# ESRC Well-Being Research Workshop at the LSE

OCTOBER 13, 2008

Last Friday I attended an ESRC Research Workshop on Well-Being held at the LSE. According to the blurb:

The time is ripe for a major expansion of well-being research in Britain â€“ in conjunction with leading overseas colleagues. Among public policy-makers, there is an increasing desire to promote well-being and a need for evidence on what works to promote it. And among social scientists there is a new capacity to throw light on well-being: its causes and its effects. Worldwide, research on these topics has already demonstrated the scope for rapid and important advances in knowledge. But the scale of such research in Britain is far too small. This one-day workshop has been organised to explore the possible intellectual content of such a cooperative endeavour.

Some of the most prominent researchers in this area were in attendance to give an overview of current work and I took some ‘impressionistic’ notes which can be found below.

## Well-being Research: the Way Forward by Daniel Kahnemann

• Living and thinking about it
• Attention
• There are 2 selves
• Experiencing self
• Remembering/Score-keeping self
• Used to think that experiencing self was what was important (Edgeworth)
• Remembering self not very accurate – cites own research on pain for medical procedures
• But now thinks remembering self is more important
• Implicit in this is an acceptance that there are at least 2 distinct dimensions
• Current well-being/happiness questions are problematic because they are mixed containing some experiencing self and some evaluative/remembering self
• New, huge, dataset from Gallup is making a big difference
• 1000 people polled a day with 40 questions on well-being
• Ladder of Life question in Gallup measures ‘Life Evaluation’
• Despite having different questions replicates existing results from DRM etc
• Attention and ‘Focusing Illusion’
• Norbert Schwarz study: how much pleasure do you get from your car
• Reasonable correlation with car monetary value
• Also asked: how much pleasure did you have in your commute this morning
• Zero correlation with monetary value
• How many dates did you have last month and how happy are you these days
• Happy first, dating second no correlation in response
• Reverse order: large correlation
• Leads to errors in prediction since we know attention alters valuation
• e.g. to predict pleasure/utility from car need to ask: how much enjoyment do I get from car when I do not think about it
• How happy would you be if you moved to be the california
• But this is mistaken [ed: is this not often taken into account as evidenced by phrase ‘always think the grass is greener’]
• Gallup data: huge correlation with money
• Remembered happiness: Money worries, health coverage, general health are main predictors
• Experienced happiness: pain + social activities
• Children: negative impact on general impact but when asked about children people are very positive – both views are correct
• Easterlin hypothesis:
• Some questioned this (Stevenson and Wolfers)
• But focus on ladder of life question
• Looking at positive/negative affect still find that within-group slope with income is steeper than across group/time effect (i.e. Easterlin hypothesis)

## Income and happiness in developed countries by Steve Nickell

• No obvious relationship on the ladder of life question
• But cross-country regressions are pretty dubious (too many variables)
• Time-series data
• Curvature for classic CES: y^{1-rho} get rho ~ 1.2 across a whole variety of countries (1.1 - 1.4)
• Time series: happiness is pretty horizontal (in the US) though income risen lot (even taking account of dispersion)
• Some reasonable support for relative income hypothesis
• But really want panel data (deals with endogeneity)
• Only one such panel: GSOEP (West Germany)
• Regress happiness on log income, log reference income, controls (state,year,individual dummies etc)
• Income alone: large +ve coefficient
• Include relative income: income coefficient disappears

## Income and the Evaluation of Life by Angus Deaton

• Gallup’s World Poll
• Why it’s great [ed: the value of having early access to proprietary data!]
• Gallup result is very similar to the World Values Survey (His paper from last year – [ed] see my comments last year)
• Could argue that steep and then flat but log sems to fit better
• Difference here with Steve Nickell
• Within country analysis
• Collecting income data within country is hard particularly in poorer countries
• Does some analysis in the US and does not get relative income affects at all (with ladder question)
• Suppose people do care about relative income. There are serious (‘ethical’) problems with a consumption tax or not worrying about GDP growth: you hurt the non-envious and help the envious

## Questions on Preceding

• My question:
1. If focusing illusion is common across goods does it actually end up leading to bias in/incorrect choices
2. Once we accept that attention has such large effects it poses difficult questions since it suggests that people’s preferences/enjoyment has a significant endogenous component.
• Several on relative income
• Replication across countries

## Workshop on happiness research by Michael Marmot, Andrew Steptoe, and Jane Wardle

Michael Marmot

• Health as a measure of well-being
• 28 year gap in life-expectancy between poorest part of Glasgow (Galton - 54) and richest (Lenzie - 82)
• Major wealth effects on health outcomes even though (e.g. in the UK) people have all got enough to have pretty good healthcare
• Relative effects of income (status?) has a major impact on health
• Relative position not relative income (income != status – at least not always)
• Control for environment
• Whitehall II study: look at poor physical health by deprived living area and grade level in civil service. Deprivation really matters when you are in the lower grades. [ed]: Suggests a) interaction effect b) that status matters more than area you live in
• Work stress: Coronary heart disease strongly linked to work stress
• Social relationships: mainly important on negative side (bad interactions are bad for you …)
• [ed: general murmurings from room throughout data presentation about what these correlations imply. Significant issues of causality and selection bias …]

