Collective Action

What are collective action problems?

Collective action problems are coordination dilemmas in which individually rational behaviour by self-interested actors produces collectively suboptimal outcomes.

This entails a wide range of social phenomena: not just extreme cases like riots and revolutions, but also more common occurrences like petitions, blood donation, and voluntary associations.

Why are large groups ill-suited to deal with collective actions problems?

Olson (1971) argues that large groups dealing with a collective action problem will typically fail to provide themselves with even minimal amounts of a collective good in the absence of coercion or select incentives. Any rational individual knows that their personal contribution is imperceptible, yet – because the collective good is non-excludable - they will still receive the benefits if others contribute, incentivizing free-riding.

What is the example of labour unions?

Consider labour unions. The benefits they seek – higher wages, safer working conditions etc. – are collective goods which, by definition, cannot feasibly be withheld from non-contributors. This creates strong incentives to free-ride.

To overcome this, unions rely on selective incentives: targeted, non-collective goods to either reward participants or punish defectors. Unions rely on coercive mechanisms such as closed shops, which restrict employment to union members, and often positive inducements such as insurance, welfare benefits, and social activities for members. 

Why are selective incentives not good enough?

However, this solution is tautological since someone has to pay for those select incentives, making their provision a collective action problem in itself.

Similarly, order-producing institutions such as a strong state, codified law, and secure property rights have also been proven to permit large-scale cooperation among unrelated self-interested individuals. However, in an evolutionary time frame, these institutions only came into place very recently and their existence is itself the result of foregoing social cooperation. We must therefore search for more basic mechanisms that could already generate social order in much simpler societies.

What is the importance of repeated interaction?

Axelrod (1984) shows how collective action can emerge among self-interested actors through repeated interaction. In Prisoner’s Dilemmas with a finite number of rounds, rational actors will always defect due to backwards induction, but where the probability of future interaction is sufficiently high – the “shadow of the future” – cooperation becomes a rational strategy.

In particular, reciprocal strategies such as Tit-for-Tat are highly successful: they conditionally reward cooperation but also immediately punish defection, making free-riding unprofitable over time.

How does Axelrod challenge Olson?

Importantly, Axelrod (1984) shows that cooperation can even emerge in a dominantly non-cooperative population, if reciprocal actors cluster together through non-random interaction, allowing them to achieve higher payoffs than defectors.

This therefore challenges Olson’s claim that an external authority or selective incentives are always required to induce collective action. Rather, under certain structural conditions, cooperation can arise endogenously from long-term self-interest alone.

What is the policy implication?

The policy implication is that cooperation can be promoted by making future interactions more likely, as the hope of future mutually beneficial exchange outweighs the short-term gain of defecting today. Further, parents should try and instil the norm of reciprocity into their children.

Also, it should be noted that cooperation in larger and more complex societies often relies on indirect reciprocity, where actors cooperate to maintain a good reputation so that third-parties will be more likely to cooperate in the future.

Caveat to Axelrod:

However, Axelrod’s account is best suited to situations resembling repeated exchange, such as Prisoner’s Dilemmas, where cooperation can be sustained through direct or reputational retaliation. It is less effective when applied to large-scale public goods problems, where individual contributions are hard to monitor and cannot easily be rewarded or punished through reciprocity.

What is there to say about social incentives? What is one example?

Alongside economic incentives, other social and psychological objectives – such as the desire for respect, friendship, and self-esteem - have been acknowledged as important motivators.

For example, it is well known that people give more to public goods when contributions are public than when they are anonymous, suggesting that people value status in public good settings.

Who develops this status idea?

Willer (2009) explores this further, providing concrete experimental evidence that status functions as a powerful selective incentive. Individuals who contribute more to a public good, relative to others, are judged by group members to have higher “group motivation” (a greater concern for collective rather than individual interests), which in turn increases their status within the group.

In turn, higher-status individuals exert greater influence over other group members, enjoy more cooperation, and are treated more generously by others (e.g. in Dictator and Prisoner’s Dilemma games).

How is status reinforcing?

Crucially, status rewards are self-reinforcing. Regression analyses show that status engenders positive feelings of identification, solidarity and cohesion. The higher status an individual receives for past contributions, the more their group motivation will increase. As such, individuals who were rated more highly by their fellow participants subsequently increased their contributions in the future, regardless of their actual past contribution levels.

This fits with the empirical finding that workers are more productive when they feel respected in their workplaces.

Two policy implications of respect:

These findings imply that organizations which broadly distribute respect, rather than enforcing rigid or highly unequal status hierarchies, are more likely to sustain high levels of cooperation and productivity.

[Trust] Contributions to a public good under threat of formal sanctions does not necessarily reflect concern for the group and therefore does not earn individual status. As such, formal sanctioning systems designed to encourage contributions to public goods may have the unintended drawback of undermining a group’s natural tendency to produce solidarity and identification among contributing members via status rewards.

