What is the secret of happiness?

Now that I have your attention I will do my best to keep it – but you cannot expect me to attempt to answer such an important question in just a few paragraphs.

It might be appropriate to begin by issuing a health warning about my ability to answer such a complex question. I can claim a great deal of interest in the subject, but my expertise is limited.

In my view human happiness is ultimately about having a meaningful life – one that is meaningful to the person living it - but that is certainly not a secret. Wise people have been saying similar things for thousands of years.

What I am about to write about now has to do with momentary happiness and the way we pursue our goals. It seems like a good idea for individuals to pursue their goals in ways that enable them to experience many happy moments and not too much disappointment along the way.

Some recent research on the links between risk-taking, expectations, rewards and happiness has produced some interesting findings. I propose to present some of those findings in a somewhat novel way and to combine them with some additional speculations.

The research by Robb Rutledge (of University College, London) and colleagues involved presenting participants in a decision-making game with choices between certain and risky options and repeatedly asking them to report their momentary happiness. The study used fMRI to examine the relationship between happiness reports and neural responses. The study also made use of the Great Brain Experiment app, to test results on large numbers of people playing the decision-making game on smart phones and tablets. (The game is actually still available the Great Brain Experiment site and good fun to play.) The research is reported in an article entitled ‘A Computational and neural model of momentary subjective well-being’, recently published in PNAS. Good summaries of the article have been published by ‘The Telegraph’ and ‘The Atlantic’.

The study found that momentary happiness is determined by the combined influence of recent reward expectations and prediction errors arising from those expectations. The happiness equation takes the following form:

Happiness = baseline average mood + what you can settle for (CR) + what you'll get on average if you gamble (EV) + the difference between that and what you actually get (RPE). 

The equation takes account of the fact that memory fades, so that more recent events have a larger impact on happiness than earlier events.

The findings suggest that happiness depends not on how well things are going, but on whether they are going better than expected.

It is a mistake to interpret the findings as in the headline in one paper: “The secret of happiness? LOWER your expectations: A good day is when things are going better than expected”. One reason the results don’t imply that people should lower their expectations is because the measure of expectation used is average reward, rather than a direct measure of optimism or pessimism. It would be slightly more accurate to interpret the findings as suggesting that people can avoid disappointment by staying within their comfort zone instead of choosing riskier options that involve high average (expected) reward. But that interpretation is also inadequate because it overlooks the pleasure that people get from anticipating a high reward (even if the outcome is not as good as expected) and also overlooks the buzz that some people get when they gamble (choose an option with high potential reward but lower average reward) and win.

I have attempted to map out some relationships between expectations, forecast errors and happiness in the following two charts.

In the first chart, we begin in Quadrant A, with expectations and happiness as indicated by the blue line i.e. a happiness rating above 7. Unfortunately, it turns out that outcomes are below expectations, so in Quadrant B it is apparent that we have a negative prediction error of -3. When we translate that to the X axis in Quadrant C (using the 45⁰ line in the south-east quadrant) we find that an error of -3 corresponds to a happiness rating of less than 6.

So, the question now arises of how you should respond to that disappointment. One way to respond is to get back into your comfort zone and adopt a strategy involving lower expectations and no prediction error. That strategy avoids disappointment but it means that you forgo the pleasure of contemplating the happiness that you could expect, on average, under the original strategy.

An alternative response is depicted in the second chart. That involves sticking with the original strategy but improving your luck.

 An improvement in luck is shown by a shift in the relationship in Quadrant B. By improving your luck you are able to achieve an outcome better than expected and end up happier than you expected.

So, the secret of happiness is to get lucky! 

Jokes aside, it makes sense to stick to a strategy that you have good reason to think will yield high returns over the longer term, even if you experience disappointing results in the short term. In other words, the secret of happiness is adopt the strategy that you expect to yield greatest rewards over the longer term and “stay the course”.

