The Economics of Happiness, Part 5: Will Raising the Incomes of All Raise the Happiness of All?

In a famous 1995 paper, Richard Easterlin asked: “Will raising the incomes of all increase the happiness of all?” His analysis involved studying the evolution of happiness through time in Japan, the U.S. and Europe. His answer? “No.”

Betsey Stevenson
and I recently returned to examining the evolution of happiness in these three important regions, and we conclude that the evidence is not so clear cut.

First, Europe. The Eurobarometer Survey allows us to track average levels of life satisfaction since 1973, in the same nine nations that Easterlin analyzed. The relationship between happiness and G.D.P. in these countries is shown below:

Growth in life satisfaction and GDP

A few observations:

1) In eight of these nine nations, satisfaction grew as G.D.P. grew, and in six of these cases, the relationship is statistically significant.

2) But there are some pretty interesting exceptions. For instance, why has happiness fallen in Belgium, even as G.D.P. has grown? And why did happiness take so long to grow in Ireland, even as the “Irish miracle” led to tremendous economic growth?

3) While the pattern varies across countries, in most cases, the time series data suggest a satisfaction-G.D.P. gradient of about 0.2, with some larger, and some smaller.

4) This pattern of satisfaction growing with G.D.P. is less clearly evident if one analyzes only the early data (solid dots). This partly explains how we are able to make stronger inferences than earlier researchers.

5) In subsequent years, this survey (and indeed, Europe) has expanded to include more countries. We have analyzed this broader sample of countries, finding roughly similar conclusions.

Now, let’s turn to Japan, which is arguably the most interesting case study, because it went from a quite poor country after the war, to become one of the world’s economic powerhouses by the late 1980’s. Moreover, Japan is unusual because the government has collected life satisfaction data since 1958. Previous researchers had interpreted those data as suggesting that this incredible economic growth had yielded no gains in happiness.

Puzzled by this finding, we had the Japanese survey questions re-translated. It turns out that there was not one continuous question yielding a flat trend in well-being, but instead four separate questions asked in four separate periods. And during each of the first three periods — when economic growth was rapid — it was matched by commensurate growth in life satisfaction. The fourth question began in 1992 and satisfaction declined as per capita economic growth was anemic (0.9 percent) and unemployment became a problem.

Evolution of subjective well-being

Looked at this way, the dramatic growth in Japanese G.D.P. from 1958 to 1991 was matched by rapid growth in life satisfaction. The decline in satisfaction since 1992 is both worrying, and worth much more study.

Finally, we turn to the United States. Happiness data from the General Social Survey show virtually no trend since 1972, despite G.D.P. having doubled over this period. We find this puzzling, and we really don’t have an airtight understanding of what is happening in the U.S. Equally the “happiness shortfall” isn’t that large — by 2006 perhaps another 8 percent of the population should be “very happy”, and the proportions “not too happy” or “fairly happy” should each be about 4 percent lower.

One possibility is that inequality may play a role.

Yesterday I showed that happiness seems to be related to the log of a person’s income level. If this is right, then average levels of happiness would rise with average log income, whereas our usual income numbers (shown in the second panel, below) focus on the log of average income.

Instead, the bottom panel shows rather anemic growth in average log income. Thus, given that the income gains over the past few decades went mostly to those who are well off, and given that these folks get somewhat less happiness per extra dollar, then perhaps it seems reasonable not to expect much of a rise in happiness. And indeed, average happiness for those at the top of the income distribution has grown through this period.

Income and happiness trends

Our research paper contains a lot more about comparisons of happiness and income through time. The broader datasets follow the contours of the discussion above: the experience of some countries points strongly to a happiness-income link, there are definitely cases that point against it, and others leave you scratching your head.

All told, there are probably more time series suggesting that growing G.D.P. is related to growing happiness than there are suggesting the opposite. Thus we conclude that the time series evidence is both weakly supportive of a happiness-income link, and also fragile.

And even if these data don’t convince you that there is a strong connection between G.D.P. and happiness, they also shouldn’t convince you that they are unrelated.

Tomorrow, we’ll turn to asking in a bit more detail just what these happiness data are measuring.


Gregg Easterbrook of The New Republic, The Brookings Institute and ESPN (how's that for a fan plug?) wrote a great book about this exact subject call the The Progress Paradox. Just finished it on a flight last week.

Russell Beckley

What does "satisfation" mean? How do I know how happy I am? IF somebody asks me how happy I am, couldn't that affect how happy I am? Perhaps money can't buy happiness, but if you are happy just to get money, it doesn't have to. IS it possible that happiness brings money?

Dave Younskevicius

You say an increase in income inequality explains the anomaly of the U.S., I say it's because past a certain point, money doesn't bring happiness. There's a famous study that says beyond $40,000, more money does not make you much happier.

As I replied in the first post on the topic, this trend is being obscured by the constant use of logarithmic scales. Constant scales would show the trend between income and happiness levelling off much faster, to the point where in a rich nation like the U.S., the relationship between the two might be flat. Money is great when you're broke and can't afford to eat. But when you already have two cars, a house, and a decent job, more money is not necessarily going to make you a lot happier.


I know if my home had a higher income we could do things we can't do now. With bills and everything else I don't have the money to stretch into vacations or anything like that.

I stay at home and my husband works. I watch our son and run a blog, it is fairly new but I am hoping some day it might bring in some cash. i choose to be a stay at home mom which does put a financial strain on our family, but benefits my son, and that pretty much out weighs the money issues.

We are happy, but calls from bill collectors, or trying to manage where everything is going to go and will we have enough money for gas and food each month is stressful.

