On the Epistemic Legitimacy of Government Paternalism

On the Epistemic Legitimacy of Government Paternalism Abstract Some contemporary paternalists argue in favor of government interventions based on how experimental psychologists and behavioral economists have found that our behavior often diverges from what would be predicted by rational-choice models. In this article it is argued that these findings can, more specifically, be used to identify decisional trouble spots where paternalist interventions may be legitimate. It is further argued that since the epistemic legitimacy of government paternalism ultimately rests on centralized decision-making having a comparative advantage, it also depends on the possibility of such interventions being governed by an ideal of evidence-based policy-making. The article asks how stringently this requirement should be understood, and to what extent government can legitimately engage in what might be called experimental policy-making of a paternalistic character. The past 10 or 15 years have seen the emergence of paternalistic approaches to public policy that rely heavily on contemporary empirical psychology in understanding our everyday choices on matters of health, wealth and safety. These paternalists might differ on which means that should be used, e.g. Richard Thaler and Cass Sunstein (Thaler and Sunstein, 2003, 2008) defend what they call libertarian paternalism and advocate nudges (setting up the choice architecture we face so that we tend to choose in a way that is beneficial over the long term), while Sarah Conly (2013) defends what she calls coercive paternalism and advocates more clearly restrictive measures. But they are united in pointing to how dual-process models in psychology highlight how many of our everyday choices are based on heuristics and biases, thus often leading to suboptimal results, an observation which can be taken as grounding the epistemic legitimacy of government paternalism: that policy-makers might sometimes be in a position of knowing better, being able to make rational decisions that we as citizens fail to do. In general, this new wave of paternalism has however tended to focus strongly on the decision-making of us as individuals, which is only one side of the kind of comparison that you need to make if you are going to argue that government paternalism has what might be called epistemic legitimacy, i.e. that centralized policy-making has an edge in terms of knowledge and rationality over the distributed decision-making of us as individuals. For instance, one literature review of articles in behavioral economics found that of 67 articles that made interventionist policy recommendations, only 2 seriously considered the possible irrationality of policy-makers (Berggren, 2012: 216). Of course, in making policy recommendations, one is forward-looking and concerned with how the policy-making process should function, so while it is important to consider how policy-makers generally function, the key issue concerns the extent to which would-be regulators in the relevant area can realistically be expected to meaningfully be governed an ideal of evidence-based policy-making. Such an analysis needs to be done before endorsing concrete policy recommendations because it can decide which recommendations that make the most sense. This article will put forward a basic framework for making comparative analyses with respect to possible public health interventions. Comparative advantage and the possibility of clustered failures Before proceeding it should be noted that paternalistic policies are rarely paternalistic in the same clear-cut way as paternalistic actions in, say, the physician–patient relationship (cf. Wilson, 2011). To begin with, since public health interventions tend to be designed to lower the rates of adverse health consequences that come with certain types of behavior, there is a strong link to what our healthcare system needs to do and the costs of what it does. Healthcare systems in most countries tend to be publicly part-financed, and this means that if a public health intervention is successful, then qua taxpayers we all benefit, and public health measures that involve taxing certain forms of consumption do not just steer behavior, they also raise revenue and thus enable us to spend more on public goods: again, we all benefit. On many standard formulations of what it means for something to be paternalistic, there is a requirement that it is done simply for the benefit of the person intervened against (Dworkin, 1972: 65). Indeed, Mill’s harm principle (1989: 13) does not say more than that a person’s ‘own good, either physical or moral, is not sufficient warrant’ for interfering with someone. Pure cases of paternalism of this sort are exceedingly rare at the level of public policy, even if we just focus on the intentions behind policies rather than on their effects. Still: many public health policies often have paternalistic elements to them and, to the extent that they do, they should concern choices where there is a comparative advantage to effectively make those decisions in a centralized rather than distributed way. When doing this comparison, then, if we look at us as individuals, there is by now a vast literature in empirical psychology and behavioral economics about how we as agents are far from the rationalist constructs assumed by many traditional economists as a basis for their mathematical models of behavior and equilibria. But even if we accept this kind of picture of human decision-making, we need not think that it points to a need for paternalist interventions. To begin with, on the level of individuals, it is quite possible that when we move from experimental setups to normal circumstances, our heuristics and biases are, at least on the whole, just efficient ways of making reasonably good decisions. We might think that satisficing is a reasonable strategy in everyday life (Simon, 1956). But this does not preclude that there are still certain types of choices we make on an everyday basis that are clearly counterproductive to our interests and that should be regarded as irrational not just in a narrow sense of the notion.1 It might also be objected that even if there are choices like these, they still need not be suitable objects for interventions. At least over time, our choices might either even out or at least not add up to much; we can in principle have a long series of irrational choices without the net effect of them being particularly serious. It lies in the nature of decision-making shortcuts that they can often be efficient ways of making decisions, but still sometimes lead us astray. We accordingly need to distinguish between two ways in which we can systematically fall short of ideals of knowledge and rationality. First, individuals might have frequent but dispersed failures, i.e. they might regularly fail to live up to the ideals but in a very wide range of ways and for lots of different reasons. In this situation there is little reason to think that centralized decision-making would improve on individual choice. But we can also have strongly clustered failures of rationality in identifiable problem areas. These would have to do with types of choice situations where several potentially problematic tendencies will all tend to pull us in the same direction, toward choices that are not just not the best, but also not even good enough. Four problematic tendencies The relevant literature in psychology and behavioral economics identifies a number of heuristics and biases. Some of them, such as confirmation bias, might however lead to relatively dispersed failures, i.e. failures which take us in many different directions and might possibly even out on a population level. A strong case for public health interventions would instead involve identifying several problematic tendencies that cluster in an area of decision-making and increase the risk of making a certain type of bad choice. In a public health context one common type of focus is on long-term health impacts of lengthy sequences of small everyday choices, where the risk increases from a single action undertaken here and now might be very slight, but where long-terms effects are significant. In these types of cases there are at least the following four tendencies that might cluster and lead to rationality failures in our decision-making:2 (1) We tend to suffer from temporal myopia. Potential future gains and losses are often more uncertain than those that are immediate or imminent, so if the two sorts are equal in magnitude, it seems prudent to value them differently, and to discount the former. Already in the 1950s, Robert Strotz (Strotz, 1955-1956) speculated that we have an in-born discount function, and later experimental studies have indicated that this is indeed the case (Shane et al., 2002; Benhabiba et al., 2010). Note that the bare operation of a discount rate is not irrational—far from it. Problems arise when the discount is the product of a relatively blind heuristic rather than a measured response. For instance, maybe the difference a day makes means that some people would rather choose a certain reward today than a larger one tomorrow, but then what are we to say about the same choice when it is made between outcomes that are a year away or a year and a day away from now? If it was just a matter of the difference a day’s waiting time makes, the day should make the same difference here as well, but this turns out to not be the case (Thaler, 1981). More recently, neuropsychological results have indicated that different parts of the brain are active, depending on whether we are thinking about present or future rewards (McClure et al., 2004). (2) We tend to have problems with differences in orders of magnitude. While the kinds of decisions that are relevant in a public health context often require us to consider temporal distances, they also often involve very different orders of magnitude. Already in the nineteenth century, Ernst Heinrich Weber had discovered that human responses to physical stimuli were more discriminating when the magnitudes of those stimuli were small; the bigger the magnitudes, the