When will social incentives be most effective?

Crucially, social incentives are only effective in smaller groups where any one member’s contribution or lack of contribution will be noticeable to others. Experimental evidence shows that face-to-face communication in a public good game – as well as in other types of social dilemmas - produces substantial increases in cooperation that are sustained across all periods, including the last.

What about large groups?

Nonetheless, social incentives can still mobilize a large, latent group if it is structured into a federation of smaller groups. The central organization can induce the small constituent groups to use their local social incentives to encourage contributions towards the collective goals of the whole group (Olson, 1971). Many solutions to collective action work simply by increasing the observability of contributions.

Who else supports this?

Similarly, Ostrom (1990), in her examination of common-pool resources, found that multiple large CPRs succeeded in solving problems of collective action through the use of nested enterprises.

For example, in the Raymond Basins, cooperation began with small, voluntary associations that enabled producers to gather and share accurate information about groundwater levels. Once these low-cost, small-scale institutions were in place, it became much easier to construct a basin-wide authority capable of setting and enforcing extraction limits.

What undermines the self-interest approach?

Recent experimental evidence has undermined the microfoundations of the rational-choice, self-interest approach by documenting people’s predisposition to reciprocity, cooperation, and punishment of defectors (Fehr and Gintis, 2007). Empirical research shows that free-riding is not the default option for a large part of the population. Rather, a high proportion of individuals act in favour of the collective interest – contributing to the common good or punishing defectors at a cost – even in the absence of selective incentives, normative pressure, or relational history.

How can we explain this?

Certain scholars have attempted to explain this reality by assuming the existence of different types of individuals. Fehr and Gintis (2007) find that 50% of subjects in public goods experiments (without a direct punishment opportunity) behave like strong reciprocators, while another 30% of participants always free-ride, regardless of what others do.

Thus, conventional explanations of collective action based entirely on either Homo Sociologicus (internalization of values inducing norm compliance) or Homo Economicus (incentives inducing selfish individuals to cooperate) are incorrect. Rather, the two types of individual co-exist and the available action opportunities determine which of these actor types dominate the aggregate level of social cooperation.

Different social structures, such as the degree to which targeted punishment opportunities are available, will generate completely different aggregate patterns of interaction.

Why does punishment opportunity matter?

Game-theoretic modelling shows that, in the absence of a targeted punishment opportunity, even a large majority of strong reciprocators (modelled by preferences for inequity aversion and reciprocal fairness) will not be able to sustain cooperation.

This is because, after observing free-riding behaviour in the first round, they will continuously revise down their beliefs about others’ expected contribution, causing them to reciprocate less and less. Further, non-cooperation is the only way to punish the free riders in the absence of a direct punishment opportunity. Thus, both the free riders and the strong reciprocators end up contributing little or nothing to the public good, albeit for different reasons.

By contrast, in the presence of a direct punishment opportunity, even a relatively small number of strong reciprocators can enforce a fully cooperative outcome by punishing defectors, creating an economic discentivise to free-riding. 

Introduce the typical perspectives on CA problems and what they predict:

Typical perspectives on collective action problems, particularly concerning common pool resources (CPRs), are generally rooted in three influential models: Hardin’s tragedy of the commons, the prisoner’s dilemma (PD) game, and Olson’s logic of collective action.

These models all predict that rational individuals will fail to produce collective benefits without some external authority coercing or offering selective incentives. Individuals in these models are perceived as being trapped in a static situation, unable to change the institutional rules affecting their incentives.

What does Olstrom argue?

However, Ostrom (1990) provide extensive empirical evidence of communities who successfully organized themselves or created their own institutions to govern CPRs without the need for coercion from an external agency.

Instead of being wrong, these are all special models that utilize extreme assumptions, rather than general theories.

Certainly, assumptions of complete information, independent action, perfect symmetry of interests, no norms of acceptable behaviour, zero monitoring and enforcement costs, and no capacity to change the structure of the situation itself help the analyst derive precise predictions.

What do these assumptions overlook?

However, these assumptions overlook many important factors in the collective action process.

For example, assuming complete information neglects issues surrounding how individuals obtain information, who has what information, and whether or not information is biased.

Rather than acting independently, individuals often consider the effects of their actions on others, and rather than being trapped in a static situation, communities can, and frequently do, change their own rules and incentive structures.

What is there to say about transformation costs?

Whereas policy models frequently assume that objective information on benefits and costs exist to be used by individuals in making choices, in reality, individuals have to invest resources to obtain this information.

Transformation costs refer to the up-front resources devoted to the process of considering a rule change itself.

These are low when institutional change is incremental, sequential, and self-transforming. Incremental change allows participants to realize intermediate benefits from initial low-cost institutional investments before they are required to commit to larger investments.

What is there to say about monitoring costs?