So yes more money = less stress = a happier me

The Baked Blogger


I think men and women have differing views on happiness. I would like to see gender broken out of the data, especially since the presence of large numbers of women in the workforce.


I saw the last graphs and was quite surprised. The second last graph, Log of (average income) is quite OK and is done regularly. But after that, the graph that you have drawn the average log of income, is problematic.

After taking the log, any addition is basically a multiplication of the original data, a division by 2 would be a square root, a division by 3 would be a cube root and so on.

So averaging after the log, has no meaning and would result in nonsensical data.


As for the Irish question: what do your date show if you use GNP in stead of GDP. In general GDP and GNP are highly correlated but in the case of Ireland GDP has outstripped GNP enormously since 1990 or so.



I think the log scales make sense, considering marginal return.

$10 matters a lot to someone with $10,000 annual income, but doesn't matter much to someone with $5,000 annual income. But $50 would matter to the person with $50,000 annual income.

Greg Flynn

In Ireland 1973-89 the economy was indeed growing but not as fast as the educated population with higher expectations than their parents. Access to the European Community and cheap air fares increased the global mobility of Irish graduates during this period. Emigration peaked in 1989 when employment bottomed out. Return migration of Irish citizens had begun increasing in 1987. I would say that in the 80s there was a wide range of satisfaction levels as the people adjusted to the new economy and investments were made in infrastructure. GDP per capita increased steadily but the population was leaking. Irish employment began increasing from 1989 and really took off in the early 90s.


Post #1 complained that Easterlin "data just didn't lend themselves to strong conclusions." Now, to imply any kind of direct correlation between income and happiness without issuing the same qualification is disingenuous.

There seems to be no controlling for availability of health care, or food, for instance. There seems to be no control of other variables that influence one's perceived place on a perceived scale other than being caused by GDP.

Also, might the Japanese 1992-2006 data and the 1972-2006 U.S. data (GDP going up, happiness staying flat) indicate there is a "satiation point"? It seems inadequate to point to growing U.S. disparity of income during the time or the growing trend of working mothers in U.S., unless that is being controlled in data from other countries where "happiness" rises with "GDP."

Also, how sure are you that lower income people are fairly represented in the happiness survey across the globe? Interesting how when you take out two poorer countries, Tanzania and Nigeria, the results are more to your liking.

I've heard ignorance is bliss, maybe wealth makes people stupid- that is not considered. Or what about the results proving the world is becoming a mono-culture of mass consuming capitalism, and income is becoming a proxy for happiness?

To repeat, to complain that Easterlin's "data just didn't lend themselves to strong conclusions" and to continue to imply any kind of direct correlation between income and happiness without issuing the same qualification is disingenuous, although maybe on par with the rest of economic theory that ignores something like environmental degradation and considers war and wasteful military spending a plus in GDP.



Thank you for this series of posts. Greatly appreciated. Am looking forward to more. This kind of material, presented for a general audience in plain English but with links to more technical material, is the best of the internet and is, I hope, a model for the way knowledge should be presented.

To explain, as you know, most academic writing is hideous. Some is just bad writing. Some is overuse of jargon that narrows the intent and meaning to a specific academic or technical point. And some is jargonized obfuscation designed to make work seem more important.

Increasing specialization and its attendant sub-languages make communication across fields more difficult. The internet, with its multi-media and links, can allow material to be understood across boundaries while preserving its depth. The difference is between an accessible but rich text and the sad reality that general or survey materials tend toward the superficial while the technical becomes incomprehensible.



If the question is: is the income effect on happyness from relative or f(absolute) changes in income - has anyone looked at other measures of income (e.g. median) and/or constrasted these changes over changes in income inequality (Gini). If relative changes are key, increasing inequity with no change to average income would be an interesing test.

Stephen de las Heras

Does having more candy make kids happier? Probably. For most kids there is no such thing as having too much candy.

But just like people's environmental awareness changes over time, so can people's materialistic awareness. Less really is more sometimes, and many people are re-discovering this simple truth. This leads to an apples and oranges comparison between people with vastly different priorities in life.


I think this is all very interesting. With respect to the peculiar case of the U.S., I think it is a strong possibility that we would have to analyze variance with various social factors given differentiation in social space. What are the factors that enable or constraint the degree to which income expands the decision space? What is the political situation in Belgium?

I will be especially interested in looking at how these questions are asked and how satisfaction is measured, as this becomes particularly difficult when we are dealing with cross-cultural comparisons (even cross-class comparisons are to some degree incommensurate without sufficient controls).


Mo money mo problems.


[ctrl. F] "Spain"... no wait!! where is Spain!! Why is Greece there instead??...


Correlation is not causation.

There's a lot more going on in the development of postwar Japan than just the GDP increasing (and happiness also). Very major cultural shifts and such.

Looking at the data, I think it's very hard to claim that per capita GDP growth equates with increased happiness.

Greece I think is the best counterexample, with the GDP staying flat for many years, but the happiness levels rising and falling in place.

Also, this data only looks at anomolies from the countries' norm happiness. I think it would be much more illuminating to look at absolute data instead. Belgium, for example, might be an exceptionally happy country, but declining more toward average over time, while the GDP rises.

This data doesn't really help with any sort of conclusion at all, not even a "weakly supportive" one.


Sorry for my ignorance, but shouldn't we be looking at median income rather than average or GDP/capita (another kind of average)? That would get around the inequality problem to some degree, no? If inequality is rising (as I've read, at least in the US), doesn't that mean that income distribution is becoming more skewed, which makes the mean a poor measure of central tendency?

Jim Birch

Here's an interesting talk on happiness that for sheds light on some of the difficulties with different happiness measures.