An increasing number of jurisdictions around the world have implemented some form of legalised assisted suicide (or are actively considering doing so). It is a highly contentious issue and one in which debates are rightly driven primarily by considerations of rights, morals, and ethics. However, one key argument in many of these debates has been a more practical issue, namely, whether such laws lead to an increase or decrease in suicide rates.
This aspect of the debate has become even more pressing, given the concerns about the effect COVID-19 restrictions and lockdowns have had on mental health, with obvious potential implications for suicide (Ribeiro 2020).
It is natural to think that legalising an activity should reduce its costs and lead to an increase in the amount of that activity. In this case, legalising assisted suicide should not only reduce practical barriers to committing suicide but may also lower societal taboos against suicide. Hence, we might expect to observe an increase in suicide rates overall (Dugdale and Callahan 2017). In 1995, however, Richard Posner proposed the alternative view that the legalisation of assisted suicide could reduce suicide rates in some circumstances. How did he arrive at such a counter-intuitive conclusion?
The Posner hypothesis
Posner (1995) used a formal economic model to explain his hypothesis, but his underlying intuition is quite simple and based on the option value of future suicide. People who have no immediate wish to die by suicide get reassurance from knowing that they can exercise the option of suicide at some point in the future, should their life experience deteriorate. Someone in the early stages of a degenerative illness might worry that when their quality of life has deteriorated to such an extent that they no longer wish to live, they will no longer be able to die by suicide without assistance. As a result, some people in that situation may choose to die by suicide at an early stage.
Now, if people have the assurance that assisted suicide will be available should they end up wishing to die in the future, some of them may be less likely to take their own lives in the earlier stages of their illness. Put another way, assisted suicide will substitute for unassisted suicide. Further, at the point when the person believed they would want to die by suicide, they may find that they no longer wish to do so. Or a diagnosis may be wrong or unduly pessimistic and the person finds that future life is less painful or debilitating than originally expected. In either event, the person may never actually take up the option of suicide.
There are two key empirical predictions from Posner’s hypothesis. First, legalising assisted suicide should reduce unassisted suicides to some extent. Second, it is possible that under some circumstances legalisation could reduce the total number of suicides (assisted and unassisted) combined.
Posner argues that the key to understanding these counter-intuitive ideas is to recognise that unassisted and assisted suicide are two distinct ‘goods’ which substitute for each other and that:
“…lowering the price of the second (by legalising it) will reduce the demand for the first and nothing in economics teaches that this reduction must be fully offset by the increased demand for the second good” (Posner 1995: 250)
Testing the Posner hypothesis
Posner’s work pre-dated Oregon’s 1997 assisted suicide law (the first in the US) and so he was unable to provide any empirical evidence to test his hypothesis. This has not stopped his hypothesis from being used as part of arguments in favour of assisted suicide laws (see, for example, Carter vs Canada (Attorney General) 2015, Dignitas 2014).
Since 1997, assisted suicide laws have been implemented in an increasing number of US states, meaning empirical analysis of his ideas is now possible. This is the task we took on in our paper (Girma and Paton 2022). We examine suicide rates in ten US states that implemented an assisted suicide law up to the end of 2019. Our basic research design is to compare changes in suicide rates before and after legal implementation relative to changes in states that did not pass such laws. The analysis considers both total suicide rates (i.e. including assisted suicides) and unassisted suicide rates.
An obvious issue with this type of analysis is whether states that legalise assisted suicides are sufficiently similar to other states to allow us to be confident that we are truly estimating a causal effect of the laws. We deal with this problem in two ways. First, we use an event study approach on panel data. In addition to including fixed effects to control for time- and state-specific unobservable factors, we control for trends in suicide rates in the years before the law was introduced and test to confirm that these trends are similar on average in other states (the so-called parallel trends assumption). We also control for other factors which are known to affect suicide rates, such as demographics (Stack and Kposowa 2007), availability of firearms (Lang 2017), substance abuse (Freeman 2007), and unemployment (Nordt et al. 2015).
Our second approach to establishing causality is to use a synthetic control method. This method involves constructing an artificial set of non-legalising states which are otherwise similar (based on a range of demographic and economic variables) to states legalising assisted suicide before legalisation. We can then compare relative trends in suicide rates after the passage of the law.
