Presentation Transcript: weaknesses, 2nd and 3rd paper, the budget constraint

We were recently required to give 15 minute presentations, so I mashed up three topics. The first is the things about my thesis that I am most worried about, the second is my other two papers (after the paper presenting the model) and the third is some thought bubbles I had recently about the budget constraint. Fears and the 2nd and 3rd papers ended up getting mashed together because I thought that would be the easiest way to discuss these issues. 

Papers 2 and 3 and big weaknesses

My first paper is going to be purely theoretical. I’m going to try and pull together the literature from economics, psychology and philosophy to produce a preliminary model of happiness.

One of the main motivations for this is that happiness studies have been very empirically driven since inception and this is now reaching the point where there is some confusion in the inferences we draw from our measurements. It would be beneficial, I think, not least of all for our scientific practice, to develop a clear, mechanistic theory that agrees with our intuitions. We can then discuss our empirical findings in the context of this model.

A related major motivation for this paper is that I worry that our admirable dedication to measurement means that we are overlooking important elements of happiness that can’t be measured. This wasn’t a problem in the early days of hedonic psychology and the like but recently I get the feeling that we know so much about some things that we’re growing increasingly comfortable acting as though that’s all there is. I am talking here especially about the differences between subjective wellbeing, which is mostly an affective concept, and psychological wellbeing, which is a much more intractable existential concept.

So I’m mostly doing theory and mostly interested in theories that can’t be measured. This brings me to my first big worry with my paper, which is that I won’t have very good empirical evidence to back up what I’m saying.

This doesn’t mean that I don’t intend to do anything empirical. Rather, it means that I am going to go looking for suggestive empirical evidence rather than gold standard identification.

Here are the two thought bubbles that I have at the moment:

First, I want to emphasise the importance of preference. Most happiness research has estimated across huge samples and taken the average. We are left with stylised facts like: married people are happier. But then Bruce has a paper where he shows that married people who are in dissatisfied marriages are actually made less happy by their marriages than unmarried people.

The issue here seems to be one of preferences.

On the flip side, we get some research into the impact of certain preferences. Tim Kasser, who I intend to write a paper critiquing, has produced several studies showing weak negative impacts from materialist preferences on happiness. Unfortunately, not one of his papers controls for whether the participants are successful materialists or not.

I’m currently poking around on this stuff in HILDA and I think success as a materialist is important. Transitional effects of income growth are more powerful and sustained for materialist than for non-materialists, and rich materialists are happier than rich non-materialists. I also have a data set of all Dutch millionaires and they seem to be an exceptionally happy bunch.

This leads me to the second paper that I want to write for my thesis. I want to use big data sets to put some people into groups based on strong preferences for things that are easier to observe, like materialism, then use these grouping to see whether preferences satisfaction is important to happiness or just doing the things that should make people happy on average.

This would allow us to determine whether the source of happiness really is subjective, or whether there are actually some things that will just work for everyone.

For this work I am hoping to be able to combine the 3 big, high quality longitudinal panels from Australia, Britain and Germany.

The hardest part will be sorting people into groups based on preferences because preferences are hard to observe.  HILDA has some questions like ‘how important is financial security to you’ which can be used to proxy for preferences, but these are often worded poorly. For example, financial security is not the same as materialism.

More importantly, you really want this preference sorting to be done in terms of what people are willing to trade off, which the HILDA variables don’t capture. There is little reason in the HILDA questions for people not to answer 8+ for community, family, financial security and leisure even though maximising all of these preferences simultaneously is likely to prove extremely difficult.

I have a lead on how to surmount these issues in a paper by Manski called ‘Identification of income-leisure preferences and evaluation of income tax policy’. This explains a method for reading trade-off preferences out of longitudinal data. Hopefully it will lead to a wider literature; I haven’t looked into it much yet.

Moving to the third paper: one of the most important findings in this literature in recent years is the consistent relationship between log-form income against happiness. I want to see whether this functional form holds for other variables that I have in my agency cluster like health and political rights, and also for other determinants of happiness like family, leisure, religion and community participation.

Perhaps most importantly, I want to see whether some factors do not exhibit diminishing returns. In particular, I am curious to see whether aspects of psychological wellbeing or meaning  do not exhibit such a truncating relationship with happiness. I haven’t thought about this paper much. Data issues will, I fear, be very difficult to surmount. 

So that covers the first of my main concerns – empirics – and the content of my second and third paper. My other two fears are the interdisciplinary nature of my research and that I’m using some things that aren’t very popular within the disciplines I’m drawing on.

There is a lot of buzz around interdisciplinary research at the moment, but my impression is that this is mostly about having specialists in each field come together to work on a complex project. That is quite different from what I’m capable of, which is having a generalist work on issues that intersect several areas.

My fears are twofold. First, because I don’t have a hugely deep training in any of the disciplines I’m working in I will miss some critical insights or lack some critical technical tools. I might, for example, not be able to express the psychological aspects of my study using the appropriate terminology, and my econometric abilities might be deemed too shallow by economists.

