14 Points

Traditional paternalism seeks to change individuals’ behavior through taxation, regulation and coercion. But by offending their commitment to self-determination, it generates resistance and resentment which ultimately render it ineffective. The voluntarist approach, exemplified by rational choice theory, presumes that individuals are best placed to make choices that most closely match their preferences. But as a burgeoning literature in cognitive psychology and behavioral economics is now making clear, humans are only boundedly rational, and their behavior is subject to a range of biases including an inclination to favor the status quo, a susceptibility to framing effects and context, loss aversion, and the use of heuristics which can lead to non-rational decisions in situations of uncertainty.

Soft paternalism seeks to retain a high degree of voluntarism for individuals to decide what to do, but, having decided, lends them some of the coercive powers of the state to help them overcome some of their internal barriers to change. The classical allusion is sometimes made to Ulysses strapping himself to his mast so he can withstand the temptations of the Sirens. Modern manifestations of the same principle include compulsive gamblers who can voluntarily apply to have themselves blacklisted from casinos,[3] or the pregnant women in Schelling’s account who commit in advance to eschew anesthesia during their labor.

To its critics, soft paternalism of the sort advocated by Schelling, and to some extent by Bertand, Mullainathan and Shafir, is conceptually as well as normatively problematic.[4] The conceptual problem is that soft paternalism presumes the existence of one or more “selves”. As Schelling argues, “They do not exist simultaneously”. Rather they take control of cognitive processing at different times under different conditions, and substantive outcomes depend critically on which self is in the driving seat at key moments. In the context of personal finance, for example, Richard Thaler argues that we do not treat all our sources of income and wealth, or all the activities on which we could spend it, in the uniform fashion predicted by economists: we are much happier spending from what we think of as our “income” account than we are our “asset” account. But critics charge that soft paternalists have no way of determining which of the various selves is the one that “should” be listened to beyond their own preconceptions (such as the idea that it is better to think long-term than short-term, or to eat healthily than unhealthily). Ultimately, say opponents, it should be for individuals to make these judgments.

One issue brings this debate into sharp focus: suicide. Suicide challenges the temporal feature of the “selves debate”: if opponents of soft paternalism are right, and there are not two selves, then individuals’ revealed preferences to take their own life are a true statement of what they felt to be in their best interests, and suicide is tragic but ultimately not something that deserves particular attention in terms of policy design. But if supporters of soft paternalism are right, and there are two selves, then suicide may simply reflect the fact that the self that no longer wanted to go on living happened to be dominant at a critical moment, and the self that thought there was more to live for never got another chance to make its case.

In this case, policies to give that other self a second chance would matter a great deal. This finds practical expression, I would argue, in the debate about gun control in the United States.

According to a famous National Rifle Association (NRA) bumper-sticker, “Guns Don’t Kill People, People Kill People.” The argument runs that tougher gun laws are unjustified because the problem is not guns but the willingness of a few errant individuals to misuse them. But while the gun debate is fixated on homicide, the bigger problem is arguably in relation to suicide. More than 30,000 Americans take their lives each year, making suicide the 11^th highest cause of death. Homicides involving a firearm represent less than a third of this figure - around 8,000 a year. More than half (55%) of suicides involve a firearm.

If the NRA - and indeed all those who argue against a view of “multiple selves” - are right, then “people kill people” and gun policies should have no effect on suicide. On this view, suicide is a clear, deliberate, individual choice that is unrelated to whether people have access to a firearm: if they don’t own a gun they will just use something else. The null hypothesis would, therefore, be that the number of people that take their own lives in a given jurisdiction is uncorrelated with the strength of the gun laws within that jurisdiction.

In the rest of this paper I test this hypothesis using state-level data on gun laws from a study by the Open Society Institute and state-level data on suicides and suicide attempts from the Centers for Disease Control and Prevention and the Suicide Prevention Resource Center.

I make four main claims based on the empirical evidence:

·         The number of suicide attempts is uncorrelated with access to firearms: gun owners are not more suicidal than non-gun owners.

·         But having more guns in homes means that more suicide attempts will involve the use of firearms. Suicide attempts that involve firearms are more lethal than those that do not.

·         Weak gun laws lead to more guns in homes. Therefore weak gun laws lead to more suicides.

·         Gun laws that “enhance self-control” lead to fewer suicides.

1. The number of suicide attempts is uncorrelated with access to firearms: gun owners are not more suicidal than non-gun owners.

It is important to my subsequent claims that I establish that gun owners are not more predisposed to suicide than non-gun owners; that is, there is no reason to think that there are any social or psychological factors which mean that the type of people who are most likely to own guns are also the type of people who are most likely to be at risk of suicide.

