With governments struggling to deliver on existing commitments is it sensible to make them responsible for something as complex as personal happiness? Moreover, much of the happiness data is faulty and where it isn't it points to conservative measuresby Paul Ormerod / April 29, 2007 / Leave a comment
Over the past 40 years, GDP per head has risen sharply in western countries; but the average level of happiness reported by individuals has shown little or no increase. This apparent failure of economic growth to make people feel happier is one of the central claims in contemporary political debate.
From this premise, prominent happiness advocates, such as Richard Layard in his book Happiness: Lessons from a New Science, argue for big policy changes. One is that taxation should be made more progressive to reduce income inequality because, it is claimed, it is relative rather than absolute levels of income that affect happiness. Further, it is proposed that the nation’s happiness should be measured regularly, just as GDP is, and used to guide policymaking.
We take issue with the validity of much of the happiness research on which such recommendations are based. We do not argue that the wider objectives espoused by happiness advocates (who are mainly on the centre left of politics), such as reducing inequality, are in themselves invalid—rather that happiness research offers no basis for such propositions.
There are aspects of happiness research that have secure scientific foundations. These, however, imply policies that are usually thought of as traditional rather than liberal. In particular, there is strong evidence that marriage and the nuclear family are important determinants of individual happiness.
There is now a large amount of academic material on happiness (a term often used interchangeably with “life satisfaction” or “wellbeing”). The 1990s saw a big increase in research into the concept, with over 4,000 articles published by the year 2000. Since then, the stream has become a torrent.
Nor is the topic being neglected by the government. Its Sustainable Development Commission is promoting research on “how policies might change with an explicit wellbeing focus.” And a cross-government wellbeing indicators group is developing a set of wellbeing measures for informing policy (to be unveiled this summer).
The idea of focusing policy explicitly on happiness is clearly touching a chord. This stems from a reasonable assessment that welfare is derived from a great deal more than material goods: from the quality of one’s personal relationships, from feelings of common cause and shared experience with others, from enjoyment of nature and from good governance, among other things. However, it is one thing to value happiness; quite another to expect the happiness of society as a whole to be measurable or to respond to policy intervention in understandable ways.
The seminal article in happiness research was published by Richard Easterlin of the University of Southern California in 1974. He claimed to show that average happiness in the US had remained at the same level over the period 1946-70, even though income per head had doubled. (The lack of any marked increase in average happiness as incomes rise seems not to be confined to the high-income countries of the west. A sample of 15,000 individuals interviewed by Gallup in China shows no increase in reported life satisfaction between 1994 and 2005, despite an increase in real income per head of some 150 per cent.)
These results are partly a function of methodology. Happiness is often measured using a three-point scale, where (3) equates to “very happy.” Average happiness can only increase when there is a net flow of people from lower to higher categories in the scale, and higher levels than (3) are not allowed. Given that the average score in such surveys tends to be around 2.2, many respondents must be answering (3). So even if they were reaching higher happiness levels over time, the survey could not track this. Furthermore, small improvements in welfare would not be enough for most people to swap “fairly happy” for “very happy.” So the tendency for such surveys to record only weak increases in reported happiness over time is partly an artefact of measurement.
But there are much more fundamental doubts about what happiness data means. In policy discussion, the devotees of happiness research focus on the lack of correlation between happiness over time and GDP, or income, per head. Yet there is a long list of other variables that might be thought to contribute to human wellbeing but that are also not related to happiness over time—for example, public spending. After allowing for inflation, public spending in the US almost doubled between 1973 and 2004. In Britain, it rose by 60 per cent. Yet in both countries, recorded happiness increased by a mere 2 per cent. If we were to rely on happiness data as a basis for policy, what is the point, one might reasonably ask, of all those schools and hospitals?
Leisure time, certainly in Europe, has increased. Between 1979 and 2002, the OECD shows that average annual working hours in Britain fell by 6 per cent. In west Germany the fall was 16 per cent; yet happiness fell by 5 per cent over this period. Crime in European countries is much higher than it was 30 or 40 years ago, yet this has had no discernible negative effect on happiness. And since the early 1990s, crime has fallen noticeably in both the US and Britain, after the sharp rises over the previous decades, yet these swings also do not register in the happiness data.
