3.8: Happiness and the capability approach
In section 3.7.2, we encountered hedonist theories of wellbeing as an important subgroup of theories of wellbeing. The debates about hedonism and the happiness approach are closely related. Hedonism is the philosophical view that wellbeing can be captured by the balance of pleasures over pains. The core aspect is the exclusive focus on mental states, and on a person’s subjective assessments of their own mental state. In recent decades, this approach has been revitalised in ‘the happiness approach’, although empirical scholars prefer the term ‘subjective wellbeing’ (SWB). The happiness and SWB literatures have in recent years gone through a revival.
On the empirical front, significant progress has been made in the last few decades by an international network of economists and psychologists, such as Andrew Clark, Ed Diener, Ada Ferrer-i-Carbonell, Bruno Frey, Richard Layard, Andrew Oswald, David Schkade, Bernard van Praag and Ruut Veenhoven. 10 Many of these scholars have concluded that sufficient scientific progress has been made for public policies to focus on subjective wellbeing. The measures of subjective wellbeing have been tested and refined, and much is supposedly known about the determinants of happiness that the government can influence. The happiness and SWB approaches are strongly focussed on empirical analysis and policy design, and this is also, therefore, the main lens that will be used in the comparison with the capability approach, although we will very briefly discuss the comparison between the theoretical happiness approach and the capability approach.
From the perspective of the capability approach, the happiness approach raises three questions: First, what is the happiness approach, exactly? Second, what are its strengths and weaknesses? Third, what role can happiness play in the capability approach? 11
3.8.1 What is the happiness approach?
The happiness approach is based on the assumption that wellbeing (or quality of life) is constituted by the subjective experiences of a person, expressed in terms of utility, happiness, or satisfaction. Satisfaction can be expressed in terms of overall satisfaction with life , or satisfaction within particular domains, such as income, health, family relationships, labour, and so forth.
In the happiness approach, life satisfaction is understood as a concept that combines two components: how we normally feel in everyday life — the affective or ‘hedonistic’ component — and how we judge the degree to which our preferences and aspirations in life have been realised — the cognitive component. In order to find out how ‘happy’ a person is, respondents are asked, for example, to rate how satisfied they are with their life on a scale from 1 to 10 (to measure the cognitive component) and to report their mood at particular moments of the day, sometimes even with the aid of a buzzer set to go off at random times (to measure the affective component). In another method, the respondents are asked to imagine the worst possible life and to give that life a value of 0, to imagine the best possible life and give that a value of 10, and then to rate their own life on a scale from 0 to 10.
The view in the happiness literature is that overall life satisfaction should be adopted as the official ‘policy guide’, and the task of the government is to aim for the highest possible average level of life satisfaction (Hagerty et al. 2001; Layard 2011). For comparisons in the long term, Ruut Veenhoven also proposes to measure the quality of life based on ‘happy life expectancy’. This is an index obtained from multiplying life expectancy in a country with average overall life satisfaction (Veenhoven 1996).
Is the happiness approach, or the SWB approach, the best basis for thinking about wellbeing and the quality of life, especially against the background of policy design? The happiness approach certainly has a number of attractive features. Firstly, it puts the human being centre stage, rather than focusing on the means that human beings use to improve their quality of life. Hence the approach satisfies the core criterion from module A that means and ends should not be conceptually confused. Secondly, in considering the means to happiness, the subjective approach is not limited to material means, which is the major shortcoming of the dominant economic empirical methods. Income has only a limited (but not unimportant) role to play in generating happiness.
In conclusion, the happiness approach does have some significant strengths, but it also gives rise to some concerns. We will briefly discuss the main theoretical problem, and then look in more detail at the worries raised for empirical research and policy making.
3.8.2 The ontological objection
The theoretical worry is an ontological worry, that is, it asks what wellbeing is , and questions whether this can be captured by mental states only. This objection was forcefully made by Robert Nozick (1974, 42–45) when he introduced the ‘experience machine’ thought experiment. Nozick ask us to imagine that we are invited to be plugged into a machine, which would stimulate our brain and make us feel as if we were having a range of experiences that we could choose beforehand; all the time, we would be floating in a tank with electrodes attached to our brain. Would we choose such a life? Nozick claims we would not, and interestingly enough the arguments he gives to justify this claim refer to our functionings and capabilities. 12 According to Nozick, three things matter to us in addition to our experiences. First, we do not only want to have the experience of doing certain things (which we could have by sitting and taking drugs) but we also want to do certain things. Second, “we want to be a certain way, to be a certain sort of person” (Nozick 1974, 43). Third, we don’t want to limit our experiences to a man-made reality (the experience machine) but also to have the opportunity to be in contact with deeper significance. The insight from Nozick’s thought experiment is thus that a good life cannot be reduced to mental states, but must also contain some genuine activities and states of being. According to Nozick (1974, 43), “someone floating in a tank is an indeterminate blob”, not a human being to whom we can describe human wellbeing. There is thus more to human wellbeing then merely feeling happy. If decent labour, knowledge, appreciating art and culture, and intimate relationships are to be valued only, or even primarily, because of their contribution to overall or specific life satisfaction, then we could say this is a misrecognition of the contribution they make to how well our lives progress. Phenomenologically speaking, this is an implausible account of wellbeing.