bigger the differences between two objects must be in order for us to notice the difference between them. In recent years, neuropsychologists have observed the same phenomenon in connection with abstract objects like numbers (Nieder and Miller, 2003; Longoa and Lourencoa, 2007). The mental number line gradually becomes more compressed, so that accuracy is indirectly proportional to the size of the numbers involved (Dehaene, 2003). Of course, in everyday life we mostly deal with small numbers, so there is no mystery about why we have evolved to be more accurate there, but this also means we there have inherent difficulties in making everyday decisions that relate both to the relatively small payoffs that we get in the present and potentially much larger effects that lie in the future. (3) We tend to be overconfident and excessively optimistic. When it comes to our possibilities to steer clear of future problems, we are in part dependent on our abilities and in part on the likelihood of different external events. In both of these areas people tend to have overly rosy pictures. For instance, with respect to self-assessments of one’s general ability levels, in one study of American drivers, 93% thought that they were better than the median driver (Svenson, 1981) and in a study of professional fund managers, 74% believed that they were above average in their job performance and 26% that they were average (Montier, 2007). When it comes to the likelihood of external events happening to us rather than others, we exhibit similar patterns. We tend to overestimate the relative likelihood that good things will happen to us and underestimate the relative likelihood that bad things will happen to us (Weinstein, 1980). As long as being only moderately delusional, these tendencies might perhaps often be a helpful coping mechanism (thinking that one is below average at one’s job is surely a depressing thought), but there is certainly a possibility that such tendencies will at times lead us to not take certain risks seriously enough. (4) Our preferences are often unstable and indeterminate. The standard model for decision-making centers on a notion of expected utility in which our preferences determine the basic utility values involved and the probabilities of achieving the objects of our preferences give us the expected utilities of all the options available. The problem is that our preferences do not appear to be determinate enough for this notion to be applied; instead we are highly sensitive to circumstantial factors, and several studies have shown that, depending on how one attempts to elicit the preferences, we make different choices, even when the methods of elicitation should be normatively equivalent (Grether and Plott, 1979; Tversky and Simonson, 1993; Slovic, 1995). We display preference reversals in which given one method of elicitation a subject will choose A over B, and given another method she will choose B over A. At least for many preferences we seem to partly make things up as we go along rather than drawing on a pre-existing preference order. This might actually be an economic way of handling many decisions, but it also means our preferences in the moment cannot always reliably be used to determine what we really want. When looking at where the comparative advantage lies it is an assessment that is impossible to do with complete precision, but having identified these tendencies, we can look at relevant choices or areas of decision-making and judge the extent to which there is a likelihood of clustered failures. There is no room here for deciding whether this is a definitive list of what we should be looking at, but it is a basic model at least. With respect to people’s ability to rationally make certain decisions, we would then ask these four questions: Is there a significant temporal distance between the relevant action(s) and possible long-term effects? Is there a significant difference in orders of magnitude between the short-term effects of the relevant action(s) and their long-term effects? Is there significant room for wishful thinking in relation to long-term effects? Is there significant vacillation or reversals in the relevant preferences or judgments? These dimensions are certainly not entirely independent of each other, but neither do they completely collapse into each other. If we strongly answer all of these questions in the affirmative, there is reason to distrust that people will know what they are doing in relation to the area of decision-making in question. This would then point to a need to take the possibility of making paternalist interventions seriously—although before we conclude that such interventions really are epistemically legitimate, we must also look at the side of policy-makers as well. The first three dimensions are of the same kind in that they involve tendencies which, taken singly and with respect to many everyday decisions, are probably evolutionary adaptive ways of thinking. They can however become problematic when they cluster, especially in a modern context where day-to-day survival is not really a central concern and where long-term health effects become more salient. The fourth dimension is different and plays a dual role. To begin with, it could in principle be the case that the first three questions are answered in the affirmative, but people in the end still tend to firmly and stably endorse the choices they made, even retrospectively. If this is the case, it becomes questionable to understand these choices as irrational or deeply problematic. But then it is also the case that to the extent that there is vacillation or reversals in our preferences, these preferences become less viable as a standard of what counts as a good outcome, and it becomes more reasonable to think instead in terms of generic goods, such as expected gains in life years, when evaluating outcomes, i.e. goods that more or less everyone can non-controversially be assumed to be better off in having more rather than less off.3 This is the kind of goods that public health interventions ultimately tend to aim at, so to the extent that outcomes for individuals are difficult to assess in the light of their precise preferences, this further opens up for centralized decision-making. Since this article is mainly about working toward a model for making certain comparisons relevant to possible public health interventions, there will not be that much room to look at concrete examples, but let us briefly look at two, albeit mainly as illustrations rather than full-scale applications. First, a relatively straightforward case: seat belts. Seat belt mandates tend to be broadly accepted, even though the choice to forgo seat belt would seem to be relatively mundane. The above model does however identify it as a problem area. Both (2) and (3) definitely seem to obtain. When it comes to (1), the temporal distance to possible ill effects is highly variable, but it is typically well past the horizon of the particular decision-making situation. It also seems like an area where it would be reasonable to expect regret in the face of those ill effects (if those affected are even in a state to have a preference reversal). In sum, when looking at choices to forgo using a seat belt, it seems reasonable to regard these as highly likely to be the product of clustered failures of rationality, and hence conclude that this is an area of decision-making where it might be quite possible for policy-makers to do better. A second possible application is about consumption choices, such as consumption of junk food, that can lead to obesity, which in turn carries with it an increased risk of type 2 diabetes, hypertension and cardiovascular disease, certain forms of cancer and an adverse impact on quality of life (Hruby et al., 2016). With respect to questions (1) and (2), choices to consume junk food clearly seem to be ones where the answer is ‘yes’. Especially with respect to our early habit-forming choices, there is a significant temporal distance to the long-term effects; the short-term pleasures and the long-term adverse health effects do also seem to lie on different orders of magnitude in terms of the values involved. Question (3) is not as straightforward. There does not seem to be much room for wishful thinking when it comes to the relation between certain habits and obesity, perhaps apart from the fact that obesity is the product of many years of certain behaviors and there is always room to think that one’s current behavior is really just temporary. It is however not really obesity that is the problem, but the health consequences that come with it, and since the connections here might primarily be visible on an epidemiological level, there seems to be ample room for individuals to think that they will be among the lucky ones. Question (4) is a more complicated one in this particular case. There are clearly negative emotions often associated with obesity, but on the other hand, those might also be linked to a stigmatization which is not only morally problematic but might even result in more of the behavior that causes obesity to begin with it (Puhl and Heuer, 2010). Arguably this is however still an area of behavior where there is often significant vacillation in people’s preferences, e.g. between consuming more junk food in the moment and feeling that one needs to cut back on it at other times. Such vacillations point to it not being an area of rational decision-making and add to the picture that when it comes to high consumption of junk food, it can be seen a type of behavior that grows out of clustered failures of rationality. Policy-making and the role of evidence The ideal of evidence-based policy-making first came to prominence in the late 1990s (Parsons, 2002). The British New Labour government was an early proponent, with Tony Blair affirming a ‘commitment to policy-making based on hard evidence’ (Cabinet Office, 2000: 3). Up to a point, more or less everyone should be able to agree that politics cannot merely be about ideology and that one must also consider the expected effectiveness of proposed measures. The interesting question, however, is just what we