The distinction between “exit” and “voice” as the two main ways for individuals to protest when they perceive a decrease in the quality of a good or service is well-known since Albert Hirschman’s essay Exit, Voice, and Loyalty.[1] What tends to be retained from Hirschman’s account is that while exit mechanisms are well-understood thanks to the work of economists since Adam Smith, voice mechanisms have largely been neglected, especially by economists. However, in many cases, especially with respect to public services, exit is not an option and voice is the only way for users to express their discontent.
Among the key ideas that emerged from Hirschman’s account, two, in particular, have had a long-lasting impact, at least in my view. First, it is often said that exit and voice belong respectively to markets and politics. Exit mechanisms are constitutive of the functioning of markets; unhappy buyers, or even those that think that they can be offered a better good or service elsewhere, just switch from one seller to the other. Voice mechanisms are rather by nature political. They are more conflictual and confrontational and can be associated with deliberative aspects of political processes. Second, voice tends to be viewed as more informative. This is interesting in light of the classical Hayekian argument for the superiority of market mechanisms over planning. The price system sends signals to buyers and sellers and incentivizes them to adjust their behavior at the margin. It does so by virtue of the information it transmits, especially about people’s preferences and firms’ costs of production. Nonetheless, voice mechanisms transmit information about the reasons for the decline of the quality and individuals’ discontent, something that the price system cannot do.
This simple framework can be complexified in several ways. Two types of questions can be asked on this basis. The first set of questions relates to which mechanisms are superior in a given context. The second set of questions relates to the different ways exit and voice mechanisms can be, or actually are combined.
I think it is fair to say that, in the same way as economists have tended to be oblivious to voice mechanisms, sociologists and political scientists have excessively underestimated the importance and power of exit mechanisms to understand long-term institutional changes. At the risk of an overgeneralization, I think there is in the social sciences (putting economics apart) a romantic if not mystical understanding of voice mechanisms and their ability to trigger long-enduring institutional changes. I think – but this is definitely impossible to prove – that over the long run, exit mechanisms are far more influential to understand institutional change, not only in the market but also in other areas of society. The rise and decline of institutions, of nations, if not of empires, is largely determined by individuals’ choices to contribute to these institutions. Exit mechanisms often are associated with the fact of “voting with one’s feet”, which is perfectly sensible. But voting with one’s feet does not necessarily mean choosing to leave physically. It can also – and most often will – consist in choosing to dedicate less effort to a particular task or to revising one’s tradeoffs between two or more objectives. In other words, exit is not necessarily about leaving (a supplier, a territory). It is also about updating one’s behavior according to changes in opportunity costs. My intuition is that, over the long run, this is this kind of marginal adjustment that has the most long-lasting impact on institutions.
As Hirschman recognized, exit and voice mechanisms can work together. In particular, one’s ability to make oneself heard by public or private authorities is enhanced if one has credible exit options. There are many trivial examples, including in professional sports. Back to the time when he was playing for the Cleveland Cavaliers in the 2010s, Lebron James was able to impose his will with respect to the composition of the coaching staff. No doubt this has been largely facilitated by the fact that had he not obtained what he wanted, he would have just left at the end of his contract or even pushed for a trade. In general, voice mechanisms are more effective if individuals have valuable exit options. This is already an indication that associating politics and especially democracy exclusively to voice mechanisms is a mistake. As it happens, the dependence of the effectiveness of voice mechanisms on exit options is a standard argument advanced by proponents of more polycentric forms of democracy.[2]
At first sight, exit seems less dependent on voice. Whether you have or not the ability to express your discontent doesn’t seem to affect the value of your exit options, if you have any. However, there is one way where voice mechanisms can be relevant for exit mechanisms that I rarely find mention. This applies in the special case where the value of exit options is variable and a function of the number of persons who effectively choose these options. Let me take a simple example. You might be unhappy with the way Twitter/X has evolved over the last year. Many people have complained about some changes in the way Twitter is working – often through Twitter itself. The problem here is that voice mechanisms are relatively weak precisely because users don’t have many valuable exit options. Is that because there is no alternative to Twitter? Not exactly. The problem is rather that the value of a social network for a single user is an increasing function of the number of users. If you don’t expect many other individuals to switch from Twitter to another platform, then the value of your exit option is low (or, relatedly, its opportunity costs are high). If Elon Musk doesn’t expect that many users will move Twitter to another platform, then he may rightfully ignore the complaints and the threats of unhappy users.
This reasoning basically applies in all cases with positive externalities, particularly network externalities. Such cases happen in markets where the value of a good or service depends on the number of users, but also in the political domain. You’re unlikely to want to move to a new jurisdiction if you’re the only one to do so, simply because the value of following some rules at least partially depends on the fact that you’re not the only one to follow those rules. In general, the perceived value of exit options is related to threshold effects of the following form:
Exit option O is valuable enough for individual I to switch from the status quo S if and only if I expects that at least N other individuals J will also switch from S to O.
This is a classical coordination problem. As we know since the work of David Lewis on conventions,[3] though simple in appearance, solving coordination problems may sometimes be difficult because it requires that individuals have common knowledge (or belief) of what each other will do. This is what seems to be required here: for I to expect that J will switch, I must believe that J believes that enough other individuals (eventually including I) K will switch. But that implies that J believes that these individuals K believe that others will switch, and so on. In principle, there is no end to this chain of iterated beliefs. There have been a lot of discussions in economics, philosophy, and elsewhere over whether common knowledge is necessary or feasible,[4] but the point is that people need a way to form higher-order expectations about what others believe everyone else will do to assess the value of exit options.
I would suggest that voice mechanisms are such a way. By voicing one’s discontent, one expresses his conditional willingness to switch to another option that would better satisfy their demands. When this voicing is public (e.g., made on a public platform), not only everyone else knows that you’re dissatisfied with the state of affairs. You also know that everyone knows, and everyone knows that. In other words, when publicly expressed, discontent becomes common knowledge. For sure, this is not enough to trigger the switch from the status quo to an exit option. Even if it is common knowledge that everyone is unhappy with the status quo, that doesn’t imply that everyone will indeed move from the status quo. But voice can at least help to make some exit options more salient and strengthen the expectations that a large enough number of individuals will switch. This number does not need to be large. If some individuals start to switch it can trigger a cascade effect where more and more, observing the change of behavior (a typical exit phenomenon) will in turn go for the exit option. As I say above, exit is ultimately the force behind institutional change. But voice can be its trigger.
[1] Hirschman, Exit Voice & Loyalty – Responses to Decline On Firms Organizations & States (Cambridge, Mass: Harvard University Press, 1990).
[2] Julian F. Müller, Political Pluralism, Disagreement and Justice: The Case for Polycentric Democracy, 1er édition (Routledge, 2019).
[3] David K. Lewis, Convention: A Philosophical Study (John Wiley and Sons, 2002).
[4] On this, I very much like the book of Michael Suk-Young Chwe, Rational Ritual: Culture, Coordination, and Common Knowledge (Princeton University Press, 2003). It suggests that common knowledge is far more important in our cultural practices than we may imagine at first sight.