Mark White’s The Manipulation of Choice: Ethics and Libertarian Paternalism
For me, the best book critiquing the “nudge” agenda is Robert Sugden’s The Community of Advantage. As he writes from within the field, Sugden engages with the most technical arguments while avoiding a romanticised view of human decision making.
In his entertaining The Manipulation of Choice: Ethics and Libertarian Paternalism, Mark White does not quite rise to that level. His three-pronged attack on nudging — the behavioural economist’s focus on preferences is too narrow, we cannot know someone’s true interests, and even if there was a way to discern their interests, we have no right to interfere with their choices to make them better — is, taken on its own terms, compelling. But by not always tackling his opponents’ strongest technical points, he leaves room for a defence of some nudges.
That criticism aside, it’s a good read, and worth working through his arguments.
My biggest disagreement is with White’s first claim — that choice is more than preferences. His other two arguments don’t depend on it, though, so I will work through those first and save the economic method for last.
Take the claim that we can’t discern someone’s true interests. White’s case for it rests largely on (sensible) thought experiments. He asks us to picture a heavyset man eating a muffin in a bakery. On first sight, we might think he should have strolled to a fruit shop. But why is he eating the muffin? Re-enacting a tradition he used to perform with his recently deceased grandfather? Celebrating weight loss from a year without muffins? To spite his wife? We can observe the behaviour but can’t know the intention behind it. Once we acknowledge that point, we can either grant the person the benefit of the doubt and assume the choice is in their best interests, or we can make assumptions about what those interests are.
I am sympathetic to this critique. Life is more than a monk-ish march to a late but boring death. Instead, it might involve extremes, choices with trade-offs or choices that someone else — or even yourself at another time — wouldn’t make.
White argues that absent knowledge of people’s interests, policy makers tend to substitute their own preferences. I agree that this is often the case, although White does not steelman this argument. Throughout Nudge — the book that serves as White’s most prominent foil — Richard Thaler and Cass Sunstein refer to the standard of “as judged by themselves”. This is an attempt to circumvent the claim of preference substitution. We could assess the “as judged by themselves” standard by later asking whether a person is happy with the choice. Do they regret it? Or we could observe what happens when barriers such as time pressure and confusion are removed. White never mentions this standard or asks whether it enables us to understand preferences. However, in fairness to White, it’s rare to see policymakers mention it either.
That lack of steelmanning is apparent in other parts of the book. For example, he doesn’t exert much effort looking for examples where it is fair to assume what a person’s preferences are. Think of “nudges” in the form of reminders about an upcoming court date to increase court attendance. I suggest it’s a fair assumption that someone would prefer to decide to not attend rather than forget. Similarly, assuming people want more money rather than less (achieved by, say, opting them into a government rebate) seems reasonable.
Absent this steelmanning, however, White’s critiques still bite. It is rare to see nudges involving an unambiguous judgement as to what is in someone’s interest. Read through the list of reports by various nudge units and you can almost always imagine people being nudged in a way that they would prefer not to be.
On this point Thaler and Sunstein would argue that a nudge does not remove choices. The nudge is voluntary. But it is here that White questions what “voluntary” means. A decision is not voluntary if it results from manipulation. It is not voluntary where circumstances are distorting. Nudges don’t pass this test.
The other limb of White’s argument — that even if we knew someone’s preferences we should still leave them alone — is largely a traditional critique of paternalism. Combined with his view on the “voluntary” nature of nudges, “libertarian paternalism” is seen as at least as bad as traditional paternalism such as taxes and bans, if not worse. At least hard paternalism is transparent.
Nudges, of course, could be made transparent. And here White suggests that “nudges won’t work if people think about them too much”. It’s not clear to me that is the case. There is literature suggesting nudges still work when disclosed. For example, Loewenstein et al. (2015) told people about a default set to sway them, and it still swayed them. If nudges still works when disclosed, this is an interesting counter to the claim that they work through manipulation. (Although to again partly concede to White, how many nudges have you seen disclosed?)
So what of corporations nudging? Here, White is more forgiving. And it is not because he sees corporate actions as any less manipulative. Rather, corporations have no power over us. If we don’t like the manipulation, we can vote with our feet and shop elsewhere. But this argument is an interesting contrast with his earlier claim that manipulation makes a choice non-voluntary. If the market is full of manipulation, is our choice to be manipulated at a different shop truly voluntary? (And that’s before asking whether the exit is even real — show me the supermarket without the lollies surrounding the till.)
His better reason to forgive corporate nudging is that we expect more of the government. We expect government to act in our interests as expressed through the ballot box. We don’t expect them to protect the interests we haven’t expressed. One way White puts it is that government should care less.
As is the case in most books critiquing nudging, White occasionally puts on some rose-tinted glasses about human decision making. White argues for presenting information neutrally, as you might see with credit card or mortgage product disclosure. He doesn’t note that no-one reads or accesses them. He suggests holding people accountable and responsible for their choices (e.g. costlier insurance if you’re tubby), without noting many still engage in those behaviours despite already bearing large costs (e.g. dying earlier with a lower quality of life). He also argues that nudges deprive people of the ability to learn, despite the lack of evidence of learning in many domains. This is especially the case for one-off decisions (e.g. a major purchase) or decisions where the consequences are felt long after the choice is made (e.g. retirement savings), domains where nudges are often found.
