Guest post by @peterroderick15
How to measure happiness at a population level? Happiness itself is a devilishly slippery concept at an individual level (is it about income, or mental wellbeing, or physical health, or social interaction, or life expectancy, or self-reported interpretations of the word happy…or something else entirely?), never mind when considering units of tens, hundreds of thousands or several million people.
It’s a worthwhile task, however, if it gives insight at a broad brushstroke into the lived, day-to-day impact of long term government policy and societal shifts, and into how people are actually feeling and experiencing the world around them.
GDP: a redundant, unhelpful notion
It’s unfortunately also a hard task, and so for the most part it has been left untried. Instead we’ve often defaulted, when trying to understand the health and happiness of populations, to that bellwether of international progress that is Gross Domestic Product (GDP), an indicator which seeks to measure a country’s total economic output by combining consumer spending, investment, government spending, and exports minus imports. First proposed by US Secretary Simon Kusnets in 1934, GDP (or its local derivative, GVA) has held the pre-eminent place when ascertaining if an area is ‘doing well’ on the national or global stage. This is despite almost universal criticism from left- and right-wing commentators, aptly summed up by Robert Kennedy’s comment in 1968 that GDP ‘measures everything except that which is worthwhile’. Things which increase GDP, for example, include house fires, burglary and car accidents, while things which have no effect on GDP include volunteering and childcare.
Other options
Given its obvious inadequacy, various other measures have been proposed to replace GDP which more adequately reflect human progress and flourishing. These include:
• The UN’s Human Development Index, which combines life expectancy at birth, mean years of schooling for adults aged 25 years, and gross national income per capita.
• A more local indicator, the Happy City Index, which combines a complex set of indicators to measure the ‘happiness’ of English core cities including general city conditions, sustainability and equality.
• The management consultancy firm Grant Thornton’s ‘vibrant economy’ index for English local authorities.
• Bhutan’s famous decision to do away with GDP and to measure ‘gross national happiness‘, using an index which combines nine distinct domains such as education, time use, cultural diversity and resilience, and community vitality. 5
The sustainability imperative
These measures are, in various degrees, far more preferable than the current hegemony of GDP. But none of them fundamentally tackle a rather perturbing question: what if countries are buying happiness in the now at the price of happiness in the future? What if we are measuring population wellbeing levels founded on ecologically unsustainable foundations? For sure, some of the above include a weighting for environmental sustainability in their judgements – but it is just that: one of many weighted components in a long list. High carbon emissions and local ecosystem depletion scores can be neatly annulled by good performance on public transport links and GCSE grades. This seems an insufficient reflection of the current global ecological crisis. We will soon need the equivalent of three planets to support human resource consumption. There are nine established planetary boundaries (such as biosphere integrity, ozone depletion etc) and we are currently overshooting into and beyond the zone of high uncertainty on four them (see Steffen et al, 2015). So the environmental imperative is surely not just one item in a list of ‘jolly nice things to have’, but the key aspect of future human flourishing.
Source: Steffen et al (2015),
There is a very simple way of creating a population wellbeing measure which adequately balances current happiness and future human existence: create a measure which takes the good stuff (e.g. equality), divides it by the bad stuff (e.g. carbon footprint), and see what comes out the other side. This would enable us to judge whether short-term happiness is being produced at the expense of the long-term happiness of the planet. Put simply, it would be a ‘One Planet’ approach to happiness.
Fortuitously, the Happy Planet Index (HPI), created by the New Economics Foundation (NEF), is a way of doing just that. It is built around a simple play-off of equality, longevity and wellbeing on the one hand and ecological impact on the other.5 The calculation of the HPI uses 4 data points: self-reported wellbeing, income equality (the Gini coefficient) life expectancy, and the Ecological footprint of a country measured by the average amount of land needed, per head of population, to sustain a typical country’s consumption patterns. NEF sum this up with a neat little graphic:
Applying the HPI locally
The HPI has only been calculated at nation state level (the UK is currently 34th). So what happens if the NEF methodology is applied at a more local level, that of English council areas: is the data there to make it happen; is the index still valid; and does it tell us anything interesting?
