PhD presentation on happiness

These are two drafts of transcripts for presentations of my thesis theme to a group of other PhD students. They were both too long so I had to cut them down, but I thought I'd post them here for anyone interested and so I can refer back to them when necessary. 

My PhD is concerned with modelling happiness. My undergraduate degree was in philosophy, and I am principally motivated by a desire to better integrate the philosophical literature on this topic into the existing research stream, which is mostly prosecuted by psychologists and economists. To date, academic happiness research, which began in the early 70s, has been principally driven by empirics rather than theory. This is understandable, because very little of what we mean by ‘happiness’ can be observed and measured. This is precisely why economics decided to focus on utility instead and to utilise revealed preferences and rational actor theory rather than something like the hedonic psychology model of behaviour. However, I am little bit troubled by this approach as I think it sometimes risks overlooking aspects of happiness that aren’t measurable. My hope is to develop a model of happiness that is more integrative and expansive and then test it empirically. Now as I just said, many of the hypotheses that I am interested in involve unobservable factors like ‘meaningfulness’ and living ethically. I am currently thinking of work arounds that will allow me to at least make a start on empirical stuff, but today I thought I would talk more about the theory itself and leave empirics to a future session.

Let’s just start with the model. It looks like the following:

Ui = log(WELLBEINGi) + HAPPINESSi + MEANINGi

I have a few quick things to point out. The dependent variable is utility, measured cardinally rather than one a 1-10 scale; I will return to this in a future session. The independent variables are all words that have been used in the literature to mean utility. This is fitting, because precisely what I am trying to do is integrate these different aspects or sources of utility. Each of the independent variables refers to a cluster of variables—this is a multi-level model in that sense—so wellbeing includes things like income and companionship while meaning includes things like church attendance and career advancement. Each of the three clusters corresponds very roughly to one of the three literature groups: wellbeing to development economics, happiness to psychology and meaning to philosophy (and psychiatry to some extent).

Wellbeing is predominantly about the absence of deprivation. I use the term wellbeing even though psychologists mean something a bit different when they use wellbeing because I can think of no better term to describe the ‘lack’ of negative factors. You can think about it in terms of Maslow’s hierarchy of needs or in terms of development economics, or indeed both. Some key components are income, health, education, companionship and/or reproduction and capabilities in the sense that Amartya Sen uses in his definition of development as freedom or agency.

Wellbeing is in log form because we have a lot of evidence that these items, especially income, suffer from diminishing returns to happiness. However, a seminal paper by Wolfers and Stevenson showed that income and happiness have a very tight linear fit on the log scale. Theoretically, we can understand this as reflecting the fact that having all the prerequisites for agency and self-actualisation is only half the puzzle—the other half is actually being able to self-actualise. Getting more ability to self-actualise doesn’t have much effect once you’ve already secured the ability.

Let me immediately pause here. I don’t want to give the impression that I have an agenda to show that self-actualisation is what matters. I used to think that but I abandoned that hypothesis years ago. It is quite possible to be supremely happy enjoying, for example, a life full of leisure, or video games for that matter. The model allows for that, but it also captures things like identity and self-actualisation, which aren’t really allowed for in hedonic models, set-point theory or the naïve income-driven models of happiness beloved of neoclassical development economics (which is not to say that I am not fond of those models).

Happiness is an exogenous shock variable modelled as an auto-regressive process. It is not captured by any sub-variables because the potential sources of these shocks are too broad and vary widely depending on what time scale you are using. For example, at the hourly time scale things like missing the bus or finding a dollar matter, while at the annual time scale we are more interested in things like divorce and promotions. This part of the literature draws heavily on studies in psychology concerning adaptation, homeostasis and set point theory. The crux of this research is that, arguably, human neurochemistry is characterised by homeostatic theory, which suggests that we have a natural base line happiness of say, 7/10 on a 1-10 scale. While we may experience divergences from this set-point owing to events like orgasm or the death of a spouse, over time we always converge back to the set-point.

