Thanks to behavioral psychology, we know that people, including voters, will respond to information differently depending on how it is presented. For example, it turns out that we have different perceptions about the fairness of a tax system depending on whether the information is provided in terms of percentages or absolute numbers. Armed with this knowledge, should a government present information to its citizens about policies in ways that will advance its preferred result? For example, if a policy is unpopular, should the government seek not to draw attention to it—that is, render it low salience?
Economist Steve Sheffrin offers a powerful argument that government should not manipulate salience. In my recent JLPE article, I argue that outright manipulation is off the table, but making prudent use of what we have learned about salience and human decision-making generally is appropriate, necessary, and already common practice.
The stakes are high: consider combatting global warming. If we are going to save the world, then we will need, somehow, to set a price on carbon. The price needs to be high if we want to get the behavioral response we need, but not so high so as to cause needless suffering, especially because in that case the whole initiative would not be politically sustainable. Also, in order to get the response we need, we would ideally pre-commit to a high price for a long time. And so one proposal would be to entrust the slow raising of the carbon price to a set of experts who could assess how the price is doing and adjust it accordingly.
How could our learning about salience—and tax salience in particular—be applied to this proposal? How can the proposal be presented? Further, when is it proper to arrive at a low salience tax (or regulatory) regime?
At one end of the spectrum is the view that we should use what we know to advance public policy through framing choices in a way that will get citizens to the right answer. If a certain presentation of the carbon price initiative is more likely to gain support, then that is how it should be presented. Furthermore, once in place, the pricing mechanism should then be made low salience so as not to attract opposition.
Steve Sheffrin has argued that the siren song of such manipulation must be resisted for reasons that themselves emerge from research into how we think. Most crucially, we have a deep aversion to unfair process generally.
Yet Sheffrin himself does not entirely rule out the use of framing or low-salience tax regimes. Indeed he offers an example of a property tax reform in Wales, in which the government engaged in a substantial effort to explain its reform to the public, as a paradigm of successful communication about complicated tax matters. The new system, moreover, was designed to operate in a somewhat low salience way going forward.
How, then, can one tell the difference between the acceptable and the not acceptable? I propose that there is no firm line, just an exercise in prudence.
When there can be more than one proper way forward, we are in the realm of prudence (Aristotle, Nicomachean Ethics, 6.5). Prudence is the virtue of mind associated with making better practical decisions when there are many reasonable options to choose from. Accordingly, I am calling an appropriate political salience in this context “prudential salience.” Choosing the right level of salience is itself a practice in prudence. But prudence does more work than that. “Prudential” also indicates that what the prudent designer is aiming to provide: the information a citizen would need to make a decision using her own practical judgment.
Prudence also has ethical components, or at least it has since first identified as a virtue of the mind by Aristotle (Nicomachean Ethics 6.12; see also, for example, here). Two components are important to highlight here. First, prudence in salience design requires a respect for the capacity of decision-makers, the voters, to make decisions for themselves. Not respecting the capacity of others to exercise their intellectual capacities is to undermine one of the bases for living in a political community to begin with. Second, prudence counsels a certain generosity as to the ends of procedural justice. The particular end I think most important here is legitimacy, particularly as permits accepting an outcome contrary to one’s preferences.
Let’s consider a common example of prudential salience in action. It has been an important part of the state and local government landscape since the nineteenth century that the voters need to specifically approve new debt issued by a state or local government. This separate election requirement had its origin in the hard experience of state and local governments taking on too much debt. There is also a sensible rationale for such special regimes as to debt. After all, the ability to saddle future generations with debt might seem to be too tempting an option for current politicians looking to get re-elected. Note that there is an argument in the other direction as well, namely that we owe future generations some debt. Consider what things would be like for our descendants if we did not borrow now for schools, roads, and bridges.
So now we have a fiscal provision meant to protect the public (but not too much!)—and especially a future public—from profligate politicians, but how to implement it? The electorate needs to be given a choice. If it is given too much information, or irrelevant information, then this will not be helpful. But what is relevant and what is too much?
Though the law of bond measures was not, so far as I know, influenced by cutting edge social science, it does seem to have reached a reasonable template. I will use the example of California school bonds. A bond measure involves borrowing a lot of money and with interest; the voters should know the total amounts, but arguably such large numbers are too abstract. Accordingly, a bond measure must present the terms of the borrowing on a scale meaningful to voters. The taxes are for something, and that too must be specified on the ballot. This description cannot be too long (75 words). Furthermore, the question the voters are asked is simple: “Bonds-Yes” or “Bonds-No.”
The county prepares a ballot pamphlet that includes an impartial summary of the ballot measure and arguments from proponents and opponents; these parties can choose to highlight different aspects of the issues that they think are important and likely to resonate with the values of voters. The county sends out a sample ballot, and provides translated ballots available at the polling place for members of linguistic minorities, provided they make up at least 3% of the local population.
I think it would be reasonable to disagree with the details on California bond election law, but strange to argue that it is normatively out of bounds. It would seem to pass the test of prudence. There is a presentation of information in a useful and inclusive way, without nudging voters in one direction or the other, even though many other options exist as to what should be presented and how.
Note that this is an election process that results in an automatic low-salience mechanism that raises local property taxes to pay for a long-term capital project. There is a strong independent reason, having nothing to do with salience, to organize taxes in this way. The county assessor does not increase taxes for debt service automatically in order to hide the taxes, but because this automatic procedure is an essential part of the guarantee of future payment that makes the whole exercise tenable.
This example shows, I suggest, that it is permissible for a government to communicate prudently with its citizens about major long-term decisions, and that voters, using appropriate processes, can permissibly decide to impose relatively low salience taxes on themselves if there is an independent reason to do so. As consumers, individuals bind themselves to convenient payment systems all the time. How can it be that someone can pre-commit to pay for Hulu but not schools or a livable environment when there are independent reasons beyond convenience justifying the less salient entrenchment? While procedural justice and prudence may dictate that taking such entrenching choices may well require more procedure because the stakes are higher, that is not the same as taking such choices off the table. And, as the example of bond elections demonstrate, such options have, in fact, been on the table for a long time. Returning to our example about the carbon pricing mechanism, we can now see that the end result would essentially be a different kind of long-term financial commitment that also must be entrenched in order to work, and thus an appropriate process can also be modeled on bond elections.