Well: some compromises have to be made and data has to be dragged in from several different places, but it’s not that hard to pull the right numbers together as a starter. A modified form of the NEF model can be built using life expectancy for English upper-tier local authority areas taken from the PHOF, self-reported happiness scores collected by the ONS as part of the new wellbeing section on the annual population survey (APS), the ‘Slope’ index of inequality (SII) between the top and bottom IMD deciles in each LA, calculated by PHE, and total carbon emissions per capita published by BEIS at local authority level. The model equation is below:
So, for example, for the city of Leeds, on the top of the equation a 3 year rolling average life expectancy at birth (2013-15) of 80.18 years is multiplied by 100 minus the SSI of 10.45 (2013-15, measured in years, and – as it is inequality – lower is better), and by 100 minus self-reported unhappiness (2015-16; the score is actually the proportion of people answering 0-4 on the APS ‘happiness’ question i.e. its an ‘unhappiness’ question, so again lower is better). All of this is divided by the tonnes of domestic- and transport- related CO2 emitted in the LA area, adjusted per capita, and a modifier is included to bring the numbers down into something sensible.
Results
When the individual scores are combined within the model, a 0-4 index is created. The graph below presents the local HPI scores for 149 English LAs
The following can be observed:
• There is a threefold variance in the index scores, with the top authority scoring 3.45 (Hackney) and the lowest scoring 1.02 (West Berkshire). The scores are normally distributed, but positively skewed towards lower scores (i.e. scores above 2.5 are outliers).
• To understand the make-up of the scores, let’s take as an example the top and bottom LA areas:
o Hackney has life expectancy at birth of 80.7 years, which is slightly lower than the national average of 81 years. The Slope Index of Inequality is 4.8 years, much less than the average of 7.3 years nationally, and the average unhappiness score is 7.66, slightly above the national score of 7.3. Hackney’s CO2 emissions are however very low, both for transport and domestic emissions (1.5 and 0.6 tCO2 per head respectively, compared to a national average of 2 and 1.7). Put simply, Hackney residents have slightly above average ‘happiness’, and very low carbon emissions, meaning that the area is, planetarily speaking, more happy than others.
o West Berkshire, on the other hand, has life expectancy at birth of 82.6 years, which is much higher than the national average of 81 years. The Slope Index of Inequality is 5.75 years, much less than the average of 7.3 years nationally, and the average unhappiness score is 7.32, bang on the national score of 7.3. West Berkshire’s CO2 emissions are however very high both for transport and domestic emissions (2.3 and 4.8 tCO2 per head respectively, compared to a national average of 2 and 1.7). Put simply, West Berkshire residents have average to high ‘happiness’, but very high carbon emissions, meaning that the area is, planetarily speaking, less unhappy than other.
• Interestingly, the top 17 areas are London boroughs. In addition, other urban areas in the south of England (e.g. Portsmouth, Reading, Luton) are near the top of the list, but you have to go as far down as the 39th area before you have anywhere remotely ‘northern’ (Nottingham).
• The eight core cities are towards the upper end of the distribution and have a higher average HPI (2.075 CI 1.997-2.221) that the average for all 149 LAs (1.969 CI 1.925-2.051), although this difference is not statistically significant:
City
Happy Planet index score
Rank
Nottingham
2.25
39th
Bristol
2.22
44th
Manchester
2.19
45th
Birmingham
2.19
47th
Sheffield
2.17
48th
Liverpool
2.06
61st
Newcastle upon Tyne
1.88
74th
Leeds
1.65
100th
• The reason London boroughs do so well, and the (mostly northern) core cities also perform their rural neighbors, is their lower-than-average carbon emissions. This may in turn relate to the theory that energy and transport efficiencies accrue to urban areas as a result of population density (for instance see here); however this idea has been challenged. It may also be the tendency for urban sprawl and minor towns surrounding city cores to be ‘out-bounded’ into other local authority areas in the UK (this would also explain the low ranking for Leeds, where this isn’t the case).
