often given by immigration advocates: why would anyone try to escape from a country with bad institutions and then cultivate those same institutions in the receiving country?

In the first place, people do not usually immigrate anywhere with a clear idea of what makes institutions tolerable or not. They simply pursue their own advantage in the context they find themselves in. It is not necessary to attribute malice to the defecting immigrant; merely habit.

Patronage networks are structured to ensure that opportunism is rarely possible; that there are very few opportunities to exchange with people outside the network. In such a society, personal obligation does the heavy lifting in preventing opportunism. In a bureaucratically organized society, however, personal obligation is comparatively rare, and anonymous one-shot interactions make opportunism very frequently possible. Even if a society where no one defects is preferable to a society where anyone would defect if given the chance, a society where no one else defects (but you do) is preferable to both. This is precisely what it means to call trust a social dilemma. The generalized trust in a bureaucratic society is a commons that, absent a strongly socialized cooperative ethos, members will be tempted to deplete, even if it results in worse institutions.

James Madison

When Does Tiebout Competition Work?

Even if the cascade model of institutions is better supported empirically and theoretically for the impact of immigration from clientelist to bureaucratic societies, Tiebout competition is still a coherent story. It will be worth asking under what conditions it might still apply.

The Tiebout model lives in the world of welfare economics, which takes entirely for granted the large sphere of autonomous preferences made possible by bureaucratic governance and mass monetary exchange. It makes no presumption that the preferences of any two people in a community align. In a clientelist system, as we have seen – where conformity to common preferences and norms bears the burden of signaling a willingness to cooperate – this will not generally be the case. For this reason, a population of humans unimpeded in their exit will nevertheless generally settle into a pattern of distinct clientelist societies with idiosyncratic norms. The difference in the content of the norms will generally prevent successful migration and integration from one clientelist society to another.

The sphere of private action in bureaucratic-monetary societies, on the other hand, creates space for comfortable expression of a migrant’s taste-based norms, a fact which makes mass immigration from clientelist to bureaucratic societies seem feasible. Even so, bureaucratic societies – despite their extremely large sphere for autonomous preferences and norms – are nevertheless built on the same human norm-conformity faculties as any other society. The sphere of autonomous preferences is large in these societies – even staggeringly so, compared to clientelist societies – but it is not infinite. “Good governance” depends, in the last resort, on a common strategy of unconditional lay cooperation (that is, trustworthiness) and the bureaucracy’s unflinching punishment of defectors in order to maintain the viability of that trustworthiness. The disruption of this common strategy, especially (though not exclusively) by clientelist strategies, is what initiates cascades of defection

Tiebout competition presumes exactly that common strategy. People’s tastes for private and public goods may differ, but they must all employ the strategy above in exchange and politics.

For this reason, Tiebout competition should be a good description of (1) federalism within a bureaucratic polity, and (2) immigration between bureaucratic societies. Indeed, Tiebout competition can be thought of as a formalization of the classic case for federalism; for having states serve as “laboratories of democracy” as Justice Brandeis put it. Jurisdictional competition has been vitiated in various ways over the past century in the U.S., but even so, city and regional governments compete with each other for residents and taxpayers – not always, but often, in socially productive ways. Similarly, the ability to move from – say – an urban center to a suburb, or from a rural to an urban area, is an important expression of preferences for public goods within a society.

Niagara 1856


Thinking of institutions as norm-based solutions to social dilemmas is a causally coherent and empirically better-supported alternative to the comparatively skeletal Tiebout-Pigou-Samuelson conception of institutions as a policy mix of concrete public goods. The latter is an important aspect of migration within developed, bureaucratic societies, but it cannot be generalized into a model of the effects of immigration from clientelist societies.

Human cooperation, as I’ve argued before, is remarkable and unlikely, even at the scale of a hunter-gatherer band. The existence of planetwide impersonal markets and bureaucratic liberal governance is even more so. It is a potentially fatal mistake to take the conditions of the latter for granted. Causal models that do so are likely to overlook, and perhaps disturb, their delicate stability conditions in ways that may not be apparent until the cascade has already started.


  1. Yes, some parks charge entry fees, but bear with the example; whether or not parks are “really” public goods is beside the point.
  2. Ethnic diversity is usually the factor tested, though it likely proxies for normative diversity
  3. Incidentally, I find it especially curious that so many open borders apologists use Italian immigration as an example of successful integration. Italian patronage networks were a thorn in the side of U.S. administration for decades in the form of organized crime. Even today, the legendary dysfunction of Chicago and New York City’s dysfunctional municipal politics (and, to some extent, state politics) is a legacy of Italian-style patronage networks. To call this a success story is to overlook one of the US’ most significant historical examples of cascading defection and institutional damage, and the fact that most Americans can no longer visually distinguish ethnic Italians is quite beside the point.


ClientelismImmigrationInstitutionsNormsPolitical EconomyPublic GoodsDoug North


Facebook Twitter Reddit Threads

Leave a Reply

More Content