Institutions: What Are They Good for?



Institutions – this was the common theme in the two workshops, in which I participated during the last two weeks (in Frankfurt and Knoxville). It is clear that institutions play an immensely important role in the rise of complex societies. But what precisely do they do? And how can we study institutions theoretically and empirically? I have been thinking very hard about these questions in the last several years, and the intense discussions during the two workshops have been extremely helpful in clarifying my thinking. In today’s, and the following, posts I will try to share some of the insights that emerged from these discussions. Right away, I want to acknowledge the huge debt of gratitude I owe to other people who participated in our subgroup on the evolution of institutions in Frankfurt and to the working group on modeling institutions at NIMBioS.

A fair warning: what follows may seem somewhat pedantic. But doing such careful development of terms, concepts, and interactions/interrelations is a central part of what theoretical scientists do. This is the all-important conceptual web that is a kind of “soft cocoon” clothing the “hard” mathematical results. It’s this conceptual work that makes it possible to interpret the math and relate it to observable reality.

First, what are institutions? There are many definitions, but they tend to fall into two classes. One is “institutions as equilibria.” As an example, this is the definition used by Wikipedia (which follows Samuel Huntington’s 1965 book Political Development and Political Decay). Wikipedia defines institutions as “stable, valued, recurring patterns of behavior.”

An alternative is to think of institutions as rules. In our working group we went this route and defined institutions as “systems of interrelated rules that incentivize and structure human behavior and regulate social relations.” Without going into details, the reason I like the institutions-as-rules approach is because it makes much more sense within the theoretical framework of cultural evolution—my preferred tool for understanding how large complex societies evolved. Definitions should match theory (in fact, when building theory you start with provisional definitions and later fine-tune them as you develop the theory).


A very simple, elementary institution can have just one rule, for example, “drive on the right side of the road.” But most institutions are a combination of many rules (for example, think about the traffic rules that you had to learn when passing your driver’s license exam). When the first cars (carts, chariots, carriages) appeared on roads, there were no rules regulating the behavior of drivers, resulting in traffic accidents and jams. This was a socially inefficient outcome, so a system of rules was invented, which told drivers what to do in different situations. This system of interlocked rules gradually evolved and became quite complex, and it also adapted to technological developments.


Most importantly, it acquired sanctions imposed on those who broke the rules (e.g., a speeding ticket). This is what “incentivize” means in the definition: following rules is rewarded, while breaking them is punished. You don’t have to follow the rule, but you know that you will pay a price if you are caught.

Driving on the right side of the road is actually a self-enforcing rule. Once everybody is doing it, it is stupid to break it, because you will get killed in a head-on collision. But in most interesting institutions there is often a temptation to break a rule.

By “interesting” I mean those institutions that regulate social relations within large-scale complex societies. These rules allow them to function (more or less) efficiently without falling apart. Most importantly, they are a key element in the ability of societies to produce various kinds of public goods.

However, the production of public goods is very vulnerable to free riding. Institutions are very helpful in solving coordination problems, such as driving on the right side of the road, in which there is no temptation to defect. But institutions by themselves are not enough to solve cooperation problems, situations in which a purely self-regarding rational agent should choose to enjoy the publicly produced good, but avoid contributing to it.

Let’s illustrate this idea with the institution of democratic governance in which rulers are elected in contested popular elections. As most institutions, this one has a multitude of interlocked rules, but I will focus on one: once the ballots are counted, the losers should gracefully yield the power to the winners.

When all parties follow this rule, the result is a public good: absence of political strife or even civil war. Everybody benefits because people are not killed, property is not destroyed, and economic growth is not stunted by the conditions of political instability. But some (the losing party) face a stiff price, and they are tempted not to accept the result.

This is what we see in many countries that are nominally democratic. For example, the results of the last presidential elections in Afghanistan were not accepted by the losing party, who claimed massive amounts of electoral fraud (and they may well be right).


The two main contenders in Afghan presidential elections in 2014: Abdullah Abdullah and Ashraf Ghani. Source

Compare this to the United States presidential elections of 2000. If the Supreme Court allowed the Florida recount, these elections would probably be won by Al Gore.



Yet Al Gore and the Democratic Party did not rise up in rebellion, or even threaten insurrection. Once the final decision by the highest arbiter was in, they acquiesced to it. (In fact, Al Gore conceded on the night of elections, although he subsequently retracted it).

The difference between the two outcomes is not due to the democratic institutions of the United States being better than those of Afghanistan. In fact, I suspect that the Constitution of Afghanistan, written in 2002–4, is a better document than the American Constitution, which was written more than 200 years ago.

In the next installment we will discuss what it is apart from institutions that makes them work well or poorly.

Notify of
Most Voted
Newest Oldest
Inline Feedbacks
View all comments
Lorenzo from Oz

John P Powellson in “Centuries of Economic Endeavour” defined institution as “an accepted mode of behaviour protected by the culture”. He contrasted it with organisation. He wrote:
“An organisation can be formally established, but an institution comes about only as patterns are repeated so many times that they become respected, trusted and even demanded.” (p.6).

Using your example, Afghanistan has organised elections, but the US also has institutionalised ones.