Figure 0 (view by clicking here) illustrates this contention . It plots the number of attempted suicides against a measure for gun ownership. Somewhat confusingly this measure also involves suicide: because administrative and survey data on gun ownership are unavailable or unreliable across states, researchers have turned to proxy measures and have found that the most reliable is the percentage of suicides involving the use of a firearm.[5] Despite the confusion, it should be clear that these two measures are tapping entirely different things: the first deals with suicide attempts and is a level; the second deals with actual suicides and is a proportion. Figure 0 shows that the two are almost totally unrelated (the Pearson correlation coefficient is 0.062).

2. However, having more guns in homes means that more suicide attempts involve the use of firearms, and suicide attempts involving firearms are more lethal than those that do not.

As I have argued, the empirically most accurate predictor of gun prevalence is the percentage of suicides involving the use of a firearm. However, as Figure 2 shows, there is also a very strong positive correlation (0.933) between the total number of suicides in a state and the number of suicides involving the use of a firearm. Put differently, it is not the case that states have roughly the same levels of suicide, and people in different states simply choose to end their lives in different ways. Where there are more guns in homes, there will be more successful suicides. This is because firearms dramatically increase the lethality of suicide attempts that might otherwise fail.[6]

3. Weak gun laws lead to more guns in homes

So far I have argued that gun owners are not more suicidal than non-gun owners, but because firearms increase the lethality of suicide, individuals who decide to take their own life are more likely to succeed if they have access to guns. Thus higher levels of gun ownership lead to higher numbers of successful suicides. The next question is whether policy is implicated in this: do weaker gun laws increase gun ownership? Unsurprisingly, the answer is yes. Figure 1 uses data from the Open Society Institute on the strength of gun laws across the 50 states. It shows that, predictably, there is a strong negative correlation (-0.764) between the strength of a state’s gun laws as measured on the OSI index and the prevalence of gun ownership.

4. Gun laws that “enhance self-control” lead to fewer suicides

To bring the analysis back to Schelling, we need to ask whether the kinds of gun laws of adopted by some states help to enhance self-control in the ways Schelling describes e.g. “removing the mischievous resource”, “submit to surveillance”, “arrange rewards and penalties” and “arrange delays”. For example, some states have waiting periods of up to 72 hours before purchasers take ownership of a gun. Others require background checks, permits to purchase or possess a firearm, and evidence that firearms will be stored securely. Minimum ages for purchase and possession also vary, which is important when suicide is a major risk factor for young adults.

To test this, we strip these features out of the OSI index to construct a new indicator. If the anti-soft paternalists are right, then this indicator should have no relationship to suicide rates: if people truly want to take their own life, they will do so. In fact, however, we find that this “enhancing self-control” indicator is reasonably strongly correlated (-0.642) with lower firearm suicide rates (see Figure 3).

Bringing it together

Table 1 brings this analysis together. It reports the result of two regression models. The first naïve regression estimates the impact of gun laws which enhance self control on total suicide rates. It finds a statistically significant effect. The second model adds covariates to control for factors that are associated with higher numbers of suicides, including the number of attempted suicides, and the geographic, age and ethnic profile of the population. These controls do not attenuate the size of the effect of self-control enhancing gun laws. To put the size of this effect in context, if a state with the average suicide rate of 12.6 per 100,000 were to achieve a 5 point increase in its score for self-control enhancing gun laws (which could be achieved, for example, by adopting a waiting period for handgun purchases) the regression would predict a drop in the suicide rate in that state of 2 people per 100,000 or about 16%.

Conclusion

To sum up, from the empirical data we can detect the following causal chain: weak gun laws mean more guns in homes; more guns in homes mean more suicide attempts using firearms; more suicide attempts using firearms mean more successful suicides; therefore, weak gun laws mean more suicides.

On this view, the tragedy is not just that so many should decide to take their own life, but that ready access to firearms means that some of those who did not “mean” to take their life end up doing so. We can therefore reject the null hypothesis that the number of people who take their own lives in a given jurisdiction is uncorrelated with the strength of the gun laws within that jurisdiction. When it comes to suicide, people kill people, by definition. But weak gun laws help. By contrast, gun laws which enhance self-control by restricting access to firearms save lives.

This analysis of the links between firearms and suicides also suggests that at least on some issues it is possible to distinguish between different “selves” to find individuals’ true interests, despite critics’ concerns to the contrary. If it is possible here I see no reason in principle why it shouldn’t be possible in other areas, as long as the necessary procedural safeguards and ex post evaluations were in place to give all “selves” a chance to express themselves. Of course, what those safeguards should be remains an important question.

Paul Skidmore is a second year MPA candidate at Princeton University’s Woodrow Wilson School.