should demand in the form of evidence. Where public health interventions are concerned, David Resnik (2014: 174) has suggested that there should be ‘substantial evidence that the restrictions are likely to be effective at addressing the health problem’. While one might discuss just when evidence is substantial and what likely means, this requirement is clearly aimed at setting the bar relatively high. Its adoption would make experimental policy-making highly problematic. This might very well be the point, but at the same time we need to remember that decisions impacting on public health are always already being made by us as individuals, and the basic question here is not about the absolute level of competence at which it is permissible, or advisable, for governments to make these decisions. It is about the comparative competence of the state and individuals. It is also important not to be misled here by an analogy between evidence-based policy-making and evidence-based medicine. The latter preceded the former and has had an enormous influence on how we now think about best practice in medicine (Howick, 2011; Solomon, 2015). Although it might be difficult to prove conclusively, it seems reasonable to think that its success has been a strong contributing factor in the ascent of the ideal of evidence-based policy-making. Already from the outset, however, there are difficulties in moving ideals from medicine to policy-making. For instance, while randomized controlled trials are usually seen as the gold standard in evidence-based medicine (Timmermans and Berg, 2003) and are comparatively manageable to perform in medical research, government policies are another matter. Policies are not enacted in controlled settings, but rather in the wild, so to speak. The number of potential confounders in that arena is much greater and the intricacies of policy implementation may mean that what on the surface looks like the same policy being implemented in another place will be a policy whose implementation is unique, or at any rate different, in important respects. While the results of rigorous medical research done in one country can often be transferred to other countries, transferring results from a policy evaluation will be more difficult simply because there are so many background factors that might play a role in making a policy work or fail. These factors often vary greatly from country to country. Something might work there, but not here (Cartwright and Hardie, 2012). Conversely, something might not work there, but still work here. Another problem with the analogy has to do with the underlying rationale for setting demands for evidence at a certain level and the extent to which demands for evidence tend to privilege inaction over action. The human body is a self-repairing system which, even when it is left alone, tends to return from sickness to a state of health (Cochrane, 1972: 5). Society is not quite like that. To the extent that there are societal states of equilibrium these are products of contingent complexes of institutions that shape the interactions between members of society. This means that while in medicine there is often a real choice between treatment and non-treatment, in politics we are often left with a choice between affecting people in certain ways or affecting them in certain other ways. Doing nothing usually still means doing something, and sometimes neither approach is backed by hard evidence. We should be careful not to fall prey to the illusion that simply acquiescing in the status quo is equivalent to not affecting people’s lives. For instance, the way the contemporary food industry operates is not just a natural fact, but is already the product of a contingent set of institutions and regulations. Finally, it needs to be remembered that at least with respect to public health policies, measures often target behaviors with strong inertia. This means that such policies usually face an uphill battle and that often it will require a package of interventions to really make healthy behavior the easy choice (cf. Gostin, 2015). For instance, many industrialized countries have now for decades pursued comprehensive tobacco control programs, and there appears to be strong evidence that these programs ‘have a dramatic impact in reducing the proportion of adolescents who start to smoke’ (Pierce et al., 2012: 262), but the comprehensive programs are packages of policies, such as increasing the price of cigarettes, requiring warning labels on packs, restricting advertising, introducing smoke-free areas, restricting accessibility by age and various kinds of school interventions. While it is certainly possibly to evaluate such policies one by one, it is also important to consider how they might function as parts of a package, where the joint effect of making it difficult enough to start smoking is what might in the long run produce the sought-after shift in behavioral patterns. Now, the argument here is that what is needed for epistemic legitimacy is both that there are clustered failures of rationality with respect to a specific area of decision-making and that policy-making with respect to that area rises to the level of being about evidence-based policy-making rather than just a matter of guesses and hunches. The idea is not that it must be possible to conclusively determine exactly which policy or mix of policies that would be the best, but simply that there is enough of evidence-basing to be able to make rational policy choices in an area of decision-making where clustered failures of rationality are likely on the level of individuals. With respect to tobacco control programs, the picture that emerges seems to be that a reasonable level of evidence-based policy-making is possible. There might still be room for discussion about exactly which policies or mix of policies are the best ones, but in the comparative analysis what is important is that a certain threshold of rationality in policy-making is clearly passed. If we look at the current state of policy-making in relation to obesity, it is similar to tobacco control in that there does not seem to be any one policy that will single-handedly provide the magic bullet that halts the rising levels of obesity and associated health problems. Many policies or suggested policies (a good example being former Mayor Bloomberg’s proposed ban on large sodas) are mainly about tweaking behavior, and while there is some evidence that some such tweaks, such as taxing sugar-sweetened beverages (Colchero, 2016) have some effect on behavior, substantial shifts in broad patterns of behavior are only likely to be accomplished by comprehensive packages of policies. So a lot of work remains to be done, and here a potential problem presents itself. To get good evidence on which policies that work, we need to enact these policies somewhere. With respect to obesity and its associated health effects, there is accordingly a significant need for experimental policy-making now for policy-making to become strongly evidence-based in the future, so the question becomes one of what we should demand for something to be a legitimate area of (at least partly) experimental policy-making. Even if there are clustered failures in that area, it still seems reasonable to require some positive qualities from such policy-making. The suggestion here is that we can frame this in terms of a demand that it should be done in a way that allows it to be meaningfully governed by an ideal of evidence-based policy-making. Here are two key dimensions to focus on in determining the extent to which this is the case. (1) Evidence-driven prospective decision-making. Given that the existence of adverse public health consequences as the likely product of clustered failures have been established, the most significant aspect of our decision-making process will be in making predictions about the likely effects of different policy measures. There might be further moral and political reasons for preferring certain such measures, presumably less invasive ones, but effectiveness and efficiency are the main concerns when setting those other reasons aside. It is important to then keep in mind that evidence-based and research-based do not mean the same thing. Much research is oriented toward understanding the mechanisms underlying what happens, and armed with an understanding of such mechanisms we might make predictions about what ought to happen in a system given a certain stimulus. The models on which such predictions are made should, of course, be based on evidence supporting our belief in the relevance of certain mechanisms, but in the end the relation between this evidence and the predictions we make will be indirect. While these predictions may not be evidence-based in a strong sense, they can still be research-based and evidence-driven, and it seems reasonable to demand that this is the case in decisions about the implementation of interventionist public health measures. Our reliance on mechanistic reasoning is also important when it comes to questions about how we identify policy goals. If we are very certain about a specific causal relation, such as the relation between smoking and increased risk of cancer, then a concrete outcome like reducing smoking will become a reasonable policy goal. Such concrete goals are much easier to track than more general states of improved health, which will tend to be multidimensional and involve many different causes. If we look at the case of obesity, we arguably have a relatively good picture of the risk factors and mechanisms involved, so at least when it comes to this point, policy interventions targeting obesity, and which have lowered rates of obesity as a concrete and trackable policy goal, would seem legitimate. (2) Evidence-oriented retrospective decision-making. While there is a tendency to question and test new policy proposals much more closely than long-running existing policies, evidence-based policy-making as an ideal points to the importance of continual policy review. There are two main sides to this. The first is about striving to support evidence-gathering. With respect to experimental policy-making this can, for