Accounting for Consequences and Claims in Policy

Fully appraising any policy requires us to capture all of its ripple effects and not simply the size of the splash when the pebble of intervention hits the water. It also requires that we weight the value of those effects according to morally relevant characteristics of people, such as their age. In this paper, I discuss the consequences that should feed into policy appraisal and the claims that different groups may have on resources.


Introduction
One of the main aims of government is to reduce misery and suffering by as much as possible, subject to resource constraints. Fully appraising any policy requires us to capture and quantify all its possible short-and long-term ripple effects, and not simply the size of the splash when the pebble of intervention hits the water. Indeed, sometimes the ripple effects might turn out to be much more significant than the initial splash, especially in the longer term.
In their response to the pandemic, policymakers have prioritised preventing deaths from COVID-19 above everything else. Concerns for lives have trumped concerns for life expectancies, let alone concerns for life experiences, which have been largely ignored. Policies have focussed on the splash with much less regard for the ripple effects. This has resulted in the use of very restrictive mandated non-pharmaceutical interventions (NPIs), or "lockdowns". Regardless of what our individual views on such responses may be, important lessons can be learned in terms of how policymakers can and should respond to future crises. Indeed, there are lessons that come out of COVID-19 for how to make better policy decisions in calmer times too. Figure 1 sets out, in a highly stylised and linear way, the ten main steps to effective decision-making. It illustrates how step 3 -the expected splash -has dominated decision-making, whilst steps 4-9 -properly accounting for the ripple effects -have largely been bypassed. But these are crucial steps if governments are to minimise the harms caused by a pandemic, as well as if they are to use their resources wisely at any time. Anyone who understands the reality of living in a society with limited resources will recognise the incongruence of trying to make good policy decisions whilst ignoring the broad aggregated effects that those decisions have across society.
In this paper, I will focus on steps 6 and 9 above, and especially distributional concerns, which can sometimes be an afterthought for economists. I will consider the ways in which we might bring together the myriad ripple effects of any policy decision into a single metric. Much as the ripples on a pond can be quantified in a single metric of displaced
Quantify effects in each sector (e.g., expected effects of loneliness) 6.
Compare to costs (i.e., cost-benefit analysis or cost-effectiveness analysis) 9.
Account for distributional concerns (equity weighted CBA or CEA) 10. Decision based on expected effects (compared to counterfactual) much like preferences, evaluative measures of SWB feed into our feelings. Indeed, reports of life satisfaction are arguably closer to a preference-based account of welfare than they are to the mental state account [13]. The extent to which this is a good or a bad thing is beyond the scope of this paper.
Whatever measure(s) of SWB we use, for the purposes of economic appraisal, it is entirely possible and legitimate to express its benefits in a single, non-monetary metric that can then be compared to its costs, so generating a cost-perbenefit unit. Cost-per-QALY estimates, for example, allow us to determine the allocation of resources that would generate the most QALYs for a given budget allocated to healthcare. If all benefits across different sectors could be captured in WELLBYs, then the resources devoted to the public sector could be distributed in a way that would generate the greatest number of WELLBYs. This would be to use resources as efficiently as possible.
For the purposes of cost-benefit analysis (CBA), benefits need to be expressed in the same monetary units as costs. This will enable us to determine whether a policy intervention was worth it per se -by generating more benefits than costs -in addition to whether it was worth it compared to other interventions. In principle, this will enable us to determine the size of the public sector and how best to allocate resources within it. There are several ways of monetising well-being, but each raises several challenges beyond the scope of this paper. Suffice to say that cost-per-WELLBY for analysis might be where we finally end up.
Wherever we do end up, we must be able to show just how much water is displaced when the pebble of policy intervention is dropped into the water. The consequences of policies introduced now are likely to have tidal waves of effect across many sectors and domains of society for years to come. These must be taken into account in deciding what action to take.

Distributional concerns
At the societal level, citizens and policymakers care not only about how many life years, QALYs, WELLBYs or whatever are being generated per pound spent but also about how those benefits are distributed across people. Just as we care about national income and about inequalities in income, we care about the size of the well-being cake and about how fairly the slices are distributed. The fairness of the distribution will be determined by the legitimacy of the claims that different individuals or groups can make on resources [14].

Five claims on resources
There are some general statements that most of us would agree with when considering where different groups of people should be placed in the queue for scarce public resources. Other things being equal, those with the greatest claim on resources (those with the most ethically justifiable reasons for being towards the front of the queue) are those who most fully satisfy the five conditions discussed later. 2 In principle, all five could be considered when setting priorities. There is no simple way to resolve any "dispute" between the conditions, and different trade-offs between competing claims will doubtless be deemed acceptable in different circumstances. In determining how to weight each competing claim, public preferences over various trade-offs can help reach conclusions, so I discuss some of this evidence here. 3

Those who have the most to gain
This is a claim to resources based on an individual's "capacity to benefit" as a result of any intervention. The more a person can expect to benefit from intervention, the closer to the front of the queue they should be. If Person A can expect to gain ten years of extra life from an intervention, then their claim is greater than Person B, who can expect to gain only ten months of extra life. Allocating resources only on this basis would mean that resources were being used so as to maximise the bang for the buck. Capacity to benefit is all that matters when we conduct CBA or cost-effectiveness analysis (CEA), for example, and it would mean generating as many WELLBYs as possible. On a fundamental level, capacity to benefit is a prerequisite for a person to even be in the queue: it would be a waste of resources to "treat" them if there was no expected benefit at all. But it is not all that matters in determining their position in the queue. There are other morally relevant claims too.

