Capabilities as menus: A non-welfarist basis for QALY evaluation

by John Q on March 8, 2014

I’m very grateful to Ingrid for setting up this discussion of capabilities. I enjoy the general discussions of social and political issues we have here at CT, but this is one of many venues for such discussion (among the best, I think, but I would say that). What’s truly unique for me is the opportunity to discuss the issues raised by my own academic work in an environment that is totally different from those offered by the economics profession.

As has already been mentioned, most of the discussion of capabilities has concerned poor/developing countries. Moreover, most of it has been qualitative rather than quantitative. One consequence is that, although the idea of capabilities has been around for a while now, its impact on the policy process in developed countries has been modest at best.

My own work on capabilities, represented by an article[1] published last year in the Journal of Health Economics has also had a modest impact, but for very different reasons. While not strictly quantitative, it’s mathematical, more so than the average reader of JHE tends to be comfortable with, and its direct relevance to policy is limited by the fact that we are, at least to start with, not addressing distributional issues.

The main objective is to explore the idea that capabilities can provide a basis for allocating health care resources based on the QALY (Quality-Adjusted Life Year) measure. in previous work, we looked at the “welfarist” idea that policy should be based on maximizing lifetime expected utility. It turns out that, considered purely as a technical problem, this can’t be done, except in very special cases. The appeal of capabilities is that they provide a non-welfarist (or at least ‘extra-welfarist’ in that it is more than a simple expected utility maximization) rationale for policies involving scarce resources like health care.

Our idea comes from the economic literature on menus, which began with work by Kreps and others. The idea here is that people value having a set of choices open to them, because they don’t know what their future preferences/needs will be. This seems very close to capabilities, if we think of capabilities as sets of possible functionings. In the health context, your health status is not an end in itself, but the capability of moving about, looking after yourself, watching sunsets and so on. Similarly, money is not valued for itself, but for the consumption choices it makes possible. So, we can regard a list of wealth and health characteristics both as a single object (a capability vector) and as the source of a set of possible functionings, which may be valued more or less highly.

This is the only point in the analysis where the math is more than routine algebra. Even though values (more precisely, ordinal rankings) are derived from sets of functionings, which are difficult to handle mathematically, we can associate them with capability vectors, which are mathematically similar to the consumption bundles economists handle all the time. There are trade-offs between improvements in different components of the capability vector, and these trade-offs can be represented by ratios, that work very like prices. In particular, if we scale all improvements in health capability against improvements in healthy lifetime, we get a Quality-Adjusted Life Years measure and, at least in principle, a trade-off between QALYs and (lifetime) wealth.

The way I think about what we’ve done here is to provide a basis for an objective notion of capabilities, which all reasonable people would evaluate more or less similarly, even while allowing for lots of individual variation in preferences, life-circumstances and so on. However, while there is a fair bit of room for variation, it’s not unlimited. For example, some participants in the health care policy discussion (mostly from the medical side) assume (implicitly or explicitly) that long life and good health are the supreme goods, and that everything else is secondary. That’s clearly not consistent with a view that places positive weight on the capability of choosing various mixes of consumption goods, education and so on.

A more serious problem is that we don’t discuss distributional issues. The analysis here works for a society in which everyone starts out with equal capabilities and chooses how to use them. We haven’t yet dealt with the case where capabilities are unequal, either because of entrenched economic inequality or because of inherent differences in health status. That problem becomes more difficult if people experience good or bad luck over their lifetimes, and there is no workable social mechanism for insurance.

