Can we use dollars to compare how much various life events affect well-being?

The life events I propose to discuss here are things like major improvements or worsening in financial situation, getting married or divorced, having a child, serious personal injury, death of a spouse, being made redundant and change of residence. I will focus on subjective well-being, as measured by surveys which ask people for a numerical rating of their satisfaction with life.

One way to compare the impact of life events on well-being is to calculate what change in income would have an equivalent impact after controlling for other factors. Some readers might recall research findings for the U.S. and Britain which suggested that the increased income equivalent of a lasting marriage is around $100,000 and an increase in income of around $60,000 would be required to compensate for the loss in well-being associated with becoming unemployed. (These numbers come from some pioneering research by David Blanchflower and Andrew Oswald published in 2000.)

There are several problems with the methodology of this early research which tend to overstate the income changes equivalent to life events. First, the methodology is based on estimates of the (small) impact that higher incomes have on current well-being without taking account of the impact of higher incomes on future well-being. Higher incomes enable the wealth accumulation (and the redistributions through tax and welfare systems) that make it possible for people to maintain their well-being during periods when earning potential is diminished (e.g. during retirement) or when they incur heavy costs or heavy costs are incurred on their behalf (e.g. education and medical expenses).

Second, the methodology focused on the impact of being in a particular state (e.g. married or unmarried) rather than on the duration or timing of the effects that life events have on well-being. Life events typically have large impacts on life satisfaction for only a relatively short period.

Third, the methodology was unable to distinguish causation. For example, it was unable to assess whether married people are happier than unmarried people because marriage tends to make people happy or because happy people are more likely to get married.

Research in this area has progressed a great deal in recent years, with the use of ongoing surveys that enable changes in the well-being of the same sample of people to be linked over time to life events. The HILDA survey (Australian data) shows that the events with the greatest positive effect on life satisfaction for both males and females included a major financial improvement in the past three months, having been married in the last three months and birth of a child (less than nine months ago). The events with greatest negative effect on life satisfaction included being detained in jail, a major financial worsening at any time in the last year, a recent separation from a spouse or partner and recent death of a relative or family member.

A recent paper by Paul Frijters, David Johnston and Michael Shields estimates the one-off windfall improvement in finances needed to compensate for various life events (‘Happiness dynamics with quarterly life event data’, DP 3604, IZA, July 2008). The windfall approach seems preferable because a comparison of the effects on current well-being of different life events avoids the conceptual and measurement problems of attempting to compare the effects of life events with the effects of differences in income levels.

The authors obtained the following estimates of compensating windfall financial gains for various life events:

Death of spouse/ child: + $178, 300
Serious personal injury or illness: + $ 59,200
Change of residence: - $ 53,000
Birth or adoption of child: - $ 18,300
Marriage - $ 16,500
Separation from spouse or partner: +$ 14,900
Fired or made redundant: + $ 6,900
Victim of property crime: +$ 2,700

(Currency: Australian dollars; $A1 = about $US 0.80. Assumed discount rate = 5% ) .

I should note that these compensating windfall estimates are additive. For example, a person who is fired might become separated from his or her spouse and experience a major financial worsening at the same time.

It seems to me that the magnitude of these estimated compensating windfalls generally make a lot more sense than do the much larger estimates of income-equivalents of life events. Nevertheless, I feel uneasy about the idea that the life satisfaction of people who suffer the death of a spouse or child would be unaffected, on average, if they received a windfall gain of around $A 178,300 at the same time. Can any amount of monetary compensation actually be sufficient to enable life satisfaction to remain unaffected while a person is mourning the loss of a loved one?

Do well-being surveys measure utility?

Economists often think of utility and well-being as the same thing. If a person chooses to buy an additional unit of good A rather than an additional unit of good B, they tend to assert that this “revealed preference” shows that the marginal utility provided by good A exceeds that provided by good B. If asked to explain what this means in simple terms an economist might say that the additional unit of good A increases the person’s well-being by more than an additional unit of good B.

At this point some readers will immediately want to bring in complications like the possibility of irrational behaviour. A branch of economics (behavioral economics) explores this possibility, but I want to put this possibility aside for the moment.

The question I want to focus on is whether well-being surveys that are conducted by asking people questions relating to their personal well-being are measuring the well-being or utility referred to by economists. Some economists assume that it is, but I think they are mistaken.