Some readers might question the wisdom of that on the grounds that most people tend to be optimistic in their expectations, relative to average reward (or mathematical expectation). However, the findings of a recent paper by Gigi Foster and Paul Frijters, which examines the formation of expectations by undergraduates at two Australian universities, suggests that optimistic expectations are benign. The results suggest that apart from their direct contribution to happiness, optimistic expectations motivate people to work harder to achieve their goals.

So, adding all that together, the secret of happiness would be to adopt the strategy that you expect to yield greatest rewards over the longer term, and to back your expectations by staying the course and working hard. But you already knew that!  And something important seems to be missing.

The real secret of happiness, in my view, is to play the inner frame games of self-acceptance and cheerfulness, and to adopt an attitude of awe and fascination about the world. 


Postscript:
I linked to the wrong article by Gigi Foster and Paul Frijters. An abstract of the article I meant to to link to can be found here.

Can the "best possible life" get much better?

Someone is sure to try to tell me that the best possible life is a bit like Groundhog Day – it is as good as life can get. If I thought that the quality of life was just about emotional states I would agree. But there is more to life than emotional states, even though emotional states are very important.

The question is meant to be about the extent to which people perceive that it is possible for the quality of the best possible life to continue to get better, e.g. as a result of further advances in medical science or communications.

In happiness surveys people are often asked to evaluate their lives according to the Cantril ladder. The relevant question, as asked in Gallup’s World Poll, is as follows:

“Please imagine a ladder with steps numbered from zero at the bottom to 10 at the top. The top of the ladder represents the best possible life for you and the bottom of the ladder represents the worst possible life for you. On which step of the ladder would you say you personally feel you stand at this time?” 

I have not been able to find research that directly relates to the question of how perceptions of the best possible life differ between countries and change over time. Perhaps I haven’t looked hard enough, or in the right places. If any readers can steer me in the direction of relevant research I would be most grateful.

A few years ago, Angus Deaton, who has been associated with the Gallup World Poll, speculated as follows about the meaning of “the best possible life”:

“A simpler interpretation of the Gallup World Poll findings is that when asked to imagine the best and worst possible lives for themselves, points "10" and "0" on the scale, people use a global standard. Danes understand how bad life is in Togo and other poor places, and the Togolese, through television and newspapers, understand how good life is in Denmark or other high-income countries.

Such an interpretation is also consistent with Easterlin's conclusion that the "best possible life for you" is a shifting standard that will move upward with rising living standards. Thus, we might expect the Danes to continue to maintain an average rating of 8 as national income rises, provided they stay in the same position in the global income rankings. If this interpretation is correct, it would be an indication of how much the globalization of information has affected the perceptions of populations worldwide - because the consistently high correlation between income and satisfaction could not have existed in its absence.”

I have some problems with the first paragraph of the quote. Given the wording of the question, it seems likely that perceptions of the best possible life are strongly influenced by knowledge of what kinds of lives it might be possible for the respondent to live. There seems to be quite strong evidence that the reference groups that rural people in China use to evaluate their own satisfaction with life are their fellow-villagers. The reference group changes when people move to the city. (See paper by John Knight and Ramani Gunatilaka.) That seems to be related to the “paradox of happy peasants and frustrated achievers” discussed by Carol Graham. If people evaluate their lives according to a rural village reference point of the best possible life, they may not be overstating their happiness but they are basing their judgement of their well-being on imperfect knowledge of what is possible. In order to shed more light on such matters it would be useful to conduct surveys of migrants from rural to urban areas using the ACSA methodology discussed in a recent post.

When we consider high income countries, it seems reasonable to expect that perceptions of the potential for the best possible life to improve would be closely related to perceptions of whether today’s youth will have a better life than their parents. Gallup’s surveys for the United States suggest that from 1998 to 2008 around two-thirds of Americans were optimistic, and only one-third pessimistic, about whether today’s youth would have a better life than their parents. Since then, however, Americans have become equally divided on that question. That probably reflects ongoing economic uncertainty in the wake of the Global Financial Crisis.