Peter Turchin

I am planning to discuss the difference between organizations and institutions in the next installment

Lorenzo from Oz

Oops, Powelson with one l.

Justin E. Lane

I would be interested to see you unpack the claim that the Afghani constitution is better than the US constitution because it is more recent? Given how the US Constitution isn’t actively part of the US framework of governance anymore but is rather superceeded by novel concepts that stand in violation of its basic framework, it would be hard-in my opinion-to make the claim that one is better or worse when neither are actually followed. So what would be the measure of success?

Peter Turchin

“the US Constitution isn’t actively part of the US framework of governance” – I thought it was the basic law on which the rest of the US legal framework is based on?

Justin E. Lane

One would think. However, when we look at what the actual text of the document says and what is currently being done by the US federal government, it is clear that many aspects of the text, such as the 10th Amendment, are not actively in use nor do they have any judicial reinterpretation that would justify the existance of so many contemporary federal powers. Another example can be found in how the US historical actions in the “War on Terror” are implicated in regards to the 14th Amendment. However, the offering of financial support for insurrections and rebellions against those who were once deemed enemies of the state are often overlooked in regards to who is able to stand for elections.


I agree that the rules approach to institutions is the correct one, and your defense (fits with cultural evolution) is insightful. More broadly, if fits with the rational choice/game theoretic approach to society, which is analytically and empirically superior to the crude “behavior” approach.

However, it is worth distinguishing an institution from a convention. Driving on the right is a convention supported by law, and one of a nexus of conventions (supported by law and morality) concerning driving a car. An institution is better confined to organizations. An organization with clear rules enforceable by sanctions and incentives is an institution. An advertising agency is an institution, and there can be many competing and cooperating society.

Peter Turchin

I am not sure how you define convention. An institution is not only a system of rules, but also sanctions that are applied to rule breakers (“incentives”). A convention, to my mind is a much weaker word. As to organizations versus institutions, on that in the next post.

Mark Lubell

I think the instititutions as rules vs equilibria views need to be reconciled and integrated and I think work of Aoki at Stanford comes close. Also the evolution of complex and interdependent institutional systems must become central focus of analysis due to the nature of real world governance.

david ronfeldt

Thanks to hgintis for mentioning “conventions”. Perhaps it may be said that this post’s view of “institutions” has become conventional, but is not institutionalized.

While I recognize this view, I have preferred the view that uses the term “institution” to refer to formal, bounded organizations that are based on hierarchy and have leaders, management structures, and administrative bureaucracies — i.e., hierarchical institutions. This view is in the tradition of Max Weber, as well as some modern management and organization theorists. I know “hierarchy” could be used instead — indeed, some literature prefers that term, and I may yet turn to prefer it. But it is not so much the presence of hierarchy, instances of which existed in early tribes and households, as the rise of the formal hierarchical institution — such as the monarchy, the state, the army, the court, the bank, the trading company — that redirects the course of social evolution. Some of today’s economists and sociologists now prefer “organization” for this — maybe that will be what’s in the next post.

The view upheld in your post is in the tradition of Emile Durkheim and Talcott Parsons, not to mention such recent theorists as Douglass North. Yes, it views any valued, structured, habitual, rule-guided pattern of activity — such as the family, the market, democracy, voting, even popular culture — as an institution, whether or not the activity involves hierarchy (tho there is normally a law-enforcing hierarchy behind it somewhere). In this sweeping view, the four forms of organization I try to focus on — tribes, hierarchical institutions, markets, and networks — are all, in some sense, institutions (tho I’ve yet to hear a “network” called an “institution”). This view makes the concept “institution” as big and broad as “culture”.

By the way, all four of those forms of organization — tribes, hierarchical institutions, markets, and networks — have definitional problems, as you surely know. The concept of “tribes” is disputed if not denied by many anthropologists, who may then uphold an alternative term that is still about kinship dynamics. The concept “institutions” has issues like those discussed here. The concept “markets” is less disputed, but I’ve experienced lots of economists who would subject all the above four forms to economic analysis, but don’t really study and maybe can’t even give me a good definition of a market. As for networks, when I started being interested in them decades ago, they were viewed mostly as a distinctive form of organization, as I like. But now, with the rise of social network analysis and network science, every organizational form above is viewed as a kind of network.

Anyway, I’m pleased and interested to see posts discussing institutions and organizations with reference to social/cultural evolution. Onward.

Peter Turchin

That’s right, I will explicitly talk about the difference between organizations and institutions in the next post. Worth noting that it’s OK for us to disagree about definitional issues. Definitions need to be internally consistent within a single research program, but may differ (in fact, always do) between programs. No point in trying to impose definitional totalitarianism!


In fact, there are many meanings for the term “institution.” All are plausible things to study. It is just important to be clear as to which one you are talking about.

Mike Alexander

I have a suggestion. Would it be possible to establish a permanent discussion thread on specific items that would remain open and active over the long term? As the first of these, I would nominate this thread on institutions, how to define them, both theoretically and operationally (i.e. how to measure them) and finally how to model them.

  1. Home
  2. /
  3. Cliodynamica
  4. /
  5. Regular Posts
  6. /
  7. Institutions: What Are They...

© Peter Turchin 2023 All rights reserved

Privacy Policy