• Of local interest: Yorkshire and Humber authorities are highlighted in red on the graph. The highest ranked is Sheffield (almost average life expectancy, slightly higher inequality, slightly more ‘unhappiness’, but better than average tCO2 per head), and the lowest is North Yorkshire (higher life expectancy, low inequality, slightly more ‘unhappiness’, and very high carbon emissions. Yorkshire local authorities are towards the lower end of the distribution, with an average score of 1.69 versus the national average of 1.97
Limitations of the model
There are a number of limitations to this model which should be noted:
• It presumes that if the same methodology is applied across all local authorities, the comparison of end results is valid and useful; thus units of measurement do not matter for the end results. However if the order of magnitude of each individual part of the top half of the equation were wildly different it would mean that, for instance, equality would contribute less to the overall score than life expectancy. As it happens they are all broadly similar, in the high 10s.
• The model places a big emphasis on life expectancy: as well as the raw indicator, the Slope Index is based on LI inequality across IMD deciles. The NEF Happy Planet index uses the Gini coefficient, a measure of income inequality, which is unfortunately not produced at lower than national level.
• The ‘Happiness’ question is abstracted from a set of 4 questions ONS use to collect data on self-reported wellbeing, and isolating one single measure from this is risky, even given that the questions themselves are fairly well validated. The NEF model uses the internationally recognised Gallup world poll, based on asking people to place them on a ladder from top (‘best possible life’) to bottom (‘worst possible life’). Again, not available at lower than national level.
• A large proportion of national tCO2 is produced by industry, and the BEIS includes this in each area’s score. This massively skews the results against areas such as North Lincolnshire, with its large industrial installations such as Singleton Birch quarry (which emitted 6 mt of Co2 in 2013). Because of this, the index presented above subtracted industrial emissions, which seemed a logical step as most large emitting industrial plants, power stations, refineries etc. produce goods and services consumed on a far wider geography than LA level. This does, however, leave these emissions out of the equation entirely, and if there were a way of accurately assigning them to local areas it would make the validity of the index stronger.
• Perhaps the largest caveat: the use of CO2 emissions as a measure of an area’s ecological footprint – without assessing other environmental impacts – is a big limitation of this model. Carbon dioxide emissions on their own are considered a poor proxy for sustainability. An average score for the bottom half of the equation derived from a set of measures such as that used by Grant Thornton in their model would be better:
o Air quality (score)
o Recycling rate (%)
o Co2 emissions per capita (Kt Co2)
o Energy consumption (GWh)
o New residential addresses created in National Flood Zone (%)
o Previously developed land usage (addresses per ha)
o Dwellings completed (no.)
o Households on LA waiting list (%)
o Planning applications (no.)
This still wouldn’t, however cover off a large swathe of the planetary boundary stuff (see Steffen et al above)
So what?
This project has used routinely available data to localise a national index, incorporating the crucial element of environmental sustainability as a key marker of a flourishing society. This reveals a few things:
• This can be done fairly easily from a data availability point of view. The accuracy/precision of the data is not bad, the validity of the raw data is fairly good, but the validity of the index as a whole in telling us what we think it is telling us must be regarded with caution, given the caveats above.
• The index suggests urban areas in England are more sustainably happy than suburban or rural areas; and (for those who are interested) Yorkshire’s local authorities are less sustainably happy (or less happily sustainable, whichever way you want to look at it).
• Differences in outcome due to deprivation and the north/south divide yet again rear their head, in this as in other wellbeing measures.
• Further work could be done to refine the model (particularly expanding the bottom half of the equation), and more deeply explore regional differences in the index.
(This piece of work was undertaken following a provocative masterclass on ‘Health in the Anthropocene’ led by Professor Trevor Hancock of the University of Victoria and hosted by the PHE/NHS Northern Sustainability and Health Network in May 2017. With thanks to Yannish Naik and Mike Gent. Any further comments on the methodology, implications or conclusions of this work are welcome – please email on peter.roderick@nhs.net)
Sheffield DPH 
Leave a comment