What is it that the surveys actually measure? They measure a variety of different things. Most commonly they measure happiness or satisfaction with life by asking people to provide numerical evaluations in response to a single question. Some more complex surveys (e.g. the ACQOL survey) measure perceptions of the quality of life by asking questions about satisfaction with various aspects of life such as standard of living, achievements, relationships and health. Others ( e.g. nef’s “National accounts of wellbeing”) incorporate a framework of questions relating to life evaluations, emotional well-being, vitality, resilience and self-esteem, and feelings of autonomy and competence etc.

I think it is fair to say, however, that the surveys measure how people feel about their current lives. (Some do include questions about future security but when this averaged with other factors most weight is given to how people feel about their current lives.) For the purposes of this discussion let us call this “current well-being” and assume that the surveys measure it accurately.

At this point the economists reading this will immediately recognize that the surveys cannot be measuring utility because people often make trade-offs between their current well-being and future well-being. This is most obvious in savings decisions where current consumption may be sacrificed to enable a higher consumption levels to be enjoyed in future. It also occurs, for example, when people decide to put up with working long hours or spending a lot of time commuting in order to make their families more financially secure.

Economists still probably learn at an early stage of their study of the subject how to picture these kinds of choices in their minds, but for the benefit of anyone else who might be reading this I will draw a relevant diagram below. Readers who prefer stories to diagrams might prefer to read an earlier post, entitled “Do good decisions always make us happy?”. For the benefit of any readers who might find an appeal to authority more persuasive I should also mention that Gary Becker and Luis Rayo have suggested that the happiness measured in surveys can be viewed as “a commodity in the utility function in the same way that owning a car and being healthy are” (comment on Stevenson/Wolfers paper, Brookings Papers, Spring 2008: 89).


The possibilities curve in this diagram (shown in red) encompasses the various combinations of “current well-being” and “security” that are attainable by the decision-maker. It can easily be seen that points on this curve are superior to all attainable points closer to the origin.

The indifference curves (shown in blue) reflect the preferences of the decision-maker between current well-being and security. The decision-maker is indifferent between the combinations of “current well-being” and “security” on particular curves. She maximizes her utility at the point of tangency between the possibilities curve and the highest attainable indifference curve.

The point I am trying to make is that as a result of the decision-maker’s preferences she views point A, where current wellbeing is at a maximum, as inferior to point B, the point at which utility is maximized.

The obvious implication is that it is foolish to rush into policy recommendations based solely on consideration of how people can improve their well-being as measured in surveys. If happiness surveys suggest that people are behaving in ways that are contrary to measured well-being we should ask ourselves whether we have an adequate understanding of what is motivating their behavior, rather than assuming that it is a result of ignorance (an information problem) or human frailty (predictable irrationality).

Should governments collect subjective well-being data?

The idea of governments collecting data on our subjective well-being might seem slightly Orwellian to many people. It could bring to mind images of officials from the government statistics office knocking at your front door and telling you that they are from the government and they have come to help you by collecting information about what is going on in your mind.

However I don’t think anyone needs to worry a great deal about the implications for their personal liberty of proposals for government collection of subjective well-being data, such as in the recently published book, “Well-being for Public Policy” by Ed Diener, Richard Lucas, Ulrich Schimmack and John Helliwell. As discussed in an earlier post, such data would be unlikely to increase the influence that paternalistic interventionists may have on the policy making process.

The important issue is whether the collection of this additional information is warranted in terms of its potential contribution to discussion of policy issues.

In their concluding chapter the authors ask themselves whether enough is known about subjective well-being for government agencies “to initiate systematic programs for measuring it”. This is how they summarise their reasons for answering “yes”:
“The measures are sufficient to reveal some of the groups in society that are suffering, and they also tell us which groups are thriving. The measures already provide strong clues about the characteristics of nations that lead to the experience of a satisfying life for citizens, along with those that predict the opposite. The measures give clear clues about the activities and circumstances that tend to lead to ill-being and well-being. And when national accounts of well-being are instituted our understanding of these issues will only grow.”

Do we really need systematic programs for collection of information on subjective well-being to tell us about such matters? The measures of subjective well-being generally tend to confirm what we know already from information on incomes and other objective indicators of the quality of life. It seems to me that the important issue is whether collection of more data on subjective well-being would add reliable information that is not available from other sources.

