Endogenous preferences, addictions, and paternalism
If our preferences are the results of our experiences, does consumer sovereignty make any sense? I say it does, with some caveats.
The other day a few of us on X had a discussion about endogenous preferences. The basic idea is that what economists call preferences are not necessarily innate and may change over time, in response to environmental stimuli. Why don’t economists talk more about this? One answer to that question is that maybe we are too wedded to our liberal/libertarian anti-paternalistic tradition, and preference endogeneity does not sit well with it. After all, in welfare economics we like to evaluate outcomes by taking preferences as given. If those preferences could not be taken as given, than what remains of traditional (indeed often non-paternalistic) welfare economics?
I think the answer lies in how exactly we want to endogenize preferences. One way to do that, and despite its limitations this is my favourite approach, is by applying the rational addiction model of Becker and Murphy. To many, this model just sounds downright craczy, but actually, as pointed out in this X post, it can make sense of changes in preferences.
Let’s say you consume two goods, x and y. Out of the two goods, the consumption of x is prone to habit-formation or “addiction”, in that past consumption has a (positive) effect on present consumption, and present consumption on future consumption. Formally, consider utility at any time t
where X is the stock of past consumption of x. Complementarity of past and present consumption implies that
That is, past consumption increases the marginal utility of consuming x in the present; the two are complements. Now you might notice that X, the capital stock, also has a direct effect on utility. In that, U is a “meta-utility function”, where we can evaluate not only the change in the (sub)utility of x, but in a way also the process of the preference evolution itself. When it comes to the latter evaluation, it is entirely possible that even if X increases the marginal utility of x, it also decreases overall utility, so that
One example is the consumption of “harmful” drugs. Consuming more today “changes my preferences” in the sense that I will crave it more in the future. At the same time, the stock of cosumption (what we can call by the name “cosumption capital”) over time takes a toll on my health (another capital stock, btw). Whether we will get addicted depends on how much utility we get from the present satisfaction of our cravings versus how much we take into account the deterioration of our health and other well-being in the long run as a result of the addiction. The addiction is said to be “rational” if we take into account the future effects as well; in that case it can be modeled as a standard dynamic optimization exercise. For instance, the rational model predicts that if the price of say cigarettes is expected to increase, then consumption might fall even in the present. Guess what? This is exactly what happens in these market! So, the rational model, while it has limitations, predicts well in a number of contexts. Empirical studies also show that drug-addicts are willing to come off the drug if they receive sufficient financial incentive to do so. Thus, they are not as irrational and limited in autonomy as most people believe.
We may thus already see why allowing for endogenous preferences is compatible with a classically liberal, anti-paternalistic worldview. Even addictions might be understood to be “freely chosen”, with an eye to the future effects, and if addicts do not cease their addiction, it is because disinvesting from their accumulated consumption capital is too costly. They would probably like to stop doing harmful drugs (they might not like their current preferences), but given their existic consumption stock, they may be better off not stopping.
But here comes the hard part: what if people get “bad preferences” not through their voluntary adult choices, but due to childhood exposure to bad habits, being abused by their parents and other adverse childhood experiences (these can all increase the consumption capital stock)? Then the idea of consumer sovereignty might indeed be compromised. Gary Becker thought about this problem but couldn’t come up with an entirely satisfying solution. However, note that in a rational addiction framework, even in these cases, applying coercion makes the lives of the people with bad habits worse, as their consumption choices are optimal given their consumption stocks. So far, therefore, we still have reasons not to be paternalistic. However, we would also like to prevent the development of bad habits in childhood, as children are generally not regarded as competent decision-makers. And here comes the really hard dilemma for the anti-paternalist: how do we keep children away from being exposed to drugs, alcohol, porn, harmful social norms, etc. while allowing competent adults to engage in bad habits, given that for them it might be “too late” to quit and be made better off for that? For instance, to cite an example very different from the drugs or cigarettes examples, it would probably harm women who veil themselves in Muslim cultures if we banned doing so, given their current stock of accumulated “religious capital”; but many people might say that young girls should be given more exposure to less restrictive norms and habits, so they would end up accumulating very differernt consumption and cultural capital stocks.
Obviously, blanket prohibitions are a blunt instrument. But taxes on bad habits, nudges, or in some cases limits on advertising them could perhaps make sense. But where exactly should we draw the line? I think this is a question we should grapple with from time to time, as sure we haven’t yet solved it. Myself as a classical liberal tend to side with anti-paternalism most of the time, as I believe governments often make problems worse, not better. Still, the questions remain, and they are important.