Those who are currently suffering the most
This is a claim to resources based on "severity of condition" in the absence of intervention. Let us take the two individuals discussed above. If Person B will die shortly without an intervention and Person A will live for another five years regardless, then our priorities might change. There are now good ethical grounds for putting Person B in front of Person A in the queue for resources because their prospects are more immediately severe in the absence of intervention. The ethical grounds for putting Person A in front of Person B on the grounds that they will benefit from treatment more remain, however. Herein lies a classic trade-off between claims based on benefit and those based on severity.
I have conducted various empirical investigations into what the public thinks about how these competing claims should be accounted and traded off against one another. In a nutshell, they suggest that a person's capacity to benefit and the severity of their condition both matter in every decision context I have enquired into, from triage decisions to macro resource allocations [15]. Context matters, of course, and it is impossible to provide a consistent value for the exchange rate between them. Suffice to say that the public would want policymakers to account for what happens to people both in the presence and in the absence of an intervention when deciding what to do.
2.1.3. Those who will suffer the most over the lifetime Capacity to benefit and severity of condition are both prospective assessments of what the profiles of future well-being look like in the presence and absence of intervention. But some people might have a legitimate ethical claim to be further up the queue based on a retrospective assessment of what happened in their past, as well as based on what is expected to happen over their lifetime. Imagine that capacity to benefit and severity of condition are the same for two individuals or groups, but that one has experienced more suffering -or less well-being -in the past than the other. We might prioritise them on this basis. Or we may wish to afford higher priority to those who are expected to experience more suffering over their entire lifetime. We might also decide to prioritise those who have had less opportunity for well-being, such as when we prefer to give a given benefit to younger over older people.
It is also clear from public preference data that members of the public wish to account for well-being over the lifetime [16]. In one of my empirical studies on priorities by age, we controlled for capacity to benefit by asking people to prioritise a fixed five-year gain in life expectancy for people at age 5, 20, 35, and 55. In the study, 70-96 per cent of people had one of the three youngest ages ranked first [17], with "having lived less" life being the main reason for prioritising younger people [18]. This is consistent with the fair innings argument (FIA) -the egalitarian principle that everyone is entitled to some "normal" span of health (usually expressed by life-years) and that anyone failing to achieve it has been "cheated" [19].

Those who are suffering because of "bad luck"
How well or how badly our life turns out is determined by factors that lie on a spectrum from being entirely outside of our control (exogenously determined) and entirely within our control (endogenously determined). Although nothing is ever truly chosen, most people would locate illness caused by a genetic cancer to be closer to the exogenous end of the spectrum than illness caused by the "choice" to go mountaineering. As such, someone suffering because of "bad luck" would have a claim to be closer to the front of the queue than someone suffering, at least in part, because of "bad choices" [20].
The cause of any need for an intervention is a hotly contested area of ethical discourse, and public preferences are mixed and malleable. In some of my own work, people became much less "harsh" on bad choices when they had been given opportunity to discuss and reflect upon why some people make very poor decisions, at least insofar as their heath is concerned [21]. For what it's worth, as I understand more about how so much of what we do is outside of our control, I am less convinced about the moral relevance of the causes of the need for resources than I was a couple of decades ago [22]. But if we accept that we have a modicum of agency (which most people do), then truly exogenous "bad luck" will reflect a legitimate claim to be closer to the front of the queue for resources [23].

Those who have the greatest impact on others
Imagine that you are responsible for allocating one donor kidney, and you have two people whom it matches equally well. One is a 40-year-old homeless man with no family, and the other is a 40-year-old happily married man with two young children and parents that he supports emotionally and financially. Whom would you choose? Either decision is morally justifiable (including tossing a coin to decide), and your instincts might be to discount the wider benefits that might come from treating one person over the other. (The homeless person might also have experienced "bad luck", which shows how interconnected the various claims can be.) In any case, the important point here is that it is legitimate to choose the married man on the grounds that he has the greatest capacity to benefit other people Unsurprisingly, the expected impact on other people is also hotly contested [24]. The utilitarian solution, which is embedded in the principles of CBA, is to treat benefits to other people no differently than benefits that go directly to the recipients of the intervention. Why treat the ripple effects any differently from the splash? Admittedly, not all of us are utilitarian, but there will most likely come a point at which most of us would want to account for spillover effects: imagine our married man was an integral part of a team that was on the verge of a cure for cancer. This somewhat extreme example illustrates how complex ethical decisions can be and how no single principle can be seen to dominate any other in all contexts.