example, be done by trying the policy out only in certain geographical areas at first, so that we obtain a relevant contrast class against which the results can be compared (although depending on the exact policy, this might not always be feasible and there can also be problems of generalizability). Additionally, exercising control over the more concrete implementations, which often involve multiple agents, will be important to reduce the number of possible confounders in future assessments. Seeing to it that relevant statistical data are collected is of course essential. The second point is about institutionalizing evidence-assessment. This can partly be an internal government process, but as already pointed out, central planners and bureaucrats are humans too and the all too human presence of confirmation bias will bring with it a tendency to interpret new data so that it fits with what one already thinks. This makes transparency and contestability into two key values in this kind of institutionalization (Hawkins and Parkhurst, 2016). Especially with respect to experimental policy-making, this orientation might be further facilitated by constructing policies so that they automatically expire at a certain date unless renewed. Now, it should be pointed out that the demand for evidence-oriented retrospective decision-making is a quite different type of dimension in that is not antecedent to the policy-making, but rather a form of legitimacy that is achieved through the way that policies are actually enacted and followed up. If we look at the case of obesity, there is obviously a very active research community with which policy-makers can collaborate in facilitating follow-up, so the basics for building this kind of legitimacy seems to be in place. Is the above too general and indeterminate? As already argued, policy-making processes do not lend themselves to strict requirements of evidence-basing, and while an exact line can hardly be drawn identifying the point at which enough has been done in terms of the above two dimensions of policy-making to indicate alignment with a reasonable ideal of evidence-based policy-making, these two dimensions can still be used to assess the degree to which we can trust that centralized decisions about policy are superior to the distributed decisions that will be made if matters are left to individual choice. These requirements are relatively weak, but that is also the point. Given that we are dealing with areas where there are strongly clustered failures, it will be enough that these relatively weak requirements are satisfied for there to be epistemic legitimacy for paternalist interventions. Acknowledgements This article has greatly benefitted from my participation in the meetings of the Bank of Sweden Tercentenary Foundation Program on Science and Proven Experience. The author is especially grateful to Charlotta Levay, Lena Wahlberg and Annika Wallin for reading and commenting on an earlier version of the text as well as the detailed and perceptive comments provided by the two referees. Funding The author completed this article while holding a position as Research Fellow granted by the Royal Swedish Academy of Letters, History and Antiquities Footnotes 1. In fact, the framework proposed here is really about identifying areas where (i) our decision-making tends to be ‘bad enough’ to raise the question of paternalist interventions, and (ii) where the possibilities of making such interventions in a knowledgeable way are ‘good enough’. 2. Note that the focus here is on tendencies where the prospects of beneficial incremental adaptations are fairly low, since they often will appear to work reasonably well within a short-term everyday framework, and the long-term feedback mechanisms are likely to be relatively inefficient. 3. Cf. Rawls on primary goods (Rawls, 1971: 62). Note that writers like Thaler and Sunstein (2008: 9) and Conly (2013: 123–5), even though they are prone to point to the many ways in which people are often irrational, still want to maintain our preferences as the standard for assessing our good. The idea here is instead that while actual preferences can often be taken to decide what lies in our interest, in some cases we need to focus on generic or primary goods instead in deciding this. References Benhabiba J., Bisin A., Schottera A. ( 2010). Present-bias, Quasi-hyperbolic Discounting, and Fixed Costs. Games and Economic Behavior , 69, 205– 223. Google Scholar CrossRef Search ADS   Berggren N. ( 2012). Time for Behavioral Political Economy? An Analysis of Articles in Behavioral Economics. Review of Austrian Economics , 25, 199– 221. Google Scholar CrossRef Search ADS   Cabinet Office. ( 2000). Adding It Up: Improving Analysis and Modelling . London: Cabinet Office. Cartwright N., Hardie J. ( 2012). Evidence-Based Policy . Oxford: Oxford University Press. Google Scholar CrossRef Search ADS   Cochrane A. ( 1972). Effectiveness and Efficiency . London: Nuffield Provincial Hospitals Trust. Colchero M. A. ( 2016). Beverage Purchases from Stores in Mexico under the Excise Tax on Sugar Sweetened Beverages: Observational Study. BMJ , 352, h6704. Google Scholar CrossRef Search ADS PubMed  Conly S. ( 2013). Against Autonomy . Cambridge: Cambridge University Press. Dehaene S. ( 2003). The Neural Basis of the Weber-Fechner Law: A Logarithmic Mental Number Line. Trends in Cognitive Sciences , 7, 145– 147. Google Scholar CrossRef Search ADS PubMed  Dworkin G. ( 1972). Paternalism. The Monist , 56, 64– 84. Google Scholar CrossRef Search ADS   Gostin L. O. ( 2015). Why Healthy Behavior is the Hard Choice. The Milbank Quarterly , 93, 242– 246. Google Scholar CrossRef Search ADS PubMed  Grether D. M., Plott C. R. ( 1979). Economic Theory of Choice and the Preference Reversal Phenomenon. American Economic Review , 69, 623– 638. Hawkins B., Parkhurst J. ( 2016). The “Good Governance” of Evidence in Health Policy. Evidence and Policy , 12, 575– 592. Google Scholar CrossRef Search ADS   Howick J. ( 2011). The Philosophy of Evidence-Based Medicine . Oxford: Wiley-Blackwell. Google Scholar CrossRef Search ADS   Longoa M. R., Lourencoa S. F. ( 2007). Spatial Attention and the Mental Number Line: Evidence for Characteristic Biases and Compression. Neuropsychologia , 45, 1400– 1407. Google Scholar CrossRef Search ADS PubMed  McClure S. M., Laibson D. I., Loewenstein G., Cohen J. D. ( 2004). Separate Neural Systems Value Immediate and Delayed Monetary Rewards. Science , 306, 503– 507. Google Scholar CrossRef Search ADS PubMed  Mill J. S. ( 1989). On Liberty and Other Writings . ed. S. Collini. Cambridge: Cambridge University Press. Montier J. ( 2007). Behaving Badly. In Behavioural Investing: A Practitioner’s Guide to Applying Behavioural Finance . Oxford: John Wiley & Sons. Google Scholar CrossRef Search ADS   Nieder A., Miller E. ( 2003). Coding of Cognitive Magnitude: Compressed Scaling of Numerical Information in the Primate Prefrontal Cortex. Neuron , 37, 149– 157. Google Scholar CrossRef Search ADS PubMed  Parsons W. ( 2002). From Muddling Through to Muddling Up—Evidence Based Policy Making and the Modernisation of British Government. Public Policy and Administration , 17, 43– 60. Google Scholar CrossRef Search ADS   Pierce J., White V., Emery S. ( 2012). What Public Health Strategies are Needed to Reduce Smoking Initiation? Tobacco Control , 21, 258– 264. Google Scholar CrossRef Search ADS PubMed  Puhl R., Heuer C. ( 2010). Obesity Stigma: Important Considerations for Public Health. American Journal of Public Health , 100, 1019– 1028. Google Scholar CrossRef Search ADS PubMed  Rawls J. ( 1971). A Theory of Justice . Cambridge, MA.: Harvard University Press. Resnik D. B. ( 2014). Paternalistic Food and Beverage Policies: A Response to Conly. Public Health Ethics , 7, 170– 177. Google Scholar CrossRef Search ADS   Hruby A., Manson J. E., Qi L., Malik V. S., Rimm E. B., Sun Q., Willett W. C., Hu F. B. ( 2016). Determinants and Consequences of Obesity. American Journal of Public Health , 106, 1656– 1662. Google Scholar CrossRef Search ADS PubMed  Shane F., Loewenstein G., O'Donoghue T. ( 2002). Time Discounting and Time Preference: A Critical Review. Journal of Economic Literature , 40, 351– 401. Google Scholar CrossRef Search ADS   Simon H. A. ( 1956). Rational Choice and the Structure of the Environment. Psychological Review , 63, 129– 138. Google Scholar CrossRef Search ADS PubMed  Slovic P. ( 1995). The Construction of Preference. American Psychologist , 50, 364– 371. Google Scholar CrossRef Search ADS   Solomon M. ( 2015). Making Medical Knowledge , Oxford: Oxford University Press. Google Scholar CrossRef Search ADS   Strotz R. ( 1955-56). Myopia and Inconsistency in Dynamic Utility Maximization. The Review of Economic Studies , 23, 165– 180. Google Scholar CrossRef Search ADS   Svenson O. ( 1981). Are We Less Risky and More Skillful than Our Fellow Drivers? Acta Psychologica , 47, 143– 151. Google Scholar CrossRef Search ADS   Thaler R. ( 1981). Some Empirical Evidence on Dynamic Inconsistency. Economics Letters , 8, 201– 207. Google Scholar CrossRef Search ADS   Thaler R., Sunstein C. ( 2003). Libertarian Paternalism is not an Oxymoron. The University of Chicago Law Review , 70, 1159– 1202. Google Scholar CrossRef Search ADS   Thaler R., Sunstein C. ( 2008). Nudge . Yale: Yale University Press. Timmermans S., Berg M. ( 2003). The Gold Standard . Philadelphia: Temple University Press. Tversky A., Simonson I. ( 1993). Context-Dependent Preferences. Management Science , 39, 1179– 1189. Google Scholar CrossRef Search ADS   Weinstein N. D. ( 1980). Unrealistic Optimism about Future Life Events. Journal of Personality and Social Psychology , 39, 806– 820. Google Scholar CrossRef Search ADS   Wilson J. ( 2011). Why It's Time to Stop Worrying About Paternalism in Health Policy Public Health Ethics , 4, 269– 279. Google Scholar CrossRef Search ADS   © The Author 2017. Published by Oxford University Press. Available online at www.phe.oxfordjournals.org http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Public Health Ethics Oxford University Press

On the Epistemic Legitimacy of Government Paternalism

Loading next page...
