Sometimes research gets left in a drawer. The virtual equivalent can happen to blog posts. In clearing out some old drafts, I found a blog post on capabilities that it seems I wrote in May 2014. I don’t remember what was going on in my life back then. But it was before I had embraced the Oxford comma (I have added one to the title). It seems that this blog post is not entirely useless or, at least, it will be no more useless if released into the wild.
So, here it is, an unintended time capsule. Why I didn’t finish it, I don’t remember. Perhaps I thought it would be superseded by my writing of a book chapter on a related topic.
I haven’t made any edits to bring it up to date, and I’m unsure of the extent to which I agree with all of its content. I have added a few notes in brackets.
Proponents of the capabilities approach in health have reached a hurdle; how to value states. Depending on what the Senists hope to achieve, it could be a hurdle of their own making. Here, I ask what it is that proponents of the capabilities approach in health seek to achieve. From there, I make some suggestions for future work depending on this choice. It’s important to note that the interpretation of utility to which I (in this post) subscribe is defined by individuals’ choices (decision utility), thus allowing for caring externalities and other unorthodox determinants of a social welfare function.
The three possible goals of the capability approach in health, as I see them, are:
- To broaden the evaluative space for health and social care
- To differentiate between functionings and capabilities and value interventions in terms of the latter
- To establish the intrinsic value of capabilities that cannot be captured by individuals’ choices and account for this in allocation decisions
Recent soul-searching by Senists, I believe, implies that some are striving for goal #3 [note: I cannot think what papers I had in mind at the time]. Here I argue that, if this is the case, they are failing miserably. If the aim is goal #1 or #2, then Senists need to embrace more modern interpretations of welfarism in their positive analyses, while maintaining the normative focus on capabilities: embracing extra-welfarism over non-welfarism.
The complete capability choice set
Imagine two possible binary functionings: A and B. There are four possible capability sets and, therefore, four possible functioning states. Under a capabilities framework – and allowing for capability sets to hold different utility values – there are nine possible states in which an individual can exist. This represents the complete capability choice set:
|Do do | Can do||Neither||Just A||Just B||Both|
Some work has been done to investigate preferences for capability sets and the utility of choice, but it’s far too complicated to discuss here. For now, let’s keep it simple. For a particular individual participating in a valuation exercise, a standard choice experiment could elicit a ranking of 9≥8≥7≥6≥5≥4≥3≥2≥1. That is to say that the individual’s preferences are such that they attach more value to B than A, both in terms of doing it and being able to do it (i.e. functioning and capability). Various valuation methods could be used to attach relative values to each of these states.
Choices vs values
If the goal of the capability approach in health is to broaden the evaluative space (goal #1), then the process is simply a matter of considering B in the valuation, where before only A was included; for example, if A was ‘mobility’ and B was ‘having love and friendship’. There is no need to quantify a capability’s value, only to assert that its value is greater than 0 and therefore included in the choice set. It would only be necessary to value states 1, 5, 7 and 9. It is quite reasonable to value states in terms of their utility, using the methods we are familiar with. The ICECAP measures succeed in this regard.
If we adopt goal #2, then only choices about capabilities hold value, and choices about functionings have zero value. However, it remains necessary to capture the utility value that people attach to their functionings. In valuing the capability, they will undoubtedly consider the choice they would make (or be able to make) about the functioning. That is to say that the ‘value’ individuals attach to any given capability set will depend upon which functioning set they would choose. As such, it is necessary to try to get an individual to differentiate between, for example, states 2 and 5. Whether an individual can conceptualise this difference is questionable. However, recent research has suggested that people can grasp concepts of capability. The ‘value’ of the capabilities lies in the difference in utility between state 1 and states 2, 3 and 4 for each capability set.
These differences exclude the benefits derived from functioning (e.g. the difference between states 2 and 5) and focus solely on capabilities. Values could be greater than, less than or equal to zero. Current methods, I suspect, succeed in producing an approximation of this value. For the individual described above, the ICECAP approach would not elicit values for states 1 through 4 or states 1 through 9 but states 1, 5, 7 and 9. For an individual with a different order of preferences, it would elicit values for different states.
Current methods fail to estimate the targeted ‘value’ because they include utility associated with functioning and do not isolate capability value. As a result, the measure is biased by the respondent’s preferences about functioning. The value associated with a capability that tends to have greater value in functioning is inflated compared to other capabilities. However, thanks to societal valuations being the norm in health economics, such an approach would at least include individuals with different preferences and reduce the upward bias associated with functionings. This approach could, of course, be viewed as welfarist. However, if the analysis can succeed in isolating and excluding from the welfare function the utility benefits of functioning, then this would clearly be non-welfarist. Research suggests that it is possible for capabilities to exceed functionings, and for people to value capabilities higher than functionings. Work like this highlights the potential value in striving for goal #2, with a reasonable means of fitting the capability peg into the QALY hole.
I do not know what Sen’s views on this approach would be. Still, it seems to address physical condition neglect by isolating the functioning value from the capability value and addresses valuation neglect by adopting a societal perspective.