The book discusses the potential contributions of subjective well-being measures in providing new information that could be relevant to discussion of policy issues relating to externalities, non-market goods, taxation, setting fines and compensation for lost welfare. Some specific examples caught my eye. It is possible that information on the extent of misery caused by different diseases could result in better allocation of public funds for medical research (p 134). Some research findings suggest that effects of airport noise on well-being of people in affected areas may currently be under-stated by its effects on residential land values (p 147). Subjective well-being information may help in assessing the value of public facilities such as parks to residents of cities who have access to such facilities (p 155).

The critical issue in considering the contribution that subjective well-being data can make to public discussion is whether this information is reliable (yields consistent results) and valid (actually measures well-being). My assessment of the relevant literature (in my draft paper on Gross National Happiness) is somewhat less optimistic than the view presented in this book. Despite all the noise in this data, however, I think the authors may be correct that enough randomness washes out in large samples to make the responses to single item questions sufficiently reliable for the purpose of creating national indicators (p74). Multiple item questionnaires such as those suggested by Ed Diener and Robert Biswas-Diener to measure “psychological wealth” (in their recent book, “Happiness”) could provide much more reliable information.

I think the authors make a fairly strong case that the surveys are measuring an aspect of well-being although I think it is an over-statement to claim that “the measures behave as they would be expected to behave given widely accepted ideas about what well-being is” (p 93). For example, the measures show a decline in well-being when people have children, despite the widely accepted idea that having children has something to do with well-being.

There is a risk that subjective well-being measures will cloud public discussion of policies rather than shed additional light on relevant issues if they come to be viewed as definitive measures of overall well-being. In interpreting these measures it is important to bear in mind that it is quite possible for people to make rational decisions to sacrifice some of their current satisfaction with life, in order to improve their own future well-being or that of their families.

Are "policy makers" responsible for our happiness?

“Societies need subjective indicators of well-being to aid policy makers and ordinary citizens in making decisions.” This is the opening line of the recently published book, “Well-being for Public Policy” by Ed Diener, Richard Lucas, Ulrich Schimmack and John Helliwell. The first three authors are psychologists (Diener has played a leading role in the field of happiness research) and Helliwell is an economist.

The final sentence of the first paragraph explains: “Overall, accounts of subjective indicators of well-being will help policy makers make wiser decisions regarding policy alternatives and help citizens be better educated about the choices that affect their lives”.

I am in favour of research to enable people to become better informed about the choices that affect their lives. I hope that what the authors mean by “help citizens to be better educated” doesn’t involve anything more sinister than publication of research findings.

While reading the book I became irritated by what seems to me to be a naive view it presents of the policy making process. Although the nature of policy making is largely incidental to the purpose of the book, I will devote the remained of this comment to policy making. I promise to focus on the substance of the book in a later post.

According to the “public interest” view presented in the book, public policies are made by “policy makers” who would make wise decisions if only they knew what policies would improve the well-being of citizens. In reality, however, the policy making process is a messy business which involves politicians seeking votes and hoping to further their careers, civil servants seeking to expand and protect empires, voters who have little interest in most policy issues and even less incentive to understand likely consequences of the proposals being considered, interest groups seeking to further the interests of the people they represent and electoral rules that may give disproportionate power to particular groups. The process also attracts ideologues of various kinds who wish to advance their particular views of the good society.

In my view, rather than attempting to persuade us that more information on the subjective well-being of citizens would help some hypothetical “policy maker” to make better decisions, it would have been better if the authors had sought to persuade us that this information would enable policy processes to produce better outcomes.

Would this have made any difference to the book? Although the basic arguments about the validity of subjective well-being measures and their potential usefulness would have been unchanged, I think this change of focus would have made some difference. In particular, it seems to me is that the authors would have had less difficulty convincing readers that they “do not advocate the idea that governments should intervene strongly to move society towards a primary goal of increased well-being” (p209). When most of the book seems to be devoted to telling “policy makers” how they can use subjective information to improve the well-being of citizens it is natural enough to expect readers to be concerned that some “policy makers” might act paternalistically in using this information. Some readers might not be entirely reassured that paternalistic “policy makers” would have regard to the findings of happiness research which show that humans tend to feel most satisfied when they perceive that they have freedom to choose how to live their lives.

The problem of how paternalistic interventionists might like to use research findings is placed in perspective once it is recognized that competing interests are involved in policy-making through discussions in a range of different forums. These discussions are about various things, but the matters discussed by vast majority of participants usually relate in some way to the effects of different policies on the well-being of people.

The important issue is whether measures of subjective well-being can make useful contributions to the discussion of policy issues.