Presentation at the HLEG thematic workshop on "Multidimensional Subjective Well-being", 30-31 October 2014, Turin, Italy, http://oe.cd/HLEG-workshop-subjective-wb-2014

1. Angus Deaton, Princeton University
HLEG working group on subjective wellbeing
Herbert Simon Society, Turin
October 2014
USING WELLBEING MEASURES FOR
POLICY: PROMISES AND PITFALLS

2. Prolog
 We have measures, especially evaluative
measures
 We have many correlates of SWB
 How to use these?
 Government regulation and cost-benefit analysis
 Deshrouding: telling people “what works” for
happiness
 Parallel with NICE for health
 Up to people what to do with this, not enforced in policy
 Subject to the “curse of the average”: what is good on
average may not be good for you
2

3. The problem of z’s
 Think of conventional utility as max u which is a function of q and
z, former chosen subject to budget constraint p.q=y, but z’s are
not chosen
 Think of public goods
 We have a standard apparatus (CV, EV, COLI) for evaluating
changes in q (i.e. p and y)
 But behavior does not reveal the value of z
 Many doomed attempts in the literature
 Three possibilities
 z is chosen, e.g. political economy
 Change in z can be priced, shadow pricing
 Measure u directly: happiness regressions
 Shadow pricing is what governments do now
 But it is difficult and badly done
 Must keep this in mind when critiquing happiness measures
3

4. Consider examples
 Correlates of happiness
 Can we use them to value z’s, for CBA or deshrouding?
 Clocks and life-cycles
 Religion
 Relative income
 SWB over space
 Is SWB or safe?
 What about Sen’s critique that it risks neglecting important
deprivations?
4

5. Clocks and cycles
 People are tired in the evening!
 Knowing that can be useful
 U-shape of wellbeing in rich countries
 People don’t seem to know this & are interested
 Perhaps elderly don’t deserve as much as they get?
 Perhaps connected with social security
 If elderly are just more efficient in producing utility, a utilitarian would
give them more, not less
 Stress, anger, worry, decline rapidly with age
 Again, people are interested, but deshrouding?
 Not much room for action
 But again relevant for wellbeing of the elderly
 Which is also why those regressions are easy to interpret
 All subject to “average critique”: because it works on average doesn’t
mean it works for you
5

6. Religion
 Those who say that religion is important in
their lives have better SWB
 Is this a causal effect? Who knows? But
perhaps
 People become religious and change their
religion often, so potentially useful
information
 But also makes it easier to challenge the finding
 Policy? Countries have established religions,
perhaps positive social policy to encourage
6

7. Relative income
 Many claims that SWB depends on your
income relative to the income of others
 If so, there is an externality of my working
hard and earning more
 Case for a high marginal tax rate to
discourage this
 Coase solution seems unlikely to work
 But the econometrics is fraught, and it could
be your own past income, or something else
7

8. Geography
 Economist standard model is that people locate to maximize
utility given their abilities
 Conditional on ability (human capital?) SWB ought to be the
same over space if no restrictions, e.g. within the US
 Psychologists see people as “stumbling” on happiness: random
allocation
 Michigan and Californians agree that happiness is higher in California,
but are equally happy
 How would I know how happy I would be in some place I have never
been?
 Evidence: spatial diffs are small locally, large over bigger areas,
e.g. states on opposite coasts in US
 Perhaps costs of relocation are larger over larger distances
 If so, policy based on spatial differences can be unproductive
 E.g. noise regulation for airports
 Disamenity regulation in general
8

9. Sen’s critique
 US regressions of life evaluation on circumstances show
without controls
 Blacks lower LE
 Women higher than men
 Hispanics higher than whites
 Control for income, marital status, education, and age
 Blacks much higher LE than whites
 Hispanics higher than whites
 Women more than men
 D0es this mean that other deprivations don’t matter?
 For blacks, no correction for quality of education, higher
mortality, imprisonment, healthcare, etc. etc.
 But this cannot mean that those deprivations do not matter!
 Plausible SWB can’t behave like this: not there yet!
9