The final cluster is meaning. This is the part that we don’t see much in the existing literature. Yet In the religious and philosophical literature we see a strong emphasis on the idea that meaning is critical for sustained happiness, which is part of why I am concerned with it: meaning might be a channel for overcoming homeostasis or permanently elevating the set-point. For example, one of the core tenets of Buddhism is that all life is suffering because, in an existential sense, because we can never sate our desires, and that asceticism is the escape hatch from this situation. Schopenhauer formalised this philosophy in the 19^th century. His pupil, Nietzsche, said that ‘man needs a reason for his suffering’ and began the development of a theory of how this reason can be created by the individual rather than received from the universe, which is the fundamental idea of the monotheistic religions of Islam, Judaism and Christianity, which find there apex philosophical representation in the work of Soren Kierkegaard. Nietzsche’s theory was further developed by the existentialists in the first half of the 20^th Century, especially Simone De Beauvoir.  

To summarise this theory very briefly, In order to be robustly and sustainably happy at a high level, the individual needs to answer three questions: who am I, what should I do and what is right? These are the questions of identity, motivation and ethics. Along with ‘where am I’ they are collectively referred to as the questions of being. Once the individual has come up with authentic answers, which means that they are their own answers rather than answers received from society, peers or religion, the individual then needs to go out and affirm those authentic values. In so doing, they will reveal to themselves, in and through their actions, to be the person they believe themselves to be and that they want to be. This is known as the coincidence of being, and is regarding in this philosophy as the end goal of the human condition and the means of its transcendence. I can’t go into any more detail here, but I do want to mention that inauthentic actualisation is referred to as bad faith, and that one can be authentically religious even though religion is an externally received value set, as evinced perhaps most clearly by Martin Luther.

My first paper elaborates this theory in much greater depth; my other two papers will try to test elements of it empirically. 

SECOND DRAFT

My PhD is concerned with modelling happiness. I have three principal motivations.

The first is to better incorporate insights from philosophy and, to a lesser extent, psychiatry, into the theoretical models of happiness we currently use.

To date, academic research into happiness has been principally driven by economists and psychologists, starting in the 1970s, and by empirics rather than theory. You take a small question, like what happens to happiness on weekends, and then check it against some data.

For two reasons, this approach is very appropriate as compared to what came before.

First, theory doesn’t become knowledge unless you test it and so we need to focus on things we can test—the old theories of happiness aren’t readily amenable to that procedure.  

This leads to the second reason, which is that happiness is extremely difficult to measure and so you are limited in how grand a hypothesis you can test.

However, I worry that at times this approach has meant that we conflate the happiness we can measure for the totality of what comprises happiness. I will give a few examples.

First, when we ask people ‘how happy are you right now?’ we are principally measuring subjective wellbeing, which is, speaking very simplistically, a mood state only partially informed by general life circumstances. If someone finds a dollar in the minutes before they are asked the question they report higher levels of happiness than people who find nothing.

Psychologists, notably the hedonic school, are fascinated by these phenomena.

However, we can intuitively grasp the idea that someone who is poor, lonely and beset by failure who finds a dollar will be less happy than someone who is rich, popular and successful who does not find a dollar. This intuition suggests another element of happiness that we might call life satisfaction.

Economics is, I would say, generally more interested in life satisfaction; hence its inclination towards objective measures of development like income and capabilities.  

Finally, you have probably come across, in yourself, in literature or among your peers, the phenomenon of someone who has all the prerequisites for happiness yet appears stuck in a kind of existential rut.

They might be struggling to answer the questions of being. These are: ‘who am I’ (the identity question), ‘what should I do’ (the question of purpose) and ‘what is right’ (the ethical question)? There is also ‘where am I’ (the question of world). You might have heard of Robin Sharma, of Oprah Winfrey fame. He was a billionaire hedge fund manager but famously sold his Ferrari and became a monk in order to find ‘a more meaningful existence’.