 
/lp/ou_press/on-the-epistemic-legitimacy-of-government-paternalism-1foYwNtmHP
Publisher
Oxford University Press
Copyright
© The Author 2017. Published by Oxford University Press. Available online at www.phe.oxfordjournals.org
ISSN
1754-9973
eISSN
1754-9981
D.O.I.
10.1093/phe/phx010
Publisher site
See Article on Publisher Site

Abstract

Abstract Some contemporary paternalists argue in favor of government interventions based on how experimental psychologists and behavioral economists have found that our behavior often diverges from what would be predicted by rational-choice models. In this article it is argued that these findings can, more specifically, be used to identify decisional trouble spots where paternalist interventions may be legitimate. It is further argued that since the epistemic legitimacy of government paternalism ultimately rests on centralized decision-making having a comparative advantage, it also depends on the possibility of such interventions being governed by an ideal of evidence-based policy-making. The article asks how stringently this requirement should be understood, and to what extent government can legitimately engage in what might be called experimental policy-making of a paternalistic character. The past 10 or 15 years have seen the emergence of paternalistic approaches to public policy that rely heavily on contemporary empirical psychology in understanding our everyday choices on matters of health, wealth and safety. These paternalists might differ on which means that should be used, e.g. Richard Thaler and Cass Sunstein (Thaler and Sunstein, 2003, 2008) defend what they call libertarian paternalism and advocate nudges (setting up the choice architecture we face so that we tend to choose in a way that is beneficial over the long term), while Sarah Conly (2013) defends what she calls coercive paternalism and advocates more clearly restrictive measures. But they are united in pointing to how dual-process models in psychology highlight how many of our everyday choices are based on heuristics and biases, thus often leading to suboptimal results, an observation which can be taken as grounding the epistemic legitimacy of government paternalism: that policy-makers might sometimes be in a position of knowing better, being able to make rational decisions that we as citizens fail to do. In general, this new wave of paternalism has however tended to focus strongly on the decision-making of us as individuals, which is only one side of the kind of comparison that you need to make if you are going to argue that government paternalism has what might be called epistemic legitimacy, i.e. that centralized policy-making has an edge in terms of knowledge and rationality over the distributed decision-making of us as individuals. For instance, one literature review of articles in behavioral economics found that of 67 articles that made interventionist policy recommendations, only 2 seriously considered the possible irrationality of policy-makers (Berggren, 2012: 216). Of course, in making policy recommendations, one is forward-looking and concerned with how the policy-making process should function, so while it is important to consider how policy-makers generally function, the key issue concerns the extent to which would-be regulators in the relevant area can realistically be expected to meaningfully be governed an ideal of evidence-based policy-making. Such an analysis needs to be done before endorsing concrete policy recommendations because it can decide which recommendations that make the most sense. This article will put forward a basic framework for making comparative analyses with respect to possible public health interventions. Comparative advantage and the possibility of clustered failures Before proceeding it should be noted that paternalistic policies are rarely paternalistic in the same clear-cut way as paternalistic actions in, say, the physician–patient relationship (cf. Wilson, 2011). To begin with, since public health interventions tend to be designed to lower the rates of adverse health consequences that come with certain types of behavior, there is a strong link to what our healthcare system needs to do and the costs of what it does. Healthcare systems in most countries tend to be publicly part-financed, and this means that if a public health intervention is successful, then qua taxpayers we all benefit, and public health measures that involve taxing certain forms of consumption do not just steer behavior, they also raise revenue and thus enable us to spend more on public goods: again, we all benefit. On many standard formulations of what it means for something to be paternalistic, there is a requirement that it is done simply for the benefit of the person intervened against (Dworkin, 1972: 65). Indeed, Mill’s harm principle (1989: 13) does not say more than that a person’s ‘own good, either physical or moral, is not sufficient warrant’ for interfering with someone. Pure cases of paternalism of this sort are exceedingly rare at the level of public policy, even if we just focus on the intentions behind policies rather than on their effects. Still: many public health policies often have paternalistic elements to them and, to the extent that they do, they should concern choices where there is a comparative advantage to effectively make those decisions in a centralized rather than distributed way. When doing this comparison, then, if we look at us as individuals, there is by now a vast literature in empirical psychology and behavioral economics about how we as agents are far from the rationalist constructs assumed by many traditional economists as a basis for their mathematical models of behavior and equilibria. But even if we accept this kind of picture of human decision-making, we need not think that it points to a need for paternalist interventions. To begin with, on the level of individuals, it is quite possible that when we move from experimental setups to normal circumstances, our heuristics and biases are, at least on the whole, just efficient ways of making reasonably good decisions. We might think that satisficing is a reasonable strategy in everyday life (Simon, 1956). But this does not preclude that there are still certain types of choices we make on an everyday basis that are clearly counterproductive to our interests and that should be regarded as irrational not just in a narrow sense of the notion.1 It might also be objected that even if there are choices like these, they still need not be suitable objects for interventions. At least over time, our choices might either even out or at least not add up to much; we can in principle have a long series of irrational choices without the net effect of them being particularly serious. It lies in the nature of decision-making shortcuts that they can often be efficient ways of making decisions, but still sometimes lead us astray. We accordingly need to distinguish between two ways in which we can systematically fall short of ideals of knowledge and rationality. First, individuals might have frequent but dispersed failures, i.e. they might regularly fail to live up to the ideals but in a very wide range of ways and for lots of different reasons. In this situation there is little reason to think that centralized decision-making would improve on individual choice. But we can also have strongly clustered failures of rationality in identifiable problem areas. These would have to do with types of choice situations where several potentially problematic tendencies will all tend to pull us in the same direction, toward choices that are not just not the best, but also not even good enough. Four problematic tendencies The relevant literature in psychology and behavioral economics identifies a number of heuristics and biases. Some of them, such as confirmation bias, might however lead to relatively dispersed failures, i.e. failures which take us in many different directions and might possibly even out on a population level. A strong case for public health interventions would instead involve identifying several problematic tendencies that cluster in an area of decision-making and increase the risk of making a certain type of bad choice. In a public health context one common type of focus is on long-term health impacts of lengthy sequences of small everyday choices, where the risk increases from a single action undertaken here and now might be very slight, but where long-terms effects are significant. In these types of cases there are at least the following four tendencies that might cluster and lead to rationality failures in our decision-making:2 (1) We tend to suffer from temporal myopia. Potential future gains and losses are often more uncertain than those that are immediate or imminent, so if the two sorts are equal in magnitude, it seems prudent to value them differently, and to discount the former. Already in the 1950s, Robert Strotz (Strotz, 1955-1956) speculated that we have an in-born discount function, and later experimental studies have indicated that this is indeed the case (Shane et al., 2002; Benhabiba et al., 2010). Note that the bare operation of a discount rate is not irrational—far from it. Problems arise when the discount is the product of a relatively blind heuristic rather than a measured response. For instance, maybe the difference a day makes means that some people would rather choose a certain reward today than a larger one tomorrow, but then what are we to say about the same choice when it is made between outcomes that are a year away or a year and a day away from now? If it was just a matter of the difference a day’s waiting time makes, the day should make the same difference here as well, but this turns out to not be the case (Thaler, 1981). More recently, neuropsychological results have indicated that different parts of the brain are active, depending on whether we are thinking about present or future rewards (McClure et al., 2004). (2) We tend to have problems with differences in orders of magnitude. While the kinds of decisions that are relevant in a public health context often require us to consider temporal distances, they also often involve very different orders of magnitude. Already in the nineteenth century, Ernst