Second, nobody will take the work very seriously because it won’t fit into an ongoing research stream. For example, the economics stuff in my paper doesn’t have much to say about reference groups and relative status, even though this is a huge part of the research agenda on happiness in economics.

These fears have been amplified during the few conversations I’ve had with happiness scholars that I’ve met at conferences. They were always nice but they were also quite dismissive of what I was doing and we had trouble communicating because I couldn’t articulate myself in a way that fit their paradigm. For example, economists keep coming back immediately back to utilitarianism in a way that I think isn’t very healthy for happiness research.

Onto the fact that I’m using stuff on the outer in my disciplines, there are three main issues here:

First, existential philosophy is generally ignored by philosophy, except for Heidegger, who isn’t really an existentialist but a metaphysician. The reason for this is that philosophy has actually never been interested in the meaning of life, only in metaphysics. Sartre, Beauvoir, Nietzsche, Kierkegaard and Jaspers did bad metaphysics, so they are largely ignored in favour of Heidegger, who did great metaphysics.

The happiness literature goes on and on about Bentham and Aristotle. These guys produced useful words—utilitarianism and eudemonia. However, Bentham does very little to link utilitarianism to the good life (quite rightly), and Aristotle provides very little in the way of content to his notion of Eudemonia.  Nietzsche wrote hundreds of pages of content for Eudemonia, but many Westerners seem to think him a boorish apologist for German nationalism, so they ignore all that.

On related notes, I’m writing an economics PhD with very little mathematics and only weak identification at a time when applied micro is obsessed with perfect measurement and macro is obsessed with dense mathematical models.

I’m also trying to make psychology take more seriously issues that dominated its earliest discussions in Freud and Jung but which it has since tried its darndest to avoid in its quest to be a science because they are hard to measure.

Getting people to take me seriously when I start bringing in all this stuff is going to be difficult.

The budget constraint

My model is a utility function:

Utility = agency + SWB + PWB

An obvious question follows: what is the budget constraint?

My initial thoughts regarding an answer were all bad. I now realise that there are two main constraints: the agency score, and information, especially about ones authentic preferences and how to satisfy them.

Recall that the agency score is basically derived from Amartya Sen’s definition of development as freedom. It contains such things as: Health, Income, Leisure time, Political power, Rights and other legal protection, ‘Capabilities’, which includes things as simple as access: no good being a billionaire on a dessert island.

This is more expansive than simply income, which is the neoclassical budget constraint. This is fitting, because maximising the good life function involves more than maximising the neoclassical utility function.

Let me give one example to illustrate how this budget constraint could be interesting. Consider the character Spartacus from the film of the same name. Spartacus is a slave who fights in the gladiator pits. He eventually leads a revolt for freedom in which he is very successful because of its fighting abilities.

Spartacus would not get far maximising an income-constrained utility function because he doesn’t have any income. Indeed, there is very little that he has because he is a slave. What he does have is health, defined expansively to include mad fighting skills and charismatic leadership qualities. Within this agency constraint he is able to maximise his PWB score by leading a revolt.

Interestingly, this suggests that the strictness of the constraint will depend substantially on what preferences the individual has that are informing their SWB and PWB.

The agency budget constraint also has substantial implications for the SWB cluster. Someone with a lot of income, health and power is more likely to be exposed to positive stimuli and less likely to be exposed to negative stimuli. They will also be able to double down on positive stimuli and rapidly work their way out of negative stimuli.

For example, someone with good health is less likely to be in pain and more capable of leaving unpleasant situations like being stranded in the woods. Someone with money can spend more to prolong the length of their massage and can escape police extortion by causally paying bribes. As such, someone with a high agency score is better able than someone with low agency to maximise an affect function, though uncertainty will be a problem.

Some other thoughts occurred to me once I took this constrained maximisation perspective. First, many of the stylised facts of the happiness literature correspond to agency—health, income and political agency (Swiss referenda literature) notably. Yet we notice that these things seem to have diminishing returns.

This makes sense if we think about these things as a budget constraint. If the constraint keeps moving outward but you don’t maximise your utility function then it’s understandable that pushing the budget constraint even further out won’t have much effect.

I recognise that I’ve just suggested people have dumb preferences or at least don’t know how to maximise their utility as derived from their preferences. This is a huge problem for politics and for a lot of modelling, but here I think it’s actually quite a reasonable position to take. The plethora of self-help books, the renaissance of religion in the West and the popularity of Alan De Botton’s guff suggests to me that people really don’t know how to make the most of our unprecedented levels of income, health and political power.

What important ramification that follows from this is that there is little argument about the agency constraint but a lot of argument about what preferences and actions make people happy. As such, government should focus on increasing people’s agency and stay away from increasing their SWB and PWB. This amounts to government expanding people’s budget constraint rather than conditioning their utility function.

One thing I really like about thinking of the model in this way is that it plugs quite nicely into economics in a way that the old model did not. It’s really just an extension of good old-fashioned rational choice theory, but bringing psychology into the fold to more realistically define the utility function in terms of SWB and PWB.