Wednesday, January 27, 2010

Some spurious regressions

This graph is one reason I don't believe in cross-country comparisons of self-reported Happiness and Life Satisfaction. It compares an international index of happiness and life satisfaction with the gini-coefficient of inequality. As you can see, the relationship between happiness and inequality is positive (barely statically significant, and clearly of the wrong sign).

I think inequality is negatively related to true well being. This relation is partially direct, and more importantly emerges as almost all social problems, be they drug use, crime, high school dropout rate, or teen pregnancy, create income inequality. As a side note this second effect leads to an endless spurious correlations as the left claims that it is income differences that is causing crime or drug use, and demand income transfers as useless remedy for deeper social pathologies.

Self-reported happiness is mainly a ratio of expectations and outcome, not an absolute measure of well-being. That is probably why the relative self-reported happiness of women has fallen, while their status has increased. They are better off, but their expectations on life have risen even faster than their status in society, so their self-reported happiness has declined.

Furthermore, reported happiness seems to depend on culture. Latin Americans report high levels of happiness, whereas for example the French give low scores (perhaps the French believe that claiming to be happy about life is low-brow). My best guess is that this odd positive relationship between inequality and happiness is created by chance, as cultural areas with high inequality (Latin America, Arab countries) tend to give high answers on self-reported happiness.

I simply cannot accept that El Salvador is better off than in France, South Africa better than off than Hong Kong, or Guatemala better off than Germany. Revealed preferences show that per capita GDP dominates indices of self-reported happiness. When people get to choose they try to move from El Salvador to the US and from Nigeria to France, as GDP predicts, but opposed to what self-reported happiness would predict.

Those on the left, who (having lost the income creating battle) now want to replace objective measures such as GDP with subjective measures such as self-reported happiness should keep these problems in mind.

7 comments:

  1. "I think inequality is negatively related to true well being."

    Could be and you provide some good reasons why this would be the case.

    I don't have a strong oppinion on this, but I would just like to note that a major factor in life satisfaction is probably a feeling of purpose, which could be increased in a inequal society that allow for different ways of living your life. (Since equality tends to be achieved by income transfers at least in the west?)

    ReplyDelete
  2. It is instructive to look at the next two words, sometimes...

    "The relationaship between happiness and inequality is ... clearly of the wrong sign. I think..."

    It's convenient, being able to disbelieve the data because I think it ought to look different. You spent your first posts demonstrating that the data do not say what Krugman et al thought it ought to look like, but when the data disagree with you, you brush it aside. Not good form.

    Better form: The graph says this. Why? Possibility 1) That's the real relationship. What would cause that? Is there good cause to think otherwise? Giving the data at least a brief benefit of the doubt enables us to learn from it.

    Possibility 2) There's a problem with how the data are collected. Maybe surveys aren't as representative in unequal societies. It is much more difficult for NGOs to reach people who earn less than $0.50/day than those earning $1/day, for instance. The more unequal the society, the less likely you are to be getting measures from the bottom of the distribution.

    Possibility 3) The survey doesn't actually measure what it purports to. (This is your first argument.)

    Possiblity 4) There are well known problems with inequality data. Your post after "Furthermore, reported happiness ..." discusses GDP and poverty, not inequality per se. So it's largely irrelevent to the graph.

    The US also has relatively high inequality, and Americans tend to be more optimistic than Europeans (I say based on 2 years in Germany listening to them complain about the optimism of the Americans they knew)

    Possibility 5) Ommitted variable bias: Cultural factors. What are they? Does the relationship disappear when you include them? If you can't do that, does the relationship disappear when you include continental dummies or some other proxy?

    As an aside, the cultural factors story is not inconsistent with the graph or the second half of your post. My culture and lifestyle has helped me to be happy. I believe I would be even happier still living in a richer country where I have greater opportunities than here.

    As another aside, even based on your interpretation, the graph does tell us something, just not what some people might interpret it as being. It tells us that outcomes/expectations are higher in more unequal societies. Could it be that people start off poorer and become wealthier over time, resulting in greater satisfaction? Could it be that while the poor in Latin America and the poor in Africa both have low expectations, outcomes are better in Latin America?

    ReplyDelete
  3. D. Watson:

    When I said it is of the wrong sign, I meant the wrong sign based on standard theory of people who believe in happiness research, not my personal beliefs. Maybe I should have be more clear.

    I *personally* have no problem at all with inequality that is caused by differences in productivity. Inequality that is caused by social problems however is a bad, although again more as a sign of deeper ills, than in out of itself. The problem of the Chicago ghetto is not that people in uptown Chicago are well off.

    ReplyDelete
  4. I don't understand how they measure those kind of things. People happiness I don't think that it is something that it can measure easily. viagra online has some great way to rate other things that I think that it is more important.

    ReplyDelete
  5. I am thoroughly convinced in this said post. I am currently searching for ways in which I could enhance my knowledge in this said topic you have posted here. It does help me a lot knowing that you have shared this information here freely. I love the way the people here interact and shared their opinions too. I would love to track your future posts pertaining to the said topic we are able to read.
    Penis Enlargement | Penis Enlargement Pills | VigRX Plus

    ReplyDelete
  6. They are better off, but their expectations on life have risen even faster than their status in society,
    thesis examples free

    ReplyDelete
  7. These information helps me consider some useful things, keep up the good work. manfaat daun salam

    ReplyDelete

Google Analytics Alternative