Heinrich Weber had discovered that human responses to physical stimuli were more discriminating when the magnitudes of those stimuli were small; the bigger the magnitudes, the bigger the differences between two objects must be in order for us to notice the difference between them. In recent years, neuropsychologists have observed the same phenomenon in connection with abstract objects like numbers (Nieder and Miller, 2003; Longoa and Lourencoa, 2007). The mental number line gradually becomes more compressed, so that accuracy is indirectly proportional to the size of the numbers involved (Dehaene, 2003). Of course, in everyday life we mostly deal with small numbers, so there is no mystery about why we have evolved to be more accurate there, but this also means we there have inherent difficulties in making everyday decisions that relate both to the relatively small payoffs that we get in the present and potentially much larger effects that lie in the future. (3) We tend to be overconfident and excessively optimistic. When it comes to our possibilities to steer clear of future problems, we are in part dependent on our abilities and in part on the likelihood of different external events. In both of these areas people tend to have overly rosy pictures. For instance, with respect to self-assessments of one’s general ability levels, in one study of American drivers, 93% thought that they were better than the median driver (Svenson, 1981) and in a study of professional fund managers, 74% believed that they were above average in their job performance and 26% that they were average (Montier, 2007). When it comes to the likelihood of external events happening to us rather than others, we exhibit similar patterns. We tend to overestimate the relative likelihood that good things will happen to us and underestimate the relative likelihood that bad things will happen to us (Weinstein, 1980). As long as being only moderately delusional, these tendencies might perhaps often be a helpful coping mechanism (thinking that one is below average at one’s job is surely a depressing thought), but there is certainly a possibility that such tendencies will at times lead us to not take certain risks seriously enough. (4) Our preferences are often unstable and indeterminate. The standard model for decision-making centers on a notion of expected utility in which our preferences determine the basic utility values involved and the probabilities of achieving the objects of our preferences give us the expected utilities of all the options available. The problem is that our preferences do not appear to be determinate enough for this notion to be applied; instead we are highly sensitive to circumstantial factors, and several studies have shown that, depending on how one attempts to elicit the preferences, we make different choices, even when the methods of elicitation should be normatively equivalent (Grether and Plott, 1979; Tversky and Simonson, 1993; Slovic, 1995). We display preference reversals in which given one method of elicitation a subject will choose A over B, and given another method she will choose B over A. At least for many preferences we seem to partly make things up as we go along rather than drawing on a pre-existing preference order. This might actually be an economic way of handling many decisions, but it also means our preferences in the moment cannot always reliably be used to determine what we really want. When looking at where the comparative advantage lies it is an assessment that is impossible to do with complete precision, but having identified these tendencies, we can look at relevant choices or areas of decision-making and judge the extent to which there is a likelihood of clustered failures. There is no room here for deciding whether this is a definitive list of what we should be looking at, but it is a basic model at least. With respect to people’s ability to rationally make certain decisions, we would then ask these four questions: Is there a significant temporal distance between the relevant action(s) and possible long-term effects? Is there a significant difference in orders of magnitude between the short-term effects of the relevant action(s) and their long-term effects? Is there significant room for wishful thinking in relation to long-term effects? Is there significant vacillation or reversals in the relevant preferences or judgments? These dimensions are certainly not entirely independent of each other, but neither do they completely collapse into each other. If we strongly answer all of these questions in the affirmative, there is reason to distrust that people will know what they are doing in relation to the area of decision-making in question. This would then point to a need to take the possibility of making paternalist interventions seriously—although before we conclude that such interventions really are epistemically legitimate, we must also look at the side of policy-makers as well. The first three dimensions are of the same kind in that they involve tendencies which, taken singly and with respect to many everyday decisions, are probably evolutionary adaptive ways of thinking. They can however become problematic when they cluster, especially in a modern context where day-to-day survival is not really a central concern and where long-term health effects become more salient. The fourth dimension is different and plays a dual role. To begin with, it could in principle be the case that the first three questions are answered in the affirmative, but people in the end still tend to firmly and stably endorse the choices they made, even retrospectively. If this is the case, it becomes questionable to understand these choices as irrational or deeply problematic. But then it is also the case that to the extent that there is vacillation or reversals in our preferences, these preferences become less viable as a standard of what counts as a good outcome, and it becomes more reasonable to think instead in terms of generic goods, such as expected gains in life years, when evaluating outcomes, i.e. goods that more or less everyone can non-controversially be assumed to be better off in having more rather than less off.3 This is the kind of goods that public health interventions ultimately tend to aim at, so to the extent that outcomes for individuals are difficult to assess in the light of their precise preferences, this further opens up for centralized decision-making. Since this article is mainly about working toward a model for making certain comparisons relevant to possible public health interventions, there will not be that much room to look at concrete examples, but let us briefly look at two, albeit mainly as illustrations rather than full-scale applications. First, a relatively straightforward case: seat belts. Seat belt mandates tend to be broadly accepted, even though the choice to forgo seat belt would seem to be relatively mundane. The above model does however identify it as a problem area. Both (2) and (3) definitely seem to obtain. When it comes to (1), the temporal distance to possible ill effects is highly variable, but it is typically well past the horizon of the particular decision-making situation. It also seems like an area where it would be reasonable to expect regret in the face of those ill effects (if those affected are even in a state to have a preference reversal). In sum, when looking at choices to forgo using a seat belt, it seems reasonable to regard these as highly likely to be the product of clustered failures of rationality, and hence conclude that this is an area of decision-making where it might be quite possible for policy-makers to do better. A second possible application is about consumption choices, such as consumption of junk food, that can lead to obesity, which in turn carries with it an increased risk of type 2 diabetes, hypertension and cardiovascular disease, certain forms of cancer and an adverse impact on quality of life (Hruby et al., 2016). With respect to questions (1) and (2), choices to consume junk food clearly seem to be ones where the answer is ‘yes’. Especially with respect to our early habit-forming choices, there is a significant temporal distance to the long-term effects; the short-term pleasures and the long-term adverse health effects do also seem to lie on different orders of magnitude in terms of the values involved. Question (3) is not as straightforward. There does not seem to be much room for wishful thinking when it comes to the relation between certain habits and obesity, perhaps apart from the fact that obesity is the product of many years of certain behaviors and there is always room to think that one’s current behavior is really just temporary. It is however not really obesity that is the problem, but the health consequences that come with it, and since the connections here might primarily be visible on an epidemiological level, there seems to be ample room for individuals to think that they will be among the lucky ones. Question (4) is a more complicated one in this particular case. There are clearly negative emotions often associated with obesity, but on the other hand, those might also be linked to a stigmatization which is not only morally problematic but might even result in more of the behavior that causes obesity to begin with it (Puhl and Heuer, 2010). Arguably this is however still an area of behavior where there is often significant vacillation in people’s preferences, e.g. between consuming more junk food in the moment and feeling that one needs to cut back on it at other times. Such vacillations point to it not being an area of rational decision-making and add to the picture that when it comes to high consumption of junk food, it can be seen a type of behavior that grows out of clustered failures of rationality. Policy-making and the role of evidence The ideal of evidence-based policy-making first came to prominence in the late 1990s (Parsons, 2002). The British New Labour government was an early proponent, with Tony Blair affirming a ‘commitment to policy-making based on hard evidence’ (Cabinet Office, 2000: 3). Up to a point, more or less everyone should be able to agree that politics cannot merely be about ideology and that one must