Saturday, May 3, 2008

Happiness is...

positively related to the log of Gross Domestic Product per head.

It's a fundamental notion in economics that expanding output of goods and services - that is, GDP - makes people happier. That is a view that has always been challenged by some individuals and philosophies, but it has always been dominant in economics and in politics, both capitalist and socialist.
In 1974, however, Richard Easterlin studied the empirical evidence from surveys that asked people how happy they were. Easterlin found the following paradox:

1) within countries rich people are happier than poor people.
2) rich countries are not happier than poor countries.
3) countries do not get happier as they get richer.

This is a finding that remains controversial to this day. One of the theories that can be drawn from it is that richer people are happier because they compare themselves to less well off people. Naturally, we compare ourselves more to people in the same country as us, which is why it is income differences within countries that have an effect. The idea that steadily increasing wealth isn't automatically making us happier - indeed it may be making us more unhappy - has become widespread, if not quite mainstream. Examples range from the Kingdom of Bhutan's commitment to "Gross National Happiness" to the concept of Affluenza, developed by psychologist Oliver James, to even the Tory party suggesting it would aim more for happiness than growth. It's certainly a commonplace notion among people I know that rates of depression are higher in rich countries, or at least the UK and US (and possible Japan and Scandinavia), than poor ones.

Recently, the economists Justin Wolfers and Betsey Stevenson have examined more up-to-date evidence on happiness. The approach they use is similar to Easterlin, relying on surveys which basically ask people to rate how happy they are. The results and discussion can be found in a series of posts on the freakonomics blog here and in a proper academic article. They're one of the most interesting things I've read this year.

Wolfers and Stevenson's conclusion is that in fact higher GDP does increase happiness, both within and between countries. Their view is that Easterlin simply didn't have the data available to make any firm conclusions. Since 1974 we have access to much more information, most notable the epic Gallup World Poll.

Plot things on a graph and you get this:

That is clearly a strong correlation between GDP and 'life satisfaction', even without looking at the proper stats (giving an extremely high correlation of 0.82).

One thing that is worth noting before moving on is that the scale for GDP is logarithmic. In other words, an extra $100 increases the happiness of someone on $500 per year more than it does someone on $5000 a year. That's common sense, and something that's accepted by economics, but it's still something that doesn't get emphasised enough.

What the logarithmic scale does show clearly is that a proportional increase in GDP has the same proportionate effect in happiness regardless of how rich you are. It doesn't matter if you are on $500 or $5000, if your income doubles, your happiness will increase by the same proportion (about a third of a unit). If this weren't the case, the best fit line would not be straight.
There is no 'saturation effect' beyond which extra wealth doesn't make people any happier. Or at least, if there is, we haven't reached it yet.

Just to be thorough, Wolfers & Stevenson show that not only are rich countries happier than poor ones, but within countries rich people are happier than poor people (here) but on the whole countries get happier as they get richer (a href="" target="_new">here). The former relationship is universal, but the latter has more problems. It holds for most European countries (but not Belgium), seems to hold (with some caveats) for Japan, but does not apply to the US. On the whole, people in the US have got marginally less happy in the last 30 years. The authors suggest that this may be due to the distribution of income - the rich have got richer, pushing up average incomes, while increasing happiness little (due to the diminishing returns mentioned above). There's more analysis of the US in the full paper, though not of poor grumpy Belgium.
This time-series data also suggests that all this is not simply the result of people comparing their income to others.

Has conventional economics been right all along on this topic? Do we romanticise poverty and the lifestyles of the poor? Figures on suicide (see next post) do suggest rich countries have particular problems, and the US is a salutary lesson not to forget diminishing returns and let the already wealthy take the most benefits. On the whole though, the empirical data suggests they are significantly happier.

(Cliffhanger ending: Or does it..........?)


Anonymous said...

Interesting post, but two notes.

Firstly, I'm no statistician, but the correlation doesn't seem strong enough to warrant your claim that "higher GDP does increase happiness". That you can be just as happy as Hong Kong (GDP 32,000) way back at GDP 16,000 - or to put it another way, just as unhappy at - seems to give the lie to that. Wouldn't a better summary be that GDP can increase happiness?

Secondly the difference doesn't seem to be huge between the countries. For starters it's a graph of average happiness. And the average of all of the values on the graph would appear to be somewhere around 5 (judging from the way the spread looks - not very scientific I know). We can see from the graph that the difference between the happiest and the least happy countries is only about a point the either side of the local average and about two points either side of the global average. If GDP was really so closely tied to wealth one would surely expect to see a few countries closer to 1 and a few closer to 10 than there are.

If anything these statistics confirm my belief that I'd much rather be happy in a poor country than miserable in a rich one. Luckily I'm happy in a rich country, so Somalis can go screw.

Only kidding.

Anonymous said...

You know, I meant 'depressed Somalis'. Otherwise the joke, which is exceedingly poor to begin with, just looks mean. Sorry.

Tim said...

Imagine how happy the people of Hong Kong would be if they had a lower GDP though.

In statistical terms, the correlation is very very high. In other words, the chances of the relationship being random are almost nil. Of course correlation doesn't prove causation - happiness could increase GDP, or they could both be increase by a third unknown factor.

However that relationship does not mean the effect is very strong. That's reflected in the fact the gradient of the line is quite flat. This means that other factors can quite easily outweigh the effect of GDP. Part of the function of a correlation test is to pick up the effects that the eye might miss when you look at a mass of dots.

In other words, high income is making the people of Hong Kong happier and increasing it would make them more happy. However other factors are dragging this happiness down and putting them out of place on the international ranking.

The fact the differences between countries don't appear that great is something I'm going to talk about a little tomorrow. Part of the problem is we don't have much of a frame of reference - so we can't say "a score of two is very very unhappy", or "a score of seven is about as happy as this well-known TV character". That aside, there are probably two statistical effects coming into play here. First is the way that large numbers tend to converge around the middle. You always get more results at the extremes if you sample smaller populations. Second is that people are well known for preferring the middling answers when asked to rate things - they're very reluctant to use the extremes of the scale. Put these two together and you have a strong bias towards the middle ground.