In the US, life expectancy for white Americans rose from 72.0 years in 1972 to 78.0 in 2003. For African-Americans, the increase was even greater, from 64.6 to 72.7, representing not merely an absolute rise, but a narrowing of the gap with whites. Gender inequality, as measured by the median earnings of women compared to men, has fallen sharply. In the US in 1972, women’s earnings were 58 per cent those of men, a figure that rose to 75 per cent by 2003. But the happiness time series is unmoved by these dramatic and desirable social changes.
There is also no connection with the level of income inequality, a factor that Layard and others argue strongly affects happiness. The US has seen the biggest increase in inequality in recent decades, yet if it is affecting the wellbeing of individuals, it is certainly not being registered by the happiness data. Using the standard measure of income inequality, the Gini coefficient, the data shows that inequality has increased throughout the period, while happiness has both risen and fallen.
So if measured happiness shows that decades of economic growth have made no difference to the welfare of western citizens, the same could equally be said of political labours of all descriptions. Public expenditure, leisure time, crime, gender inequality, income inequality—none of these are in any way correlated with measures of happiness over time. And although measured depression has risen in the postwar period, this is also not reflected by a downward trend in the happiness index. So one could conclude either that the attempt to improve the human lot by social and economic policy is futile or the data is not telling us anything of value.
Why may the data be telling us little of value? Stepping aside from economics for a moment, consider the closing scene of George Orwell’s 1984. Winston Smith, the central character, finally achieves happiness. He loves Big Brother. The years of attention devoted to his mental health by O’Brien have paid off. Smith is finally content and reconciled with the society in which he lives.
Of course, this is fiction, but things almost as strange have actually happened. The death of Stalin in 1953 created mass grief throughout the Soviet Union. Contemporary accounts make it clear that this was a genuinely spontaneous outpouring of emotion. The fictional happiness of Winston Smith and the apparently genuine happiness of many of the citizens of the Soviet Union under Stalin suggest that happiness cannot be separated from its social context. People adapt to their circumstances, and it is possible to be happy in very different contexts.
This view is confirmed by some of the more serious research within happiness economics. Easterlin himself, the doyen of happiness research, recently provided evidence that there is complete adaptation to levels of income over time. In other words, although an increase in income does usually lead to a rise in happiness, the feeling is only temporary, and people adapt rapidly to their new income. In social psychology, Philip Brickman and colleagues showed almost 30 years ago that not only did lottery winners report comparable life satisfaction levels to non-winners, people who had become paraplegic within the previous year had only slightly lower levels of life satisfaction than healthy individuals. A much more recent study has found that although the average life satisfaction of people who sustain a moderate disability initially falls, within two years it recovers completely. So it appears that reported happiness over time can adjust rapidly to changes in economic or other circumstances, which explains why there is no trend of increase over time.
Despite these serious shortcomings, happiness advocates are pressing for a set of “national wellbeing accounts” to supplement the more traditional national economic accounts that measure GDP, or national income. One of the main criticisms of GDP as a measure of welfare is that it increases even when economic activity produces things which detract from our welfare, such as pollution and congestion. GDP has long been acknowledged to be an imperfect measure of the wellbeing of a society, but it does contain information of real value. GDP growth is known to be correlated with increases in non-material benefits, such as longevity, reductions in infant mortality, environmental protection (once a certain point has been reached)—even indices of political liberty. Wellbeing is referred to by the Sustainable Development Research Network as accounting for “elements of life satisfaction which cannot be defined, explained or primarily influenced by economic growth.” Given the correlations between GDP growth and many benefits, wellbeing seems to omit rather a lot.
From a policy perspective, attempting to maximise almost any numerical measure—whether GDP or “gross national happiness”—will have perverse consequences. Increasing a measure is not the same as increasing whatever it is supposed to be measuring; the more complex the concept, the more this is the case. Witness, for example, the massive grade inflation and devaluing of standards that has taken place in education. Centralised systems of maximising indicators and setting targets increase the remoteness of those making decisions from those affected by them, reducing feedback and making decisions more vulnerable to partial or inaccurate information. An alternative approach is to apply the principle of subsidiarity—to make decisions at the lowest appropriate level, and forgo the central gathering of information wherever practicable.