It would be interesting to know to what extent people who consider themselves to be living the best possible life at present are optimistic about the potential for life to get even better. I don’t have access to the detailed Gallup data, but data from the World Values Survey relating to people who are “completely satisfied” with their lives (score of 10 on the WVS rating scale) may be relevant. The following chart is based on data for high income countries in the latest round of surveys (2010-14) who claim to be completely satisfied with their lives.

The chart shows that being completely satisfied with life does not prevent people from being in agreement with the proposition that because of science and technology there will be more opportunities for the next generation. Optimism about such matters might even help to explain why some people are completely satisfied with their lives.  

Postscript: November 2019

Recent research by Alan Piper has suggested that optimism and pessimism about the future have an important influence on life satisfaction that is not attributable to personality differences among individuals. The use of German panel data covering several decades enabled the author to control for individual fixed effects (including those attributable to personality differences). Controlling for fixed effects reduces the estimated coefficients, but their size suggests that views of the future have an important impact irrespective of personality differences. After controlling for fixed effects, the estimated reduction in life satisfaction experienced by individuals who report being pessimistic is comparable to that associated with unemployment.

The paper: ‘Optimism, pessimism and life satisfaction: an empirical investigation’, February 2019, German Socio-Economic Panel (SOEP)

The results suggest that while many people who have high life satisfaction might think that the life they are living is the best possible at present, that is consistent with them being optimistic about the potential for their lives to become even better in future.

Is your altruism a scarce resource?

In What Money Can’t Buy, Michael Sandel argues that markets and market values have come to govern our lives as never before. He suggests two reasons why we should be worried about this: fairness and corruption.

He is concerned about fairness because the distribution of wealth matters more when money is able to buy things that were previously available free of charge to individual consumers. I don’t think this argument gets to first base because the main examples cited – quality of medical care, quality of schooling, the ability to live in safer neighbourhoods, the ability to avoid queues, the ability to avoid socializing with poor people – are things that wealth has always been able to buy.

The “corruption” issue has to do with the possibility that our attitudes towards the good things in life may change when we put a price on them. In other words, some good things are degraded or corrupted when turned into marketable commodities. One example the author cites is blood donation. He refers to a study by Richard Titmuss which suggested that purchase of blood by commercial blood banks in the US had tended to displace voluntary donation. As blood came to be viewed as a commodity that was bought and sold, this apparently had a corrosive effect on norms of altruism.

Does this matter? I think it does matter when a government decides to provide a service which displaces the efforts of unpaid volunteers and voluntary money contributions. In that instance norms of altruism are displaced by coercion, since the government services have to be paid for from tax revenue. 

Situations can also arise where commercial activities displace services previously provided by unpaid volunteers and voluntary money contributions. However, commercial suppliers would need to be seen to have considerable merit, in terms of value for money for services offered, to succeed in markets dominated by voluntary activity. I find it difficult to see a case for preventing commercial suppliers from attempting to compete in sectors currently dominated by voluntary activity. And I also find it difficult to see a case for preventing people from making monetary contributions to charitable organisations rather than donating their time, if that is what they would prefer to do.

Sandel takes exception to the views presented by economists - such as Dennis Robertson, Kenneth Arrow and Lawrence Summers – who have argued that the altruistic motive should be treated as a scarce resource that should be relied upon only where the market system breaks down. He seems to be particularly upset by Summers’ view that we should save our altruism for our family and friends, “and the many social problems in this world that markets cannot solve”.

Sandel draws attention to Aristotle’s argument that virtue is something that we cultivate with practice. He suggests that altruism is like a muscle that develops and grows stronger with exercise.

That seems to me to be beside the point. Humans also develop intellectual skills through exercise, but still seem to insist that their intellectual skills (human capital) should be treated as a scarce resource.

Perhaps this is an appropriate time for me to make a personal confession. My altruism is definitely a scare resource. While I can see merit is developing my altruistic muscles, my desire to do that tends to evaporate when I feel that my efforts are being wasted. My time should not be treated as a free good, just because I choose to donate it.

Now, it is possible that I hold that view because I am an economist and have spent too much time over the last 50 years, or so, thinking about the opportunity cost of time. But I suspect that many non-economists hold similar views.