Do the data support Posner?
Both approaches yield similar results. There is very strong evidence that the legalisation of assisted suicide is associated with a significant increase in total suicides. Further, the increase is observed most strongly for the over-64s and for women. To give an idea of the size of the effect, the event study estimates suggest assisted suicide laws increase total suicide rates by about 18% overall. For women, the estimated increase is 40%.
There is weaker evidence that assisted suicide is also associated with an increase in unassisted suicides. The effect is smaller (about a 6% increase overall, 13% increase for women). It is still statistically significant in the main estimates but not in all of the robustness checks, meaning we have less confidence in that result. However, we find no evidence that assisted suicide laws are associated with a reduction in either total or unassisted suicide rates.
The results from our paper are consistent with another recent paper looking at trends in European countries that have introduced either assisted suicide or euthanasia (Jones 2022) and which concluded that there is “no reduction in non-assisted suicide relative to the most similar [non-legalising] neighbour and, in some cases, there is a relative and/or an absolute increase in non-assisted suicide”.
The finding that assisted suicide laws have a relatively bigger effect on women is also consistent with previous research (Canetto and McIntosh 2022). Given that unassisted suicide rates tend to be much lower amongst women than men, one interpretation of this finding is that the higher take-up of assisted suicide reflects women being empowered to take control over end-of-life decisions. Canetto and McIntosh (2022) propose an alternative view. They suggest that higher take-up may reflect the disempowerment of those who are more vulnerable to social pressure to die by suicide – for example, through feeling a burden to relatives or society – and that women are overrepresented in such groups.
To date, there seems to be little if any real-world evidence in support of Posner’s hypothesis that assisted suicide laws might reduce suicide rates. That does not necessarily mean that Posner got it completely wrong. It is possible, for example, that assisted suicide laws do induce some substitution from unregulated to assisted suicide, but that this effect is neutralised by an increase in unassisted suicide arising from, say, a reduction in societal taboos associated with suicide.
So where does this leave debates over assisted suicide? The ethical basis for those in favour of assisted suicide laws is that they promote the right to self-determination and bodily autonomy. In contrast, those opposing assisted suicide argue that such laws infringe the rights of vulnerable people, such as the elderly, disabled and terminally ill, and those who may be placed under undue pressure to choose to end their lives prematurely. The empirical evidence that assisted suicide laws increase suicide rates overall does not resolve the principle of the underlying ethical debates, but we can hope that it helps to put those debates into some sort of practical context.
Canetto, S S, and J L McIntosh (2022), “A comparison of physician-assisted/Death-with-Dignity death and suicide patterns in older adult women and men”, American Journal of Geriatric Psychiatry 30(2): 211–20.
Dignitas (2014), Submission 67 by Dignitas to Medical Services (Dying with Dignity) Exposure Draft Bill 2014, 20 August, Parliament of the Commonwealth of Australia.
Dugdale, L S, and D Callahan (2017), “Assisted death and the public good”, Southern Medical Journal 110(9): 559–61.
Freeman, D G (2007), “Drunk driving legislation and traffic fatalities: New evidence on BAC 08 laws”, Contemporary Economic Policy 25(3): 293–308.
Girma, S, and D Paton (2022), “Is assisted suicide a substitute for unassisted suicide?”, European Economic Review 145: 104113.
Jones, D A (2022), “Euthanasia, assisted suicide, and suicide rates in Europe”, Journal of Ethics in Mental Health, open volume (Feb): 1–35.
Lang, M (2017), “Firearm background checks and suicide: Learning more about firearms and suicide”, VoxEU.org, 6 October.
Nordt, C, I Warnke, E Seifritz and W Kawohl (2015), “Modelling suicide and unemployment: A longitudinal analysis covering 63 countries, 2000–11”, Lancet Psychiatry 2(1): 239–45.
Posner, R A (1995), Aging and old age, Chicago: University of Chicago Press.
Ribeiro, E (2020), “Fighting suicide during COVID-19: Lessons from past pandemics and recessions”, VoxEU.org, 12 August
Stack, S, and A J Kposowa (2007), “The association of suicide rates with individual-level suicide attitudes: A cross-national analysis”, Social Science Quarterly 89(1): 39–59.