When government intervention is required—and there are many situations when isolated decisions by individuals may not produce the best outcomes—it should be based on indicators that reduce the vulnerability of the policy process to information failure. Numbers and statistics can take on lives of their own, so it is vital that they contain real information which is easy to interpret. At least GDP, for all its faults, exhibits these properties.
Many people assume that happiness-based policy would advance causes which they already champion, such as environmental protection or social justice. But what would environmental policy based on happiness evidence look like? It is unlikely that a meaningful inverse statistical relationship would be found between overall life satisfaction and, say, nutrient pollution, which causes overgrowth of plant life in water bodies, thereby starving the water of oxygen. Yet surely most people walking by a stream would rather see fish, ducklings and dragonflies than uniform green scum. As it happens, conventional cost-benefit analysis has been taking such environmental preferences into account for some time, and would be much more likely than happiness data to make a good case for reducing nutrient pollution.
Further, many of the more soundly based findings of the literature on what contributes to happiness could broadly be described as “conservative paternalism.” Stable family life, being married, and having faith contribute to happiness, while divorce detracts from it. Yet the proponents of happiness policy are rarely heard promoting marriage and religious faith. Andrew Oswald of the University of Warwick is probably the leading original researcher on happiness in Britain. In a recent survey, he considers 95 academic papers on happiness and marriage. The literature shows that compared to the unmarried, people who are married are far less likely to suffer psychological illness, live much longer and are healthier and happier. The benefits are confined to those who are married rather than cohabiting—and they are large. In terms of health, the longevity effect of marriage may even offset the consequences of smoking.
Much of the research on happiness points to social models which are not attractive to modern liberals. For example, many studies on happiness and income inequality imply that a society in which social classes are static, where people only befriend those in a similar income bracket and where nobody tries to improve status or wealth relative to his peers may be less prone to “status anxiety” and discontent than a more socially fluid, meritocratic model.
And there is a country where happiness arguments have already produced morally questionable outcomes. The kingdom of Bhutan is cited approvingly by leading happiness advocates for being the first country to use the concept of gross national happiness as the basis for policy. In this fortunate nation, national dress is compulsory and, until 1999, television was banned. Bhutan wants to protect and maintain its culture, so in the 1980s and 1990s the government effectively expelled many of the minority ethnic Nepalese population, who fled to camps in Nepal.
Bhutan highlights the fact that it will always be a political and moral judgement as to what kind of happiness-increasing measures are acceptable; there is no objective measure. Instead, we should presume individuals to be arbiters of what does and doesn’t make them happy and suppose them to have some ability to convey this to their fellow citizens. That is the democratic way. The resulting compromises may be slow to emerge and may have some negative side-effects. For example, the “status anxiety” produced by meritocracy is as much an outcome of social and political choice as it is of liberal economic policy; yet people have conveyed that they prefer meritocracy (however imperfect) to the alternatives—static social classes or enforced equality.
It is entirely reasonable to want to live in a society not totally dominated by commercial or material concerns. It is entirely reasonable to want a feeling of shared values with one’s fellow citizens, and to feel that both you and they are not entirely driven by individual self-interest. The attention that happiness research has attracted is a manifestation of these aspirations. However, there are very many developments that both happiness research and common sense would expect to have affected (for good or ill) the contentment of citizens in western societies over the past 60 years. That they appear not to have done so suggests either that trying to increase the sum total of human happiness is futile, or that there are real problems with measuring happiness.
Certainly, to claim that only one particular change—increased material living standards—makes no difference whatsoever to happiness seems a case of choosing one’s evidence to fit one’s argument. We would have expected the huge increase in public spending in the second half of the 20th century, or the increase in leisure time, to have had an impact on the happiness time series. They didn’t.
Giving unhappy people the message that they are entitled to be happy without any effort on their own part will not make them happy. Moreover, if the state struggles to be held to democratic account for its current responsibilities, the last thing it should do is ambitiously expand them. In 1944, Friedrich Hayek predicted that as the state expanded its responsibilities, it would become sclerotic and exceed its capacity to respond to people’s demands and aspirations. As a result, they would become disillusioned with democracy and calls would be made for decisions to be “taken out of politics” and placed in the hands of experts. The aspiration that governments should promote something as complex as happiness is likely to provide a further twist to this disillusionment.