There are far too many factors affecting monitoring and enforcement costs to discuss here but, in general, higher costs are incurred when monitoring technology is expensive or if the rules are complex and require frequent measurements.

For example, unambiguously closing the common-pool resource for a certain period is less costly to monitor than imposing a specific quota on each appropriator.

What is there to say about norms?

Where individuals repeatedly communicate and interact in a localized setting, they develop shared norms of reciprocity and trust, generating the necessary social capital with which to build institutional arrangements for resolving CPR dilemmas. These shared norms also reduce monitoring and enforcement costs.

Two final factors worth noting:

Appropriators with low discount rates are more likely to invest time and resources into institutional maintenance. For example, if they expect that themselves and their children will be present to reap the benefits.

The likelihood of successful self-organization increases when the proposed rule changes affect most appropriators in similar ways. High heterogeneity of interests makes organization difficult because new proposed rules often tend to benefit one subgroup of another.

Why do strong reciprocators cooperate?

Introduce the puzzle:

Given it is costly for a person to sanction others, and the benefits of punishment are diffused (a public good), the act of sanctioning is generally not individually rational.

Research suggests that the motivational forces behind the punishment of norm violators have deep neurobiological roots. Neurobiological experiments have shown that the human brain’s reward system is activated when subjects are given the opportunity to punish those who cheated in a social exchange. Singer et al. (2006) shows that even merely observing the punishment of a cheater activates important components of the human reward system.

What is there to say about personal interests?

Collective action becomes possible when individual and collective interests overlap. Most formal models treat these interests as exogenously given. This assumption is misleading, because what individuals come to see as politically salient is shaped by the interplay between their own preferences and the patterns of relations in which they are embedded. The local context functions to activate certain interests and reduce the significance of others.

What is there to say about the collective good?

Any consistent course of collective action ultimately depends on the availability of a shared representation of the collective good. Formal models of collective action tend to assume the public good – the common goal for which people mobilize – as something exogenous to the model.

However, particularly with extraordinary forms of collective action such as riots, revolutions, or the rise of new political parties, the definition of the public good itself is often a by-product of mobilization rather than its starting point.

As the collective enterprise unfolds, micro-level dispositions (interests, identities, motivations) and macro-level structures (patterns of alliance, the political arena, and public opinion) co-evolve, with new issues, actors, and goals emerging over time. Consequently, neither individual dispositions nor structural conditions can be assumed as stable.

Why is network structure important?

Network structure plays a central role in these collective action dynamics. For one thing, structural differences in the distribution of social relations can lead to very different collective behaviours, despite identical interest distributions at the individual level.

This is because actors are more likely to participate in collective action when they receive consistent signals from their social environment rather than when they are subjected to cross-pressures (Marsden 1981).

What type of network is most conducive to collective action?

However, standard formal models yield conflicting results about whether dense or sparse networks are more conducive to mobilization, partly because they assume static relational structures.

A more realistic approach models social networks as purposeful and dynamic, allowing actors to adapt who they interact with and what they discuss based on past interactions and perceived similarity.

What is the dynamic model?

Baldassarri (2001) constructs a dynamic model where actors hold preferences over multiple public goods, are likely to discuss the public good that is most salient to them, and can change who they interact with over time.

When people share the same view, interaction leads to a reinforcement of their belief. Where people differ over the focal good, either compromise (reduced salience) or conflict (reinforced salience) can result depending on whether they share similar opinions on the remaining goods.

What happens when we run the model?

Simulating the model, Baldassarri finds two qualitatively distinct outcomes.

In most cases, goods are discussed at comparable rates leading to a cohesive but non-mobilizing network.

However, in very rare cases, attention disproportionately concentrates on a single public good, opinions on the good polarize, and social relations crystallize into segregated clusters, leading to perennial reinforcement and no compromise. This simultaneous polarization of interests and social ties is a precondition for collective action: neither is sufficient alone.

Meaningful social partitions are less likely to raise in the absence of polarizing goods, whilst interest polarization is of little consequence if not encoded into crystallized social identities. In sum, Baldassarri’s model shows that collective action is made possible by the simultaneity of collective identity and collective interest.

What is there to say about deliberate action?

Furthermore, these processes are driven by purposeful actors not by blind structural forces. Typical models of collective action assume that the process leading to collective action occurs behind actors’ backs. However, in reality, people involved in collective action foresee the benefit of said collective action and act consciously to achieve it.

This may involve persuading other actors, building coalitions, or modifying the group strategy and ultimate collective goal in order to recruit others who may have differing preferences.

Conclusion on dynamicism:

It follows that at the aggregate level we do not observe the simple composition of individual actions, or of interdependent actions, but the outcome of a deliberate bargaining process shaped by actors’ rational understanding of the situation.

As such, to fully capture these dynamics, we need a dynamic model in which individuals’ attitudes, social structure, and the collective interest itself are not predefined, exogenous aspects of the model but rather shaped in interaction sequences.