3.8.3 Mental adaptation and social comparisons
How about the worries related to empirical research and policy making? The first worry at this level is raised by processes of mental adaptation and social comparisons. Our satisfaction is to some extent influenced by mental adaptation issues that emerge from comparisons with the situations of others. This can have problematic implications for public policies aiming at the highest happiness for the greatest number. Take the mental adaptation processes first. How do these emerge? First of all, there can be shocks in our lives that have a major effect on our wellbeing, such as immobility after an accident. People confronted with a major setback in health and mobility through such an impairment will first experience a strong deterioration of their subjective wellbeing, but after a while this effect will weaken. Obviously, this adaptation to circumstances is good, since a disabled person will not remain deeply unhappy for the rest of her life due to her limited abilities to move around without pain. However, the question is what this implies for policy. A utilitarian will say that the government has to limit itself to creating provisions such that a disabled person can return to an acceptable level of life satisfaction, taking into account the corresponding welfare costs for others. 13 A utilitarian would even say that there is no reason to invest in prevention if this is more expensive than rehabilitation. But one could also argue that a cost-sensitive policy has to try to reach an acceptable level of functioning for a disabled person, even if this makes little difference to her subjective judgement about her wellbeing after adapting to the accident. Subjective indicators focus automatically on the first goal, but this may imply that the quality aspects that relate to the things a person still can do after the accident remain out of sight.
Secondly, people can adapt to an objective disadvantage that is not caused by an external shock, but that shows a more stable pattern. This is the problem of ‘adaptive preferences’, which is particularly relevant for the happiness approach. 14 Amartya Sen has pointed out repeatedly that people living at the very bottom of the social ladder (such as ‘exploited labourers’ or ‘oppressed housewives’) adapt to their situation and come to suffer less intensely. Another example is the effect of racism. If a society becomes gradually less tolerant towards cultural minorities and increasingly accepts racist practices, then cultural minorities might get used to a racist social climate. Perhaps they will change their behaviour, in order to avoid contact with openly racist people. By changing their behaviour and mentally preparing for racist practices, it is possible that after a while the negative wellbeing effect of racism on minority groups will be partially wiped out. However, a policy that anticipates such adaptation processes is morally and politically problematic: racism should not be tolerated in society, even if it were not to have a significant impact on the subjective wellbeing of its victims. In order to judge that racism is not morally permissible and hence that policies should try to minimize racism, we don’t first need to investigate whether racism makes its victims less happy: that’s simply beside the point. Even if the victims of racism acted stoically and didn’t let racism affect their happiness levels, that wouldn’t make racism any less undesirable.
Another form of mental adaptation which is relevant for the government is the adjustment response to income changes. Subjective wellbeing judgements about income have been shown to adapt asymmetrically to income changes. Income increases go together with higher aspirations for the future, with only one third of the increase being reflected by improvements of subjective wellbeing (Frey and Stutzer 2002). Panel-analysis over a period of ten years shows that we adapt strongly to an increase in income, but much less so to a drop in income (Burchardt 2005). Thus, if people change positions in an income distribution which itself remains unchanged, then aggregate satisfaction of the population will decrease. The people who move up the ladder will be more satisfied for a short time, but quickly adapt to the new situation, whereas people who move downwards experience a larger drop in satisfaction — and this effect lasts longer as well. Tania Burchardt (2006) argues that due to similar phenomena of adaptation, people’s positions in the distribution of income, health and marital status should preferably remain immobile, according to utilitarianism. Clearly this is a policy conclusion that goes against the principle that people should receive equal opportunities, even if the effect of one person’s upward social mobility is not compensated by the effect of another person’s downward social mobility.
How serious are these problems of mental adaptation for the subjective approach? In part, the response to this question depends on our normative judgements about the counter-intuitive and sometimes perverse implications of a policy that single-mindedly aims to promote maximal average utility. It also depends, however, on the empirical question of how strong these mental adaptation processes are in reality. According to Veenhoven, overall life satisfaction is primarily determined by the affective component, and therefore it is much less vulnerable to the effects of mental adaptation than satisfaction in particular domains, which he judges to be much more vulnerable to adaptation. However, the work of Kahneman and Krueger (2006, 17–18) shows that mental adaptation processes are clearly present even when predominantly affective measures of overall happiness experiences are adopted.