Claims in the context of COVID-19
Based on the foregoing discussion, we would expect policymakers to account for a person's capacity to benefit, severity of condition, and lifetime suffering in their prioritisation decisions. And yet in the case of COVID-19, arguably only severity has been considered. It has been very interesting to me that any attempts to raise concerns about capacity to benefit and lifetime well-being have been greeted with moral outrage. Many of those who have died from COVID-19

Dolan: Accounting for Consequences and Claims in Policy
Art. 3, page 5 of 8 would have died within the next year -up to half of them, according to Professor Neil Ferguson, a government adviser on SAGE. Despite the public outrage, this is morally relevant.
This in no way suggests that we simply let people die, but rather that we manage their deaths properly so as to create as much benefit as possible both for the dying person and, crucially, for those left behind. We must do more to accept death, especially in old age, and to minimise the impact death has on family and friends. Some of you might have baulked at the idea of treating a cancer specialist over a homeless person, but the utilitarian arguments for maximising the benefits from an intervention become much more compelling when they are framed around minimising the suffering of all those affected by someone's death.
The impact of a death is unquestionably affected by the age of death. The average life expectancy for an 18-year-old in the UK today is around 81. According to the ONS, around 60% of the deaths from COVID-19 in the UK have been in people who are this age or older. Substantively, most of those bearing the biggest burden from COVID-19 won't live for as long as those who are dying from it. If people's lifetime prospects are an important measure of human welfare, then we have engaged in one of the biggest redistributions of resources from those who have the least to those who have the most in human history.
It is baffling to me that there has not been more -or any -real discussion of the potential injustice of this. Ever since I worked with Alan Williams at the University of York in the 1990s, I have supported the FIA. I contend that it is a sentiment shared by most of the public. You could say that we are ageist, and we are. But not all -isms are unfair. If we afford a 35-year-old priority over a 70-year-old now, then in 35 years' time, the current 35-year-old will be given less priority as a 70-year-old, so everyone is treated equally over the lifetime. This is assuming that we are all lucky enough to live that long, which about one in five current 35-year-olds won't. When we look at a cohort of older people, we are subject to "survivorship bias" -we see only the lucky ones who have survived and do not properly consider all those who have been cheated out of a fair innings.
I should again stress here that the FIA I have in mind here does not suggest that older people should simply be allowed to die because they have already achieved a fair innings, but only that they should be afforded less priority for life-saving interventions than those who have not yet lived as long. During the pandemic, younger people have been denied access, for instance, to cancer diagnoses and treatment to prevent the deaths of older people from COVID-19. This represents a gross violation of the FIA. You might say that it is all well and good to have rational principles in calmer times, but that a pandemic represents an immediate threat to life, so all efforts should rightly be directed towards mortality risks (severity of condition). Perhaps, but times of crisis arguably make rational considerations even more important.
Besides, concerns for lifetime well-being don't just disappear during a pandemic, as empirical work I conducted with Amanda Henwood and Aki Tsuchiya over this last year attests [25]. We conducted an online discrete choice experiment (DCE) to elicit the preferences of the UK public on two occasions: May 2020 (n = 6,153) and February 2021 (n = 1,024). The DCE asked people to make trade-offs across four attributes, including excess deaths above age 70 and excess deaths below age 35. We found that the relative value of mortality above 70 to below 35 is roughly 1:24. These preferences were stable across the two surveys and across respondents of different ages.
A more sophisticated version of the FIA would account for life experience as well as life expectancy. It is impossible to know what ripple effects COVID-19 would have caused if we had pursued a different set of policy options (step 10 in Figure 1 above), such as the focussed protection of older people [26], so we must all be measured in our judgements of what should have been done. But we do know that lockdowns have disproportionately affected families of low income [27], that domestic violence has increased for those at risk [28,29], that school closures will further widen the attainment gap [30], and that social distancing causes greater harm to those with pre-existing mental health conditions [31], lower starting well-being [32], younger people [33], women [34], and children from disadvantaged backgrounds [35]. Some of these effects on mental health may not be easily restored [36].

Conclusion
This paper is not only, or even primarily, about COVID-19. The 10 steps to effective decision-making set out in Figure 1 are always required for policymakers. The process of going through each step, especially steps 4 and 5, ensures that the important downstream effects of policy are properly accounted for. These consequences might affect a population group that is largely ignored (such as young adults who do not go to university), a dimension of well-being that falls between the cracks of government departments (such as loneliness), or that will occur some time into the future (such as the effects of childhood development on later life). We should be doing all we can to ensure that equity-weighted well-being measures are developed as quickly as possible and used as widely as possible.
The significance of these steps has been magnified during the pandemic. During the last year, around the world, the dominant response to COVID-19 was to seek to significantly reduce social contacts through MNPIs, or "lockdown". This pebble of intervention -perhaps the biggest stone that ever has been dropped into the water -has been assessed almost entirely in terms of its effects on the splash of mortality risks (step 3 in Figure 1). Most of the significant ripple effects have barely been listed (step 4) let alone quantified (step 5). We must do better in the future. In this paper, I have focussed partly on generating a single index (step 6) and mostly on distributional concerns (step 9), which will come after the single index has been monetised (step 7) and compared to costs (step 8).