also consider the expected effectiveness of proposed measures. The interesting question, however, is just what we should demand in the form of evidence. Where public health interventions are concerned, David Resnik (2014: 174) has suggested that there should be ‘substantial evidence that the restrictions are likely to be effective at addressing the health problem’. While one might discuss just when evidence is substantial and what likely means, this requirement is clearly aimed at setting the bar relatively high. Its adoption would make experimental policy-making highly problematic. This might very well be the point, but at the same time we need to remember that decisions impacting on public health are always already being made by us as individuals, and the basic question here is not about the absolute level of competence at which it is permissible, or advisable, for governments to make these decisions. It is about the comparative competence of the state and individuals. It is also important not to be misled here by an analogy between evidence-based policy-making and evidence-based medicine. The latter preceded the former and has had an enormous influence on how we now think about best practice in medicine (Howick, 2011; Solomon, 2015). Although it might be difficult to prove conclusively, it seems reasonable to think that its success has been a strong contributing factor in the ascent of the ideal of evidence-based policy-making. Already from the outset, however, there are difficulties in moving ideals from medicine to policy-making. For instance, while randomized controlled trials are usually seen as the gold standard in evidence-based medicine (Timmermans and Berg, 2003) and are comparatively manageable to perform in medical research, government policies are another matter. Policies are not enacted in controlled settings, but rather in the wild, so to speak. The number of potential confounders in that arena is much greater and the intricacies of policy implementation may mean that what on the surface looks like the same policy being implemented in another place will be a policy whose implementation is unique, or at any rate different, in important respects. While the results of rigorous medical research done in one country can often be transferred to other countries, transferring results from a policy evaluation will be more difficult simply because there are so many background factors that might play a role in making a policy work or fail. These factors often vary greatly from country to country. Something might work there, but not here (Cartwright and Hardie, 2012). Conversely, something might not work there, but still work here. Another problem with the analogy has to do with the underlying rationale for setting demands for evidence at a certain level and the extent to which demands for evidence tend to privilege inaction over action. The human body is a self-repairing system which, even when it is left alone, tends to return from sickness to a state of health (Cochrane, 1972: 5). Society is not quite like that. To the extent that there are societal states of equilibrium these are products of contingent complexes of institutions that shape the interactions between members of society. This means that while in medicine there is often a real choice between treatment and non-treatment, in politics we are often left with a choice between affecting people in certain ways or affecting them in certain other ways. Doing nothing usually still means doing something, and sometimes neither approach is backed by hard evidence. We should be careful not to fall prey to the illusion that simply acquiescing in the status quo is equivalent to not affecting people’s lives. For instance, the way the contemporary food industry operates is not just a natural fact, but is already the product of a contingent set of institutions and regulations. Finally, it needs to be remembered that at least with respect to public health policies, measures often target behaviors with strong inertia. This means that such policies usually face an uphill battle and that often it will require a package of interventions to really make healthy behavior the easy choice (cf. Gostin, 2015). For instance, many industrialized countries have now for decades pursued comprehensive tobacco control programs, and there appears to be strong evidence that these programs ‘have a dramatic impact in reducing the proportion of adolescents who start to smoke’ (Pierce et al., 2012: 262), but the comprehensive programs are packages of policies, such as increasing the price of cigarettes, requiring warning labels on packs, restricting advertising, introducing smoke-free areas, restricting accessibility by age and various kinds of school interventions. While it is certainly possibly to evaluate such policies one by one, it is also important to consider how they might function as parts of a package, where the joint effect of making it difficult enough to start smoking is what might in the long run produce the sought-after shift in behavioral patterns. Now, the argument here is that what is needed for epistemic legitimacy is both that there are clustered failures of rationality with respect to a specific area of decision-making and that policy-making with respect to that area rises to the level of being about evidence-based policy-making rather than just a matter of guesses and hunches. The idea is not that it must be possible to conclusively determine exactly which policy or mix of policies that would be the best, but simply that there is enough of evidence-basing to be able to make rational policy choices in an area of decision-making where clustered failures of rationality are likely on the level of individuals. With respect to tobacco control programs, the picture that emerges seems to be that a reasonable level of evidence-based policy-making is possible. There might still be room for discussion about exactly which policies or mix of policies are the best ones, but in the comparative analysis what is important is that a certain threshold of rationality in policy-making is clearly passed. If we look at the current state of policy-making in relation to obesity, it is similar to tobacco control in that there does not seem to be any one policy that will single-handedly provide the magic bullet that halts the rising levels of obesity and associated health problems. Many policies or suggested policies (a good example being former Mayor Bloomberg’s proposed ban on large sodas) are mainly about tweaking behavior, and while there is some evidence that some such tweaks, such as taxing sugar-sweetened beverages (Colchero, 2016) have some effect on behavior, substantial shifts in broad patterns of behavior are only likely to be accomplished by comprehensive packages of policies. So a lot of work remains to be done, and here a potential problem presents itself. To get good evidence on which policies that work, we need to enact these policies somewhere. With respect to obesity and its associated health effects, there is accordingly a significant need for experimental policy-making now for policy-making to become strongly evidence-based in the future, so the question becomes one of what we should demand for something to be a legitimate area of (at least partly) experimental policy-making. Even if there are clustered failures in that area, it still seems reasonable to require some positive qualities from such policy-making. The suggestion here is that we can frame this in terms of a demand that it should be done in a way that allows it to be meaningfully governed by an ideal of evidence-based policy-making. Here are two key dimensions to focus on in determining the extent to which this is the case. (1) Evidence-driven prospective decision-making. Given that the existence of adverse public health consequences as the likely product of clustered failures have been established, the most significant aspect of our decision-making process will be in making predictions about the likely effects of different policy measures. There might be further moral and political reasons for preferring certain such measures, presumably less invasive ones, but effectiveness and efficiency are the main concerns when setting those other reasons aside. It is important to then keep in mind that evidence-based and research-based do not mean the same thing. Much research is oriented toward understanding the mechanisms underlying what happens, and armed with an understanding of such mechanisms we might make predictions about what ought to happen in a system given a certain stimulus. The models on which such predictions are made should, of course, be based on evidence supporting our belief in the relevance of certain mechanisms, but in the end the relation between this evidence and the predictions we make will be indirect. While these predictions may not be evidence-based in a strong sense, they can still be research-based and evidence-driven, and it seems reasonable to demand that this is the case in decisions about the implementation of interventionist public health measures. Our reliance on mechanistic reasoning is also important when it comes to questions about how we identify policy goals. If we are very certain about a specific causal relation, such as the relation between smoking and increased risk of cancer, then a concrete outcome like reducing smoking will become a reasonable policy goal. Such concrete goals are much easier to track than more general states of improved health, which will tend to be multidimensional and involve many different causes. If we look at the case of obesity, we arguably have a relatively good picture of the risk factors and mechanisms involved, so at least when it comes to this point, policy interventions targeting obesity, and which have lowered rates of obesity as a concrete and trackable policy goal, would seem legitimate. (2) Evidence-oriented retrospective decision-making. While there is a tendency to question and test new policy proposals much more closely than long-running existing policies, evidence-based policy-making as an ideal points to the importance of continual policy review. There are two main sides to this. The first is about striving to support evidence-gathering. With