There are additional concerns related to the subjective wellbeing findings in particular domains, such as income or education. Recall the literature reporting on the findings that subjective wellbeing is also strongly influenced by social comparisons with reference groups. In particular, the wellbeing effect of income, but also of education, is affected by the levels reached by members of the reference groups to which individuals compare their own situation. As a consequence, increases in income, or additional educational credentials, contribute less to satisfaction in these domains, the more income or educational progress is achieved within the reference groups. Apparently, these resources have a stronger positional component than other resources do, in particular leisure time, where the comparison effect appears to have a much weaker impact on wellbeing obtained from an additional unit of free time.
In conclusion, there seems to be little consensus in the subjective wellbeing literature on the question of whether, and to what extent, phenomena of mental adaptation and reference groups cause problems for the measurement of overall life satisfaction. However, all researchers do acknowledge that satisfaction in some domains is susceptible to these phenomena, and this may result in the counter-intuitive policy implications we mentioned earlier.
3.8.4 Comparing groups
The second worry about the happiness or SWB approach concerns the effect of group differences, which will be a problem if we need an account of wellbeing to compare wellbeing levels between groups. The subjective wellbeing approach focuses on the affective and cognitive responses of people to their lives overall, or in particular domains. If groups differ on average in their responses to a situation, then this may cause problems for policies, if those differences correlate with the objective circumstances that one would intuitively judge as important. There are two symmetric possibilities: (1) groups who are in the same objective situation have different levels of life satisfaction, or (2) groups with the same level of life satisfaction are in different situations, whereby it is clear that one situation is worse than the other independently of subjective wellbeing.
Research has indeed shown that the average level of life satisfaction between demographic groups differs systematically. In other words, if we control for the relevant factors, then some groups are significantly less satisfied with their lives than others. For example, recent Australian research (Cummins et al. 2003) shows that women report a higher level of overall life satisfaction than men, after taking a number of control-variables into account. 15 The researchers cannot pinpoint the exact causes of this finding, but they do not exclude the possibility that women are ‘constitutionally’ more satisfied than men. This may have a biological explanation, but it may also be the consequence of processes of adaptation that men and women experience differently over their lifetimes.
If the aim of the account of wellbeing is to inform public policy, then the question is how government should deal with these findings. From a utilitarian perspective, it would be efficient to develop a policy that is advantageous to men. For example, if due to unemployment men experience a larger drop in happiness than women, as reported by Frey and Stutzer (2002, 419), then a policy that gives men priority on the labour market will minimise the average wellbeing damage in terms of happiness. But the fact that one demographic group (women, the worst off, the elderly, and so forth) are made less unhappy due to a certain event than other groups can cause perverse policy implications if life satisfaction is declared to be the guideline for policies. Fundamental political principles such as non-discrimination and equality of opportunities for all citizens are thereby put into jeopardy. This would also be true in the symmetric case where the average level of life satisfaction of discriminated or marginalised groups does not differ significantly from the average level of a group that is not faced with these disadvantages. I do not want to claim here that the subjective wellbeing approach will always lead to such injustices. But I do think that a central focus on subjective wellbeing will make policies less sensitive to signalling and combating these injustices. Hence Burchardt (2005, 94) is right in pointing out that “satisfaction — the best proxy we have for the concept of utility — is unsuitable for assessing current wellbeing, justice or equality”.
3.8.5 Macro analysis
A third worry concerns the applicability of the subjective wellbeing approach at the national or regional levels of policy making. One may agree that the happiness approach can be very helpful when it can offer persons with low affect (negative moods and feelings) concrete strategies to change that, such as engaging in mindfulness training and practice. Yet what about policy making? Are the happiness indicators sufficiently refined and sensitive for policy at lower levels of aggregation than the level of a country? In their discussion of the criteria that an index of the quality of life should meet, Hagerty and his co-authors (2001, 2) include the criterion that the index must help policy makers to develop and evaluate policies at all levels of aggregation. Thus, the index should not only be useful for the national government, but also for governments in cities, communities, and regions. As Robert van der Veen and I argued in earlier work, overall life satisfaction does not satisfactorily meet this criterion (Robeyns and Van der Veen 2007); it is too crude for these purposes. It is even less suitable for the evaluation of specific policy interventions (Cummins et al. 2003). The effect of one policy measure such as improved child care facilities will be reflected hardly or not at all in reported overall life satisfaction, even if such policies have significant effects on the real opportunities of parents to organise their lives as they think best — hence on their capabilities. Overall ‘happy life expectancy’ is, by contrast, well-suited for comparing the effects of fundamental political and economic institutions on subjective wellbeing. This emerges clearly from the work of Veenhoven (1996), which concentrates on studies whereby the unit of analysis is the country. In other words, Veenhoven mainly uses happy life expectancy as an indicator for macro-analysis. The variables that emerge as the determinants of happy life expectancy are therefore typically system variables such as the degree of political freedom, or the presence of rule of law. But the quality of life in a micro-situation (say, living in a particular community or neighbourhood) is also influenced by many other variables.