That’s not to say that any of these are bad ideas. It’s just that they’re not going to shift the dial much on decision making quality. A stronger argument would be to simply accept that more people will have bad outcomes, but that’s ok.
White also suggests we should help people overcome their cognitive biases and heuristics. Although not referenced directly, this thread has some similarities to the nudge versus boost debates. How can we improve decision making? Maybe it’s through creating more time (e.g. cooling off periods). However, even here the rose-tinted glasses reappear. Perhaps consumer demand will deliver information presentation from merchants and lenders that enables easier comparison? The evidence out there doesn’t provide much support.
Putting it together, White builds an attractive case. His critique of our ability to know a person’s true interests lacks the technical strength of Sugden’s The Community of Advantage, but the result is a more readable and full-throated critique of nudging.
Before closing, I want to come back to White’s first couple of chapters — his critique of the economic method itself. It is the weakest part of the book, and it contains three complaints: that preferences are too narrow to capture everything that matters, that economics reduces everything to a number, and that it ignores willpower. I flagged the first earlier. Let me take all three together.
White argues that preferences do not include “principles”, such as the moral principles of honesty and courage. Principles capture the idea that a person may not want to buy a car made by a company with poor labour practices or from a car dealer who has made racist comments. These principles can’t be measured.
White gives the following example:
Jodi normally gives $100 each month to her local animal shelter out of her love and sympathy for abandoned pets. If this choice were based on a preference, we would say that she chose to donate the $100 because the value of that donation to her is greater than its opportunity cost (the value of whatever else she would do with the money). If Jodi’s circumstances changed and she wants (or needs) to use the money for something else—or simply has less to give—that trade-off may no longer be attractive to her. She might give only $75 to the shelter next month, because she now has more pressing uses for the other $25. …
But now let’s say Jodi promised to give $100 to the animal shelter each month. Her promise is a commitment, which is supported by a principle—specifically, a moral principle—to honor commitments, and possibly also by a principle that motivated the promise in the first place. In this case, when Jodi’s circumstances change, she will be less likely to lower her donation. … This is not to say that there is no circumstance that would lead Jodi to break or compromise her promise: she might lose her job, for instance, or find another cause that she feels serves abandoned animals better. Commitments are not absolute, but neither are they subject to the small changes and adjustments that preferences are, wherein we devote a few more resources to this and a few less to that when our circumstances change.
I struggle to see how Jodi’s desire to hold to the promise can’t be captured by economic preferences. There’s nothing about preferences to say that a slight change in opportunity cost should cause her to break her promise (although I would predict that at some point she would).
Standard smooth preferences won’t give that behaviour however strongly Jodi feels, as she would shave her donation with each small change in circumstance. White pictures her holding at $100 until something big hits. But what if the promise creates a reference point, a kink where falling below it carries a cost of its own? That reference-dependent preference is part of the behavioural economics toolkit and could produce just the kind of behaviour White describes. The promise isn’t external to Jodi’s preferences but a modification of them.
White also argues that economists equate preference satisfaction with wellbeing and that “everything is reducible to a numerical value”. Here White is conflating different approaches to analysis of economic decision making.
If I am analysing individual consumer choice, there is no such assumption about what is welfare enhancing. Further, basic consumer choice analysis is ordinal rather than cardinal — there is no such “numerical value” but rather a ranking. Outside the assumptions of completeness and transitivity, nothing else is assumed about the shape of these preferences or what those preferences are over.
When we move into market analysis — say, by examining the shape of a demand curve — we normally do reduce to a single dimension (typically dollars). But this approach is not designed to analyse Jodi’s choice, but rather the aggregate choices of many people. And that is where we see the power of the economic method. In times of economic stress, the donations to that animal shelter probably go down, despite some people having “promised”. We move along the demand curve as some people change their action.
Then if we consider welfare economics, where you are considering the situation of a policy maker balancing the interests of many citizens, they often do want a common basis for that analysis. What are the welfare implications of policy X? But even that translation into dollars doesn’t mean that caring about animals or the like are ignored. They try to value everything, through ideas such as Willingness to Pay. How much would you pay to save 10,000 rare birds?
So whichever of these angles you examine — consumer, market or welfare analysis — White’s critique doesn’t stick.
White’s strangest critique of the economic approach is that it does not account for “willpower”. Fair enough for classical economics, but he levels the same charge at behavioural economics. He points to hyperbolic discounting, one of the main ways behavioural economists model self-control, and says it doesn’t capture what he means by willpower.
I can see the point he is trying to get at. Economics is good at the behaviour — the time-inconsistent choices, the procrastinating then acting. White is concerned about the mechanism, with willpower being a resource you can run down. Economics is not overly concerned with the black box in the mind.
But White then claims economics can’t explain why someone might delay for a while before suddenly doing a task — a behaviour. There are oodles of papers on precisely that (e.g. O’Donoghue and Rabin (1999)). Maybe he doesn’t like them, or has a different conception in mind, but willpower is hardly being ignored. (And believe me, the nudgers are always thinking about willpower!)
Having said all this, nothing in White’s argument hangs on his critique of the economic method. Cut the first two chapters and the argument would still hold up fine.