The health vs wealth trade-off and COVID-19 policymaking
Valuating health vs wealth: The effect of information and how this matters for COVID-19 policymaking
Shaun P. Hargreaves Heap, Christel Koop, Konstantinos Matakos, Asli Unan, Nina Weber06 June 2020
In the absence of medicine-based responses to COVID-19, policymakers have to rely on behavioural interventions to control the transmission of the virus. Yet, at least in the short run, these interventions create a trade-off between saving lives and saving the economy – a ‘health–wealth trade-off’. The lockdown restrictions on individual freedoms, for example, saved lives by preventing transmission, but they came at the cost of lost output and income (Baldwin and Weder di Mauro 2020).
Relaxing a lockdown involves the same trade-off but in reverse: the economy recovers but at a cost of an increased threat of virus transmission. There are similar potential medium-term trade-offs in the choice between strategies of herd immunity and so-called ‘hammer and dance’ (Pueyo 2020). Faced with such health–wealth trade-offs, how should policymakers decide where to locate the balance?
This question has led policymakers to dust off risk-valuation manuals to see whether the proposed movement makes cost-benefit good sense. For instance, the OECD (2012) suggests, on the basis of evidence, that a figure between $1.5–$4.5 million should be used as the statistical value of life, with a base value of $3 million (the EU uses a slightly wider range and a higher base value of $3.6 million).
Armed with such a figure, consider the UK lockdown. The Imperial College modellers predicted that lockdown would save several hundred thousand lives (Ferguson et al. 2020), whilst current estimates point to a lockdown loss of more than 10% of GDP (Bank of England 2020); 10% of GDP is £221,489 million. Thus, even if only one, rather than several, hundred thousand lives had been saved, the lockdown would have cost about £2 million per life saved and would have produced a net benefit, using the OECD’s base figure valuation of a life at $3 million. Beyond 100,000 lives saved, the net benefit from lockdown quickly becomes enormous.
Setting aside arguments against using cost-benefit analysis in such decisions, there are two problems with recycling OECD (or other) estimates of the value of a statistical life. First, the figures vary widely and – although the OECD proposes a range, any part of which would probably make the lockdown yield a net benefit – they refer to the value of a statistical life. COVID-19, however, produces mortality disproportionately in the old and those with co-morbidities.
So perhaps the alternative quality-adjusted life-year (QALY) approach – whereby quality-adjusted life-years saved are valued – should be used. With as much as 20% of those over 70 who die from COVID-19 being likely to die within one year anyway, and a QALY value of between £20–30,000, a positive net benefit from lockdown becomes much less likely.
The second problem is deeper. The policy choices relate to COVID-19 deaths; not imaginary deaths in the past and not some medical intervention that might prolong life by one year. Thus, what plausibly matters is how people currently value such deaths. This is for two reasons.
First, how people value health over wealth in this period obviously matters for the politics of these decisions. Politicians are re-elected by the people who assess government performance by their own lights; not by how well the government executed a manual’s implicit assessment.
Second, people’s current valuation may matter because there is evidence that people are more likely to follow rules when they agree with them (Tyler 1990). We don’t know whether this is the case for COVID-19 lockdown policies, but if it is, then the efficacy of any lockdown or its relaxation also depends on whether its movement on a health–wealth trade-off accords with the public’s relative valuation of health over wealth.
Of course, although these considerations are real, they would not matter if we had reason to believe that people’s assessments in such matters were relatively constant (in that case, the OECD’s summary might apply). Yet we don’t. In fact, we know such assessments are extremely sensitive to exposure to the risk in question (e.g. Gigerenzer 2014) – think of the exodus from flying in the US post-9/11.
So, features specific to COVID-19 will influence people’s current COVID related health–wealth valuation. This is why we conducted a survey experiment in the UK and the US to see how people actually value health over wealth in the context of COVID-19 and the lockdown (Hargreaves Heap et al. 2020).
The survey consists of a sequence of binary choices between pairs of health and wealth outcomes for each country. The sequence is designed so that anyone who values both ‘lives saved’ and ‘average household income’, and has a preference ordering over their various combinations, will switch at one point from the option that always does as well or better than the other on health, to the option that does better on wealth. Where they switch is an indicator of their relative valuation of health over wealth and can be interpreted as giving the lower bound for the value of life.
The experimental element comes from asking participants to make this sequence of choices a second time. After the first occasion, the participants engage in another task and are then divided into three sub-groups. One sub-group receives an information prompt on the predicted number of COVID-19 deaths (per million) by August; another receives an information prompt on the predicted loss of GDP per capita through lockdown measures in 2020. The final sub-group is our control group; they hear a short piece of instrumental music instead.
The participants then make the decisions over health and wealth combinations again. This experimental element allows us to test for the robustness of the health–wealth valuations (do they change much in the control?) and for whether they respond in predictable ways to information on deaths and income losses (in the two information treatments).
We find that in the peak of the pandemic in the UK and US, the average implicit value of a COVID-19 life was extremely high by the standards of the OECD figures. The average value is a factor of around ten times greater than usual. This means that the lockdown policy accorded very well with the public’s preferences for health over wealth at that time. Indeed, the lockdown policies were met by increased support for the governments imposing them (Blais et al. 2020).
We also find that the individual variation in this valuation predicts individual compliance with the lockdown measures in both countries. So, policymakers do need to care about what the people think in this matter, not just for political reasons of re-election, but also because it affects the efficacy of any policy change through its influence on compliance.
This experimental evidence gives us a clue as to whether the valuations will change as the experience of COVID-19 deaths recedes and income losses mount (see Midoes 2020): they will.
On the one hand, the high valuation of COVID-19 lives saved implies that any relaxation of the lockdown – which will produce a spike in COVID-19 deaths – will court unpopularity and lack of compliance. On the other hand, information on COVID-19-related fatalities and growing income losses will lead people to adjust their relative valuation of health over wealth towards wealth (see Hargreaves Heap et al. 2020).
In short, the longer that relaxation of the lockdown is delayed, people will become more tolerant of any increased deaths associated with the relaxation because income losses increase with delay. This insight complements an earlier one on the gains from the delay that come from better information on how to relax a lockdown (Santos-Pinot and Mata 2020).
Moreover, whilst unexpected deaths will, in general, put a brake on the drift to valuing wealth relatively more highly, they also tend to polarise the valuations of health over wealth in the UK (Hargreaves Heap et al 2020); this is in a context where there is already evidence of polarisation in other respects (see Gambacorta et al. 2020). Both effects reinforce the argument for caution in relaxing the lockdown because a delay will help avoid the braking and polarising effect of unexpected deaths on people’s valuation of health relative to wealth as it otherwise drifts – in both the UK and the US – towards wealth.
The politics of loosening (and potentially re-tightening) lockdowns are not straightforward (see Aksoy et al. 2020), not least because standard tools employed, such as information provision (see Fetzer et al. 2020), might generate unintended consequences that undermine policy objectives. At a minimum, they should at least be facilitated by analyses of the public’s relative valuations of health and wealth.
Ferguson, N M, D Laydon, G Nedjati-Gilani, et al. (2020), “Impact of non-pharmaceutical interventions (NPIs) to reduce COVID-19 mortality and healthcare demand”, Imperial College COVID-19 Response Team, Report 9, 16 March.