3.8.6 The place of happiness in the capability approach
The previous sections argued that happiness can’t be taken to represent a person’s wellbeing for many purposes, including policy purposes. Yet it would also be deeply counter-intuitive to say that happiness doesn’t matter at all. It may be the right concept of wellbeing for other aims. How, then, can happiness be given a proper place within the capability approach?
The first possibility is to see happiness, or some more specific capabilities that are closely related to the affective component of subjective wellbeing, as one important dimension to be selected. In fact, Amartya Sen has for many years argued that we could take ‘feeling happy’ as one of the functionings to be selected. For example, Sen (2008, 26) wrote:
happiness, however, is extremely important, since being happy is a momentous achievement in itself. Happiness cannot be the only thing that we have reason to value, nor the only metric for measuring other things that we value, but on its own, happiness is an important human functioning. The capability to be happy is, similarly, a major aspect of the freedom that we have good reason to treasure. The perspective of happiness illuminates one critically important element of human living.
Indicators of happiness are already included, for example, by incorporating dimensions of mental health into capability applications. Certain specific functionings, which make up an overall ‘mental health’ functioning, already contain such affective items: whether, over the last week, the respondent has felt down or worthless, for example, or whether one is free from worry.
Second, we can try to capture the cognitive aspect of happiness, that is, a person’s satisfaction with her capability set, or with particular options from that set. We can then compare that person’s level of satisfaction with her capabilities, which should allow us to compare the objective situation with a person’s satisfaction with that situation. For example, the absence of criminality is a valuable functioning. 16 But there is a long-standing finding in criminology that there are discrepancies between the objective incidence of being safe versus the subjective feeling of being in danger of becoming a victim of crime. For example, even if the incidence and impact of criminality goes down, it can still be the case that the population is more worried about crime than before and feels less safe. The discrepancy between objective outcomes and subjective perception is instructive here; it may imply, for example, that the government should communicate more effectively about its success in reducing crime, so as to make the subjective perception more in line with the objective reality, or it should work directly on factors that impact on the subjective experience of safety feelings.
Third, capability scholars would, of course, hope that an enlarging of people’s functionings and capabilities would, as a further effect, increase their feelings of happiness and satisfaction, and serve as a (sometimes rough) indicator of people’s satisfaction with their functionings and capabilities (Sen 2008, 26–27). Not all functionings will lead to people becoming happier, yet their lives may still be better: more flourishing, or more meaningful, or with a higher quality of life, or with a greater degree of freedom that could be realised. An interesting example is the case of writing a PhD dissertation. Very few PhD students would say that this is what makes them happy in the sense outlined above. So why, then, do so many graduates want to earn a doctoral degree, and give some of the best years of their lives to what often becomes a stressful time? It is hard to understand this, if one doesn’t take into account the meaning it brings to their lives. The capability approach can capture this — taking on a difficult and challenging project such as writing a PhD dissertation can plausibly be conceptualised as a general functioning (consisting of a set of more specific functionings) that we may want to include in our capability analyses, including in our capabilitarian theories of wellbeing for public policies.
I would thus defend the position that various roles for happiness are potentially possible within capability theories, and that it depends on the exact purpose and scope of the capability theory or application, as well as the aim that wellbeing plays in that capability theory, what the best role (if any) for both the affective and cognitive aspects of happiness would be.
11 This section draws on, yet also modifies and expands, the analysis presented in Robeyns and Van der Veen (2007, 33–42).
12 Obviously, Nozick didn’t use the terminology of the capability approach, but his account of what is valuable in life could nevertheless be seen as capabilitarian.
13 A more fine-grained analysis than that presented in this section would need to make the distinction between act-utilitarianism and rule-utilitarianism, and ask whether both are vulnerable to these critiques to the same extent. I am assuming here that the critiques apply to both types of utilitarianism to such a degree that it leads to worrying consequences.
14 Yet the phenomenon of ‘adaptive preferences’ can also potentially create problems for some capability theories, as we will analyse in section 3.9.
15 A similar strong and significant gender effect has been found by Eriksson, Rice and Goodin (2007).
16 Perhaps criminals would disagree.