Would you be as willing to donate your time to good causes if your altruism was not viewed as a scarce and valuable resource? 

Should life evaluations be anchored to the best and worst periods of our lives?

Survey data on life satisfaction is a reliable tool for measuring some aspects of well-being. There is evidence (including in some research reported on this blog) that survey measures of life satisfaction are closely related to perceptions of achievement, personal relationships, standard of living, links to community, health and future security. There is also evidence that when people are asked to evaluate their life satisfaction they tend to think about things such as their career, romantic life, family, standard of living and health (See: Maike Luhmann et.al).

However, as previously discussed on this blog (here, here and here) there is a problem in comparing life satisfaction ratings from surveys conducted at different times because reference norms do not remain static.  When we are asked to rate our satisfaction with life we do so relative to reference norms, such as by comparing the quality of our lives with those of people we know, or by comparing our current lives with ‘the best possible life’. That means that we are rating our lives against benchmarks that can change as a result of such things as technological advances. For example, if we perceive that people living the best possible life have access to useful communications devices, such as iPhones, we may rate our own lives less highly if we cannot afford such devices.

If we see our lives improving in line with our perceptions of the best possible life, it is hardly surprising if we give similar ratings to their lives in successive surveys. It should be obvious that it would be a mistake under those circumstances to interpret stable ratings as implying that there has been no improvement in the quality of our lives. Before claiming that people are on some kind of hedonic treadmill that requires them to run faster to stay in the same place, researchers should be sure that the measurement tools they are using are actually capable of measuring progress.

Similarly, when researchers see life satisfaction ratings return to previous levels after people suffer some misfortune, they should be sure that they are using appropriate measurement tools before they claim that this means that the people concerned have not experienced any lasting loss of well-being. It is commendable and often therapeutic to “look on the bright side of life”, but if life evaluations reflect frames of mind it may be problematic to interpret them as a judgement about overall well-being.

That point can be illustrated by reference to a South African study in which conventional life satisfaction and happiness ratings were compared with scores on the ACSA scale. (See: Valerie Møller, Peter Theuns, Ida Erstad and Jan Bernheim, ‘The Best and Worst Times of Life: Narratives and Assessments of subjective Well-Being by Anamnestic Comparative Self Assessment (ACSA) in the Eastern Cape, South Africa’, Soc Indic Res, 89(1) 2008.)

The ACSA approach to measuring well-being and changes in well-being was first suggested by Jan Bernheim about 30 years ago. Its distinctive feature is to ask survey respondents to rate their current well-being by comparison with their memory of the best and worst periods of their own lives, with the best period being given a rating of +5 and the worst period being given a rating of -5. The scale is left open so that subsequent scores higher than +5 or lower than -5 are possible if individuals consider that their lives have improved or deteriorated sufficiently.

The South African study involved 46 respondents, 20 of whom were patients in a TB hospital. In general, the three measures of subjective well-being were strongly correlated with each other, but there was weak correlation between life satisfaction and ACSA for the TB patients. Again, while there were no significant differences between the averages of conventional life satisfaction and happiness ratings of the TB patients and others included in the survey, on the ACSA scale the average ratings of the TB patients were about 30 per cent lower than for the others included in the survey.

The authors note that while they did not ask for life stories when they were obtaining ACSA scores, they observed that respondents tended to string the momentous events in their lives together, using anchor periods as reference points to shape a coherent narrative. Interviews lasted between half an hour and 45 minutes and respondents “generally had no difficulty with the task”.

A couple of years ago I wrote a post wondering why more use has not been made of ACSA. I am still wondering. It seems to me that it would be potentially useful to incorporate an ACSA question for a sub-sample of respondents whose well-being has been monitored over an extended period in longitudinal studies (such as GSOEP, HILDA and BHPS). The accounts of changes in well-being over time provided by those surveys would be more plausible if life evaluations were anchored to consistent reference points corresponding to the best and worst periods of the lives of respondents.