In summary, this is really a very preliminary approach to making the capability idea workable. Still, I was excited to work on it, and would really like to see others pick up on it. So, I’m very happy to be able to present here at CT, and hoping it will generate some discussion

fn1. It’s joint work with Han Bleichrodt of Erasmus, with whom I’ve worked on health economics regularly over the last 20 years or so. The interpretation here is my own, though,



Brian 03.08.14 at 11:59 pm


This is an interesting approach, but I think it misappropriates the idea of capability/ies put forward in previous work. Take these two claims you make:

1. “The idea here is that people value having a set of choices open to them, because they don’t know what their future preferences/needs will be. This seems very close to capabilities…”

2. “The analysis here works for a society in which everyone starts out with equal capabilities and chooses how to use them. ”

It seems that all you have done is replace “choice” with “capabilities”, and suggest that people maximize multiple outcomes simultaneously. While this makes some sense on the face of it, I think the value of the capabilities approach is that it has shifted attention from observed outcomes (which economists often assume reflect ‘rational’ choices) toward the social, economic, and political resources (“entitlements”) that define the opportunity set people and households face. In short, it seeks to critique the assumption that your study relies upon (#2 above).

While I appreciate that you’ve drawn upon this concept to broader the perspective of your study, I think it’s problematic to claim you’re offering an “objective notion of capabilities” – economistic, yes, but impoverished conceptually. Not to mention, much of the work on capabilities has been empirical and in some cases quantitative.


Val 03.09.14 at 12:15 pm

(hi this is the same Val who occasionally comments particularly from a feminist viewpoint at johnquiggin, though using my Monash email in this case)

I´m interested in this issue though a bit reluctant to display my ignorance, even though Ingrid suggests the discussion is open to people who aren’t experts in CA – as I haven’t read as much Sen et al as I would like. However do have a couple of comments or questions.
Practically, as someone who has worked in health policy, I wouldn´t find your linked article useful, and it´s not particularly because of the maths, but because it is such specialised economics speak. Do you think that such articles might be more useful if they were more accessible to non-economists, and particularly if they included some real life examples of what they would mean in practice? Perhaps more of the detailed maths and economics speak could be in appendices?
While acknowledging my limited understanding, it still seems to me that your approach is, in spite of some disclaimers, intellectually framed primarily in terms of individuals who make choices. This seems to be a limitation, since as Ingrid says below, CA is about “social arrangements”, as well as, or as much as, individual choices. I cant see how this is addressed in your discussion.
It´s true that in health promotion we could still use QALYs to justify or promote spending in our area (eg why walkable environments might be a better investment than lap band surgery or something), but normally I would just talk about years of life lived or years of life lived without disability. I mean I guess we are implicitly making some choices there (its better to be able to walk streets safely rather than being able to drive faster than 40k on those same streets or whatever), but because freedom of choice isn’t inherently set up as a value in those measures, we don’t have to get into that debate in that context . However once you start including freedom of choice as a value I think you get into all sorts of difficulties. Free market advocates (and this is very political because it may become enshrined under the TPP) say that the freedom to choose whether to be healthy or not is more important than, eg, plain paper packaging of cigarettes, and I cant see how your approach deals with those practical issues.
Anyway sorry this is turning into a bit of a rave, but I´m heading more and more towards the idea that any approach that uses the individual as its basic conceptual building block just doesn’t work. Not that it isn’t subjectively meaningful, but it´s too limited.


Jameson Quinn 03.09.14 at 1:08 pm

Why are values ordinal and not cardinal? In my work on voting systems, I’ve seen how that apparently innocuous choice can doom you to merely picking which horn of the Arrow’s-theorem dilemma you want to die on.

…Quickly skimming your paper, this choice seems unforced. You seem to be considering only a single decision agent, so collapsing to ordinality is immaterial. Still, were someone ever to expand your theory to multiple deciders (probably an inevitable step if they are to deal with distributional issues), I they would be ill-advised to maintain ordinality.


John Quiggin 03.09.14 at 9:21 pm

Thanks for this suggestion, Jameson. Cardinality requires a fair bit more information (about intensity of preference, or choices under uncertainty) and may bring in things we don’t want (for example, pushing us back to welfarism). So, given that it’s immaterial in the single-agent case, it seems best to leave it out.