If the Senists are striving for goal #3, things become a lot more difficult.
Goal #3 insists that things are of value not (simply) because they enhance utility, but in and of themselves. They have intrinsic value as well as instrumental value. So how do we elicit this extra ‘intrinsic’ bit? One thing it cannot be based upon is choice. Even if we ask about the particular intrinsicities [is that a word?] of a capability, whatever is derived from choices is by its very definition a reflection of its utility value. Sen isn’t very definitive in his writing, but one thing he has written is that “The choice approach to well-being is… really a nonstarter”. As a result, some Senists are at pains to distance their methods from the choice approach, suggesting that goal #2 is insufficient.
Cookson suggests that we might be able to work around this by interpreting willingness to pay questions for capability sets as value judgments, rather than as desires or choices. This ‘reinterpretation’ simply brings us back to goal #2, which I believe to be theoretically sound. Nevertheless, it seems there is a desire to move away from choices. If we adopt goal #3, it becomes necessary to identify the difference between choices and values and accept that the two are mutually exclusive. If we accept that individuals, by definition, (try to) choose the utility-maximising option available to them – hypothetical or otherwise – then we are presented with a paradox. Sen argues (as far as I can tell) that this utilitarian perspective fails to capture certain things that we value, pertaining to capabilities. Therefore, the capabilities approach proposes that there are potential states of affairs in which our well-being would increase, but that we would not choose these states because they are not utility-maximising. This issue is quite separate from our inability to accurately estimate the utility gain from a state but rather argues that, even in experiencing the state, it would hold value that would not be expressed in our choices. To identify such situations is a necessary condition of goal #3. The ranking and valuation exercise described above would not capture this value. So how might it be done? I have a suggestion.
Goal #3 is really about identifying the intrinsic value of a capability that cannot be reflected in our utility-maximising choices. However, much of good economics involves compromise, and I believe we can compromise here, by presenting a choice that is distanced from utility. To do this, it’s necessary to exhaust an individual’s choices. Any ‘value’ judgements that can still be made about capabilities can then be captured. We can begin with some convenient assumptions:
- Willingness to pay experiments reflect the full instrumental value of capabilities (identifiable in the complete capability choice set)
- Willingness to pay experiments do not reflect the intrinsic value of unrealised capabilities
- Money has no intrinsic value
Having exhausted an individual’s preferences in relation to the complete capability choice set, we know the monetary difference in value between any two states. If an individual prefers state 2 to state 1, they will be demonstrably indifferent between state 2 and state 1 plus some monetary value. Yet there may still be intrinsic value in state 2 which makes it superior to state 1. For an individual who attaches no intrinsic value to money, their willingness-to-pay for a capability’s intrinsic value is necessarily zero. If individuals are subsequently asked to choose which state they think is ‘best’, this might bring us closer to satisfying the concerns of the anti-choice Senists.
The question posed should obviously not ask the individual to choose or state which they prefer, but rather it should request that the individual indicate which they think is ‘best’. If there is any credence in the capabilities argument, then it is possible – likely, even – that people will be more likely to select 2 over 1, even accounting for the necessary reimbursement. This would signal the ‘value’ of the capability for functioning A that is not reflected in choice. It is not possible to value A and B against anything but themselves. Individual’s values cannot be traded-off when these values have been stripped of their instrumental worth.
Carrying out the valuation from a societal perspective helps prevent the potential underdog problem identified by Sen, whereby individuals adapt their preferences within inferior states. The strength of different values could be based on the tendency of a population to display them. If differences exist – and we can exclude any bias – then this represents a ‘value’ that traditional welfarist approaches have failed to capture.
If current approaches [as of 2014] to the use of capabilities in health are aiming for goal #3, then they are comprehensively failing. Firstly, the valuation process uses best-worst scaling (BWS). It is hard to argue that this is not a preference elicitation technique. Given that the ICECAP descriptors relate specifically to the individual, it is clear that they consider existing in the states being compared and choose – in keeping with standard welfarist economics – the option that would maximise their utility. As currently used for the ICECAP, BWS does not exclude or control for the instrumental value reflected in the choices. Secondly, the ICECAP-A dimensions were defined in much the same way as other (more utilitarian) measures; by asking people what is important to them, what they like and dislike, and what they desire. It is hard to see how this process identifies capabilities with intrinsic value. Thirdly, the ICECAP state descriptors force individuals to consider desire fulfilment by asking whether they can ‘have’ various levels of capabilities. To have something – even if something relatively intangible – is to satisfy a desire for it. The ICECAP questions elicit the extent to which an individual has the ability to increase their utility through various aspects of their life.
Measures like the ICECAP need to decide what their purpose is. If their purpose is to capture wider benefits of health and social care beyond health alone, then they are valuable and useful in their current form. If they seek to estimate the extent to which capabilities – rather than functionings – affect a person’s utility, they could succeed with a little more work. If, however, their purpose is to capture the intrinsic value of capabilities that cannot be reflected in individuals’ choices, they have comprehensively failed in their design and operation.
My question is, where are we?