Philosophy and theology made enormous theoretical inroads into this group of issues in the late 19^th and early 20^th centuries in the works of people like Kierkegaard, Nietzsche and Simone de Beauvoir, but the happiness literature seems stuck on Aristotle 2000 years earlier, and Bentham, who was hardly a philosopher of the good life.

Speaking very generally, it is these three aspects of happiness that I would like to integrate into one model. I will present it next time.

So that covers my first motivation—to better integrate philosophy and to create a model that captures all the different sources of utility. I say utility because we typically think of utility as a cardinal concept while happiness is almost invariably measured on an ordinal scale, usually 1-10.

This brings me to my second motivation, which is to investigate the possibility of re-scaling.

One of the most puzzling findings in the happiness literature is that a goat-herder in the foothills of Pakistan is as happy (7/10) as a young middle class Briton in London. The existing explanations for this phenomenon are reference group effects and adaptation to a set-point.

Reference group effects owe their origins to the fact that humans are a social species and we typically compare our circumstances to those around us. Basically, if I am rich my world standards but poor by the standards of my immediate peers I will feel poor and this will make me sad.

The goat herder in Pakistan does not have the same reference group as the young Englishman. He is rich by his local standards while the boy is poor.

Adaptation refers to the fact that we ‘get used to’ things, or at least we seem to.

One of the most famous papers in this literature is a study by Brickman and Coates in 1971 of lottery winners and paraplegics.  They found that these individuals initially experienced spikes and dips in their happiness immediately after the wins and accidents. However, after two years their happiness had converged back to 7/10. This is adaptation and 7/10 is the set-point.

There is some evidence to suggest that there are some things we don’t completely adapt to. They include marriage and the death of a spouse.

One aspect of adaptation that was mentioned in the early days of that literature but seems to have recently fallen by the wayside, perhaps because of measurement issues, is that rather than adaptation, people are actually just changing their scales—7/10 no longer means what it used to mean.

Let me explain with a diagram:

We may see set-point phenomena but that doesn’t mean the set point hasn’t shifted up or down.

Critically, if we think in terms of cardinal utility rather than ordinal happiness we might get a clearer picture, at least a clearer theoretical picture, of these issues. This is because ordinal happiness essentially imposes a set point rather than allowing for, say, an Auto-regressive process with structural breaks.

As I mentioned, this phenomena is extremely difficult to identify empirically, but I have some leads that I will share in the future.

My final motivation is to better account for preferences in happiness modelling. I am a little bit surprised by how little economists have discussed this dimension.

For the most part, happiness regressions are estimated on very large samples and the average parameters noted. We find in such studies, among other things, that church attendance makes people happy.

Yet it seems intuitive that this result would only hold for people who have faith. An atheist like me might find church a less than pleasant experience. I would like to test whether average effects hold across sub-samples of people with different preference sets.

They very well might, or some might at least. For example, we might find that money has sharp diminishing returns even if you are a materialist.

As well as looking at the impact of circumstances on happiness given certain preferences, I would like to investigate the effect of preferences on happiness given certain circumstances.

In initial research on this angle I am going to use HILDA, which is Australia’s longitudinal panel. Its first wave includes question about preference: how important is financial security, family, religion, community, career etc. to you? I can use these questions to sort people into groups.

In the long run, I would like to get a better picture of people’s preferences. In particular, I would like to find preferences in a trade-off form i.e. who not only values career but would be willing to sacrifice family to that end. Such data would open up a lot of interesting avenues.

There is some literature on how to infer preferences from longitudinal data that I hope will furnish me something adequate.    

In tackling this preferences stuff I am going to be referring to the existentialist philosophy I mentioned earlier, psychology literature on self-actualisation and human striving and basic economic theory. I will discuss this stuff in greater detail next time.

Thanks.