Things may change when we come to consider distribution. However, I would start from a presumption that justice requires equalization of capabilities, so far as this is reasonably possible. With this starting point, we may not need any cardinality assumptions.


John Quiggin 03.09.14 at 9:27 pm

“It´s true that in health promotion we could still use QALYs to justify or promote spending in our area (eg why walkable environments might be a better investment than lap band surgery or something), but normally I would just talk about years of life lived or years of life lived without disability.”

Unless you are treating years lived with disability as valueless, I think these are just different names for the same concept, or very similar ones.


John Quiggin 03.09.14 at 9:28 pm


“It seems that all you have done is replace “choice” with “capabilities”, and suggest that people maximize multiple outcomes simultaneously. While this makes some sense on the face of it, I think the value of the capabilities approach is that it has shifted attention from observed outcomes (which economists often assume reflect ‘rational’ choices) toward the social, economic, and political resources (“entitlements”) that define the opportunity set people and households face. In short, it seeks to critique the assumption that your study relies upon (#2 above). ”

I think you’ve missed the point here. Capabilities, as we’ve defined them, are exactly what you call “entitlements”


Ed Herdman 03.09.14 at 11:58 pm

Do economists actually call “social, economic, and political resources” “entitlements?” That would be new on me.

Very interested to see this development, and coming as close as possible to a quantitative method is very welcome. If this succeeds, what chances does it have of supplanting various “easy” methods like Cost-Benefit Analysis? (Which is not to say CBA actually is easy, but people and organizations often use it often in a minimally involved way.)


John Quiggin 03.10.14 at 3:31 am

Ed, I quoted the term “entitlements” from Brian @1, as equivalent to “capabilities”. But economists in general aren’t using capabilities, just that capability fans are over-represented at CT.

CBA is highly problematic in the health context which I hope will make some room for a capability model


Val 03.10.14 at 7:57 am

(Thanks JQ I can see all comments now, must be something to do with order of refreshing or something)

my previous comment was a bit unclear – YLLs are a metric tied to individuals, but they are not tied to individual subjectivities in the sense that ´quality´ is, and ´choice´certainly is. It´s the notion of the choosing individual (central to much economic theory as I understand it) that is the problem, because both individual subjectivities and the choices they make are socially constructed. I think if you try to abstract them from this context it becomes a problem and your approach still seems to be doing that. I think my comment is similar to Brian´s in fact.


John Quiggin 03.10.14 at 10:36 am

And my response is the same. It’s functionings, not capabilities that are the objects of individual choice here. The valuation of capabilities, at least as we’ve done is the same for everyone. Of course, the valuation arises from the fact that capabilities allow you to make choices. And, of course, those choices are made in a social context – we are social beings.


Val 03.10.14 at 5:53 pm

Yeah but I’m a health policy person and I can’t see how this works – so I guess I go back to my first point – how would it help in the junk food debate, or alcohol/smoking debates – where the argument us that an individual’s right to ‘choose’ to stuff up their health outweighs my right to ‘choose’ as a health policy person to stop this stuff being actively promoted to people?

Just as an example, I once prepared a brief about why we should restrict junk food advertising on tv – very well reasoned etc, though I can’t remember how far I went into economic arguments – and it went nowhere, because certain tv execs met with a certain premier. So how would your work help with that kind of thing?