respect to experimental policy-making this can, for example, be done by trying the policy out only in certain geographical areas at first, so that we obtain a relevant contrast class against which the results can be compared (although depending on the exact policy, this might not always be feasible and there can also be problems of generalizability). Additionally, exercising control over the more concrete implementations, which often involve multiple agents, will be important to reduce the number of possible confounders in future assessments. Seeing to it that relevant statistical data are collected is of course essential. The second point is about institutionalizing evidence-assessment. This can partly be an internal government process, but as already pointed out, central planners and bureaucrats are humans too and the all too human presence of confirmation bias will bring with it a tendency to interpret new data so that it fits with what one already thinks. This makes transparency and contestability into two key values in this kind of institutionalization (Hawkins and Parkhurst, 2016). Especially with respect to experimental policy-making, this orientation might be further facilitated by constructing policies so that they automatically expire at a certain date unless renewed. Now, it should be pointed out that the demand for evidence-oriented retrospective decision-making is a quite different type of dimension in that is not antecedent to the policy-making, but rather a form of legitimacy that is achieved through the way that policies are actually enacted and followed up. If we look at the case of obesity, there is obviously a very active research community with which policy-makers can collaborate in facilitating follow-up, so the basics for building this kind of legitimacy seems to be in place. Is the above too general and indeterminate? As already argued, policy-making processes do not lend themselves to strict requirements of evidence-basing, and while an exact line can hardly be drawn identifying the point at which enough has been done in terms of the above two dimensions of policy-making to indicate alignment with a reasonable ideal of evidence-based policy-making, these two dimensions can still be used to assess the degree to which we can trust that centralized decisions about policy are superior to the distributed decisions that will be made if matters are left to individual choice. These requirements are relatively weak, but that is also the point. Given that we are dealing with areas where there are strongly clustered failures, it will be enough that these relatively weak requirements are satisfied for there to be epistemic legitimacy for paternalist interventions. Acknowledgements This article has greatly benefitted from my participation in the meetings of the Bank of Sweden Tercentenary Foundation Program on Science and Proven Experience. The author is especially grateful to Charlotta Levay, Lena Wahlberg and Annika Wallin for reading and commenting on an earlier version of the text as well as the detailed and perceptive comments provided by the two referees. Funding The author completed this article while holding a position as Research Fellow granted by the Royal Swedish Academy of Letters, History and Antiquities Footnotes 1. In fact, the framework proposed here is really about identifying areas where (i) our decision-making tends to be ‘bad enough’ to raise the question of paternalist interventions, and (ii) where the possibilities of making such interventions in a knowledgeable way are ‘good enough’. 2. Note that the focus here is on tendencies where the prospects of beneficial incremental adaptations are fairly low, since they often will appear to work reasonably well within a short-term everyday framework, and the long-term feedback mechanisms are likely to be relatively inefficient. 3. Cf. Rawls on primary goods (Rawls, 1971: 62). Note that writers like Thaler and Sunstein (2008: 9) and Conly (2013: 123–5), even though they are prone to point to the many ways in which people are often irrational, still want to maintain our preferences as the standard for assessing our good. The idea here is instead that while actual preferences can often be taken to decide what lies in our interest, in some cases we need to focus on generic or primary goods instead in deciding this. References Benhabiba J., Bisin A., Schottera A. ( 2010). Present-bias, Quasi-hyperbolic Discounting, and Fixed Costs. Games and Economic Behavior , 69, 205– 223. Google Scholar CrossRef Search ADS   Berggren N. ( 2012). Time for Behavioral Political Economy? An Analysis of Articles in Behavioral Economics. Review of Austrian Economics , 25, 199– 221. Google Scholar CrossRef Search ADS   Cabinet Office. ( 2000). Adding It Up: Improving Analysis and Modelling . London: Cabinet Office. Cartwright N., Hardie J. ( 2012). Evidence-Based Policy . Oxford: Oxford University Press. Google Scholar CrossRef Search ADS   Cochrane A. ( 1972). Effectiveness and Efficiency . London: Nuffield Provincial Hospitals Trust. Colchero M. A. ( 2016). Beverage Purchases from Stores in Mexico under the Excise Tax on Sugar Sweetened Beverages: Observational Study. BMJ , 352, h6704. Google Scholar CrossRef Search ADS PubMed  Conly S. ( 2013). Against Autonomy . Cambridge: Cambridge University Press. Dehaene S. ( 2003). The Neural Basis of the Weber-Fechner Law: A Logarithmic Mental Number Line. Trends in Cognitive Sciences , 7, 145– 147. Google Scholar CrossRef Search ADS PubMed  Dworkin G. ( 1972). Paternalism. The Monist , 56, 64– 84. Google Scholar CrossRef Search ADS   Gostin L. O. ( 2015). Why Healthy Behavior is the Hard Choice. The Milbank Quarterly , 93, 242– 246. Google Scholar CrossRef Search ADS PubMed  Grether D. M., Plott C. R. ( 1979). Economic Theory of Choice and the Preference Reversal Phenomenon. American Economic Review , 69, 623– 638. Hawkins B., Parkhurst J. ( 2016). The “Good Governance” of Evidence in Health Policy. Evidence and Policy , 12, 575– 592. Google Scholar CrossRef Search ADS   Howick J. ( 2011). The Philosophy of Evidence-Based Medicine . Oxford: Wiley-Blackwell. Google Scholar CrossRef Search ADS   Longoa M. R., Lourencoa S. F. ( 2007). Spatial Attention and the Mental Number Line: Evidence for Characteristic Biases and Compression. Neuropsychologia , 45, 1400– 1407. Google Scholar CrossRef Search ADS PubMed  McClure S. M., Laibson D. I., Loewenstein G., Cohen J. D. ( 2004). Separate Neural Systems Value Immediate and Delayed Monetary Rewards. Science , 306, 503– 507. Google Scholar CrossRef Search ADS PubMed  Mill J. S. ( 1989). On Liberty and Other Writings . ed. S. Collini. Cambridge: Cambridge University Press. Montier J. ( 2007). Behaving Badly. In Behavioural Investing: A Practitioner’s Guide to Applying Behavioural Finance . Oxford: John Wiley & Sons. Google Scholar CrossRef Search ADS   Nieder A., Miller E. ( 2003). Coding of Cognitive Magnitude: Compressed Scaling of Numerical Information in the Primate Prefrontal Cortex. Neuron , 37, 149– 157. Google Scholar CrossRef Search ADS PubMed  Parsons W. ( 2002). From Muddling Through to Muddling Up—Evidence Based Policy Making and the Modernisation of British Government. Public Policy and Administration , 17, 43– 60. Google Scholar CrossRef Search ADS   Pierce J., White V., Emery S. ( 2012). What Public Health Strategies are Needed to Reduce Smoking Initiation? Tobacco Control , 21, 258– 264. Google Scholar CrossRef Search ADS PubMed  Puhl R., Heuer C. ( 2010). Obesity Stigma: Important Considerations for Public Health. American Journal of Public Health , 100, 1019– 1028. Google Scholar CrossRef Search ADS PubMed  Rawls J. ( 1971). A Theory of Justice . Cambridge, MA.: Harvard University Press. Resnik D. B. ( 2014). Paternalistic Food and Beverage Policies: A Response to Conly. Public Health Ethics , 7, 170– 177. Google Scholar CrossRef Search ADS   Hruby A., Manson J. E., Qi L., Malik V. S., Rimm E. B., Sun Q., Willett W. C., Hu F. B. ( 2016). Determinants and Consequences of Obesity. American Journal of Public Health , 106, 1656– 1662. Google Scholar CrossRef Search ADS PubMed  Shane F., Loewenstein G., O'Donoghue T. ( 2002). Time Discounting and Time Preference: A Critical Review. Journal of Economic Literature , 40, 351– 401. Google Scholar CrossRef Search ADS   Simon H. A. ( 1956). Rational Choice and the Structure of the Environment. Psychological Review , 63, 129– 138. Google Scholar CrossRef Search ADS PubMed  Slovic P. ( 1995). The Construction of Preference. American Psychologist , 50, 364– 371. Google Scholar CrossRef Search ADS   Solomon M. ( 2015). Making Medical Knowledge , Oxford: Oxford University Press. Google Scholar CrossRef Search ADS   Strotz R. ( 1955-56). Myopia and Inconsistency in Dynamic Utility Maximization. The Review of Economic Studies , 23, 165– 180. Google Scholar CrossRef Search ADS   Svenson O. ( 1981). Are We Less Risky and More Skillful than Our Fellow Drivers? Acta Psychologica , 47, 143– 151. Google Scholar CrossRef Search ADS   Thaler R. ( 1981). Some Empirical Evidence on Dynamic Inconsistency. Economics Letters , 8, 201– 207. Google Scholar CrossRef Search ADS   Thaler R., Sunstein C. ( 2003). Libertarian Paternalism is not an Oxymoron. The University of Chicago Law Review , 70, 1159– 1202. Google Scholar CrossRef Search ADS   Thaler R., Sunstein C. ( 2008). Nudge . Yale: Yale University Press. Timmermans S., Berg M. ( 2003). The Gold Standard . Philadelphia: Temple University Press. Tversky A., Simonson I. ( 1993). Context-Dependent Preferences. Management Science , 39, 1179– 1189. Google Scholar CrossRef Search ADS   Weinstein N. D. ( 1980). Unrealistic Optimism about Future Life Events. Journal of Personality and Social Psychology , 39, 806– 820. Google Scholar CrossRef Search ADS   Wilson J. ( 2011). Why It's Time to Stop Worrying About Paternalism in Health Policy Public Health Ethics , 4, 269– 279. Google Scholar CrossRef Search ADS   © The Author 2017. Published by Oxford University Press. Available online at www.phe.oxfordjournals.org

Journal

Public Health EthicsOxford University Press

Published: Apr 1, 2018

There are no references for this article.

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off