Andrew Steptoe

• Meta analysis of positive affect and health
• 18% reduction in prob. of mortality (even when controlling for other variables: smoking, BMI, social position etc)
• Issues:
• Confounding: even though have controls direction of causation goes the other way (health to positive affect)
• Genetics: simple correlation
• Lifestyle: happier people lead healthier lives (or vice-versa)
• Biology: positive affect associated with
• Lower cortisol over working and non-working days
• Lower heart rate over day
• Lower systolic BP over the day
• Reduced inflammatory responses
• Independent of socio-demographic factors
• Happiness measure matters (a lot)
• Using retrospective questionnaire measures find no relationship of positive affect with other stuff
• But using EMA or DRM (i.e. more instantaneous stuff) find relationships
• Cross-cultural comparisons: Japan vs. the UK
• Japan reports less +ve affect then UK (e.g. Gallup)
• Find this in DRM studies of university women
• And, importantly, find impact on cortisol levels (UK women lower than Japan)

## Mapping Pain and Well-Being in Real Time and Yesterday by Alan Krueger

• Study in the Lancet (w/ Arthur Stone) on pain in general population (using diary study)
• Data came from PATS, ~3900 people (by Gallup)
• Pain rises with age but very flat 45 - 65 (for men and women)
• Correlated with SES: poorer people in more pain (~20% of people with income under $30k in reasonable to severe pain compared to 7% for >$100k)
• People in pain work less and watch more television
• Now doing EMA-PATS study + biological info (Krueger and Stone)
• Check EMA and PATS are related (strongly correlated ~ 0.94 corrected for pain, and 0.92 corrected for happiness)
• Not a representative sample (v. hard to get participants)
• A world of pain – use Gallup survey to look at pain across countries
• Strong connection of GDP per capita and pain (~ -0.42 correlation)
• Questions:
• Source/duration of pain
• Biomarkers

## Knott and Scott

• Examples of kids with cerebral palsy and some other bad thing: expectations matter (despite having serious disabilities kids evaluated their life as as good as others)
• Support only: no effect
• Homestart: no effect or negative!
• Surestart: also been shown not to work
• Skills and support: slightly better
• Child Antisocial behaviour: benefits
• Quality of mental health professionals: matters a lot
• Very little long-term follow-up data
• Perry pre-school: good effects at age 27
• 10 years follow-up of Scott et al (2001) finds some long-term effects
• More evidence based psychiatry
• Quite a lot we can do if we do it in a skillful way

## Well-being and Aging by Felicia Huppert

• Negative stereotyping has large impact
• Older people exposed to -ve stereotypes do worse on stress, cognitive performance etc
• Causes of well-being
• Separate +ve and -ve in GHQ (found a big difference in impact of e.g. unemployment on +ve vs -ve affect)
• Magnitudes (as opposed to pure significant)
• What are important drivers
• Environmental affects likely to be large (much larger than genetic affects)
• Study in US IT company: RCT of mindfulness meditation found substantial impact
• How much is society losing from people not flourishing [ed: losing seems to mean losing money/GDP here]

## Work, Stress and Well-being by Richard Freeman

• Questions
1. Does working environment affect worker well-being
2. Can we specify workplace policies/practices make work lives better
3. Do measures of job satisfaction and well-being provide different information
4. Moving beyond survey measures
• Job satisfaction - one of most widely studied variables
• Correlated with health and turnover (people leaving associated with dissatisfaction)
• Two-factor model needed to explain some patterns
• Puzzle: unionized workers quit less but also less satisfied (expectations?)
• Job satisfaction and well-being
• When people quit and go to a new job their satisfaction goes up
• Results from various datasets they used (WERS - several people per workplace + a lot of detail)
1. Working environment matters a lot (could be workplace policy, culture, or selectivity)
• Workplaces bad (good) in one dimension often bad (good) in others
• [ed: so not some simple trade-off/optimization]
• Large changes in well-being after quitting and moving elsewhere (bigger than money impact)
1. Policy/practices matter but causality unclear
• Major endogeneity problems (if i have a job satisfaction policy is that because people are miserable)
• Well-being related with job attributes (hazardous, stress etc) in normal way
1. Job satisfaction and worker well-being
• Not that correlated
• Job-satisfaction important for well-being but less important that health and various other variables
• 3 things to do
1. Biomarkers at high/low satisfaction workplaces
2. Impact of change of jobs
3. Harvard network on work, family and health
• Check company work policy carefully
• Look at health outcomes, stress, sleep
• Found big correlation of manager’s attitudes and practices correlated with cardiovascular outcomes

## Co-operation and well-being by Armin Falk and David Skuse

David Skuse (development neuro-psychiatrist)

• Individual differences in happiness
• Role of genes and brain on behaviour
• Mechanisms of mental functioning underlying mental health
• Compensation for deficiencies …

Armin Falk

• [ed: computer battery ran out so this is very partial]
• Relative pay and fMRI results. Big impact of relative pay (Science 2007)
• Unfairness in principal agent setup (dull task and unfair division of revenue. impact on heart rate variability)
• Oxytocin study: look at genetic variations affecting oxytocin and see how they impact on trust in trust game (amount sent at stage 1)
• Mentioned current/future research on cultural formation on preferences