ChrisB 03.10.14 at 11:45 pm

The problem I’ve encountered in QALYs previously is that if one gives years lived with disabilities a lesser weighting than other years that does automatically discount the interests of people with severe disabilities. If, say, one finds that one could either spend an amount on adding five years to the lives of 100 people with Down syndrome or on adding the same amount to the lives of 100 people without, then the normals get the gelt because they have more QALYs. It’s known in the literature as double jeopardy, and Singer, for one, has argued that it’s perfectly reasonable.
I wrote an article in opposition to this view for (Singer’s) Monash Journal of Ethics some years ago;
“a statistical and QALY-based approach does not represent the elimination of irrelevant information to allow the underlying regularities to emerge more clearly; it represents the elimination of the only information there is or can be, and a destruction of material that should correctly be taken into account in decision-making. QALYs do not represent a distillation of the raw data, still less an enrichment. They represent an impoverishment. This can be readily seen for the individual; if I am offered a choice of two 20-QALY treatments, I can reasonably wish to undo the mathematical transformation in order to know which of two treatments with identical QALYs has fewer years but less pain. A policymaker can, and should, ask much the same thing. The QALYs number cannot hold the information that it is necessary it should if the maximum satisfaction of preferences is to be our goal.
We must give up the dream, stretching back to Leibniz, of being able to lift our disputes out of the realm of argument into the realm of logical, arithmetical, demonstrable, unarguable, eternal truth.”

I’ve put it on the blog at


John Quiggin 03.11.14 at 12:43 am

@ChrisB On your first point, I think this is valid, but hopefully won’t arise as an objection to the egalitarian capability model we are working on.

Your second point is exactly opposite to those of Val and Brian, who want even less of a role for preferences than is allowed by the standard QALY approach. And it seems to me to be somewhat incoherent. If, on average, people want to trade off pain and longevity in some particular way, and this is embodied in the QALY formula, on what basis does a policymaker over-ride this in the allocation of health resources between, say, palliative care facilities and hospitals? Treatment of individual cases is a different matter, but no one (AFAIK) suggests using QALYs as the basis for individual medical advice.


ChrisB 03.11.14 at 2:46 am

Read the whole thing…


John Quiggin 03.11.14 at 5:24 am

The “whole thing” just seems to say that the problem is too hard, and that it should be left to political/market processes. In practice, that usually means that politicians and doctors make the decisions in ways that let them pretend that no decisions are actually being made.

Except for the further layer of complication that a lot of politicians, doctors, insurance companies etc are actually eager to get the problem off their plate and use an “objective” criterion like QALYs.


John Quiggin 03.11.14 at 5:33 am

Val, you can certainly use the general approach of our paper in a context where some functionings (such as drug use) are negatively valued, so we want the capability to preclude them. There’s actually quite a big literature on this, using the “menu” concept of the title, but I’ve never looked closely at it. My main interest in health policy is the allocation of resources to improve health, and the paper reflects that.

But, even if the model was expanded to allow for negatively valued functionings, I doubt it would have much effect on the kinds of decisions that are made in meetings between politicians and media moguls. That’s the negative side of the political/market processes discussed in previous comments.


Val 03.11.14 at 6:29 am

The point about the media moguls and advertising of course is that they represent one of the forces that shapes people’s “choices”.

Our law privileges the right of corporations to promote unhealthy food (eg McDonalds in Tecoma case) above the rights of residents or health policy makers to resist. This is done in the name of “consumer choice”.

So far as I understand the theory of capabilities, this is relevant because it affects both capabilities and functioning.


Brian 03.11.14 at 11:09 am

Re: #6

Thanks for the response. I think that my comment appeared to miss the point precisely because you’ve equated entitlements with capabilities, which I think is problematic. Individuals might have the same capabilities – say, the ability to acquire some subsistence level of food – but this may be a function of a different set of entitlements (e.g., their savings account, government food assistance, NGO aid, etc.).

I think that part of the usefulness of the capabilities approach is interrogating the multiple entitlements (e.g., is NGO aid really an entitlement – can people really make claims to it?) that underlie those capabilities. These underlying entitlements are qualitatively and quantitatively heterogeneous within and across populations. Given this, it seems like a stretch to even assume that capabilities are homogeneous.

One other, not related point that might stoke some discussion: the focus on entitlements and capabilities approach is completely at odds with the notion of “revealed preferences” among economists. Has this been articulated in the academic literature?



Sam Clark 03.11.14 at 12:38 pm

I’m way outside my competence here (and I admit to having only skimmed the paper), but there are a couple of things about this that look odd from my outsider’s perspective:

1) ‘Welfarism’ is defined as the view that ‘policy should aim at maximizing a social welfare function, the only arguments of which are the lifetime utility profiles of the individual members of society’ (128). But that pulls together several different ideas: consequentialism, the view that rational action is maximizing some result value (compare e.g. the view that rational action is publically meaningful action); value monism, the view that the value result to be maximized is measurable on a single scale – in this case the utility scale (compare value pluralism); and (what I’d call) welfarism, the view that value is only instantiated in the lives of individual people (compare e.g. the intrinsic value of natural beauty). This may just be a terminological difference between disciplines, but even if so, it’s useful – to me at least – to notice it.

2) There’s something strange, on the face of it, about basing what looks like a way to do cost/benefit analysis on the CA: I took it that the CA is value pluralist in the sense that it claims that there are a number of distinct, incommensurable values, and therefore resistant to any such move to arithmetic measurements and trade-offs.


John Quiggin 03.11.14 at 8:50 pm

Brian, I didn’t mean to say that entitlements are *exactly* the same as capabilities. As you say, the same capabilities may arise from different entitlements. And of course people with the same capabilities may/will have very different functionings

My idea of the capability approach is that it’s capabilities rather than entitlements or functionings that we should be concerned about equalizing.

As regards the conflict with revealed preference, that’s implicit (to me) in the “non-welfarist” part of the title. But I don’t think it’s been explored much in the economic literature on menus and choice. There would be some nice (in the academic econ sense, which roughly means “lots of clever and tricky math”) papers to be written there.


John Quiggin 03.11.14 at 9:01 pm

Sam, it’s always striking to me how differently terms are used in different disciplines, even relatively closely related ones.

What you call “welfarism”, I would call something like “methodological individualism”. This is so much part of the core assumptions in mainstream econ that it’s almost never spelt out, and similarly with consequentialism and value monism. The bit of welfarism we are challenging, which isn’t in your unpacking, is the “revealed preference” idea discussed in the previous comment, that individual good arises from preference satisfaction.

As regards incommensurability, it would be good to ask Ingrid if she sees this as a feature of CA, and if so whether it is essential or optional. For myself, I want to be able to answer questions like “should we fund a palliative care centre or a new oncology ward”, so a response like “this involves incommensurable values that can’t be subjected to arithmetic values and trade-offs” is entirely unhelpful. If you are going to say that, it would be better to go the whole hog with Chris B and say “leave it to politics and the market to sort this out”, in which case discussion of capabilities just clouds the issue.


Brian 03.12.14 at 10:29 am

Agreed Dr Q., thanks for your comments. Interesting stuff!


Sam Clark 03.12.14 at 4:44 pm

Thanks John – yes, terminological differences can get pretty startling (and just to add even further to them, I’d take methodological individualism to be a claim about explanation: social action must ultimately be explained as the aggregate of actions of individuals, with no appeal to historical forces, classes, etc. except as shorthand. But welfarism[Sam] is a claim about how actions should be justified, not about how they should be explained).

I also defer to Ingrid on whether (or how necessarily) CA is pluralist, but I do want to say: that a decision can’t be made by weighing quantities doesn’t mean that it can’t be made reasonably. I would want questions about public and incommensurable goods to be made through democratic deliberation (is that ‘leave it to politics’? Not politics-as-usual, anyway). The market is another way of asserting value monism, I think: price is taken as commensurating, wrongly in the case of goods which shouldn’t be commodified.


Val 03.12.14 at 6:37 pm

ChrisB @12
I looked at your blog and I think the points you are making are really important, although your post is awfully long (maybe you should do an executive summary?) – I have to admit I started skimming after a while.
I believe that mathematical analysis can be meaningfully applied to almost everything, but in order to do that you have to have really thoughtful qualitative and conceptual analysis to ensure you are measuring something meaningful, and that so often seems to be left out.

Comments on this entry are closed.