Bryan Caplan  

Money and Happiness: Double-Check With the GSS

Contrarian Like a Conformist... Health Care in California: Sic...

Arnold and I have a running debate on the connection between material wealth and happiness. He's skeptical of the whole subject; I'm not. He thinks that people's behavior shows that money brings happiness; I've claimed that the standard conclusion that money doesn't bring much extra happiness is introspectively plausible.

Nevertheless, I'll admit that I'm a bit weird, so I'm especially eager to double-check my introspection against a more representative sample. Not so eager, of course, that I'll track down a lot of data and replicate existing research. But fortunately, the General Social Survey website lets us do statistical analysis on the cheap.

I'm frankly surprised at what I found. In the GSS, the estimated effect of income on happiness is several times bigger than I would have guessed. Admittedly, my statistics cut a few corners. For example, the GSS gives people who refuse to state their income the highest possible coding, rather than excluding them from estimation, and I don't bother to fix this. But if anything, this ought to obscure the happiness-income link.

Quick background: The GSS contains the variable HAPPY, which has three possible responses: 1="very happy"; 2="pretty happy"; 3="not too happy." Let's call each of these categories "a step."

Basic finding:

Without any control variables, moving from the lowest to the highest 1998 income category makes you about half a step happier.

Advanced findings:

  • Controlling for education and IQ does not reduce the effect of income. (Education slightly increases happiness, and IQ matters not).

  • Controlling for the happiness of your marriage drastically slashes the t-stat on income, but mainly because the number of observations shrinks. The coefficient still implies that moving from the lowest to the highest income category makes you about a quarter step happier. (By way of comparison, moving from a "not too happy" to a "very happy" marriage makes you almost a full step happier).

  • Controlling for both the happiness of your marriage and your job satisfaction, moving from the lowest to the highest income category makes you about a fifth of a step happier. (By way of comparison, moving from "very dissatisfied" with your job to "very satisfied" makes you about a half step happier).

    Bottom line: Either the GSS results are atypical, or the literature surveys I've read have understated the joy of income. Income isn't as important to happiness as having a good marriage or a job you like, but even in a rich country like the U.S., people who earn more feel noticeably better about their lives.

  • Comments and Sharing

    COMMENTS (14 to date)
    Carl writes:

    I thought that the major issue with income and happiness was the importance of *relative* income, i.e. sure an individual can become happier by increasing their relative income (which is a proxy for social status) but economic growth cannot increase overall happiness by lifting all boats.

    dsquared writes:

    But surely you've just discovered a connection between relative income and happiness, which nobody (least of all the wealth-doesn't-make-you-happy crowd) has ever denied. The controversial assertion which Arnold backs is that absolute increases in income make you happier.

    conchis writes:

    the common finding from this sort of analysis in the literature is that the relationship between happiness and income massively declines or disappears when you look over time rather than in the cross section... leading credence to either the relative income hypothesis, or alternatively, the possibility that we adapt relatively quickly to new levels of consumption. I remember seeing a paper somewhere (I think by Ruut Veenhoven) that tried to separate out the different effects effects of absolute and relative income and adaptation... but canĀ“t remember exactly what it came out with.

    Mr. Econotarian writes:

    The great confounder may be that being more educated may allow you to have higher income and a better job, and thus potentially a better marriage (clear for men, unclear for women, perhaps you argue about money less either way).

    Timothy writes:

    I think job satisfaction might be pretty correllated with income: better jobs pay more, you move up and you're more satisfied. The income effect may be understated due to this.

    Helmut writes:

    What you neglect is that high income is a desirable trait in a mate. Men with higher relative incomes tend to retain more attractive wives. Thus the two factors are linked, no?

    Will Wilkinson writes:

    Bryan, Everyone accepts that within a society higher income correlates positively with higher self-reported happiness. Indeed, moving from the top to the bottom of the distribution has a huge effect. The trick, sometimes called the "Easterlin paradox" (after Richard Easterlin), is that the income/happiness correlation doesn't hold between societies or within a society over time. The main explanations, as conchis mentions, are (1) the importance of relative income to happiness and (2) adaptation.

    Asbolute income has a very strong positive effect up to around 10K, at which point, the principle of diminishing marginal utility kicks in pretty hard. A number of folks argue that this means that absolute income doesn't matter after this threshold has been crossed. But it would be pretty incredible if more money didn't buy you anything, even if adaptation effects are strong. And, it turns out, if extra income is spent on stuff that we don't easily or swiftly adapt to, then extra money does buy some extra happiness. So absolute income does matter, but not a lot.

    Adam writes:

    Others have commented on the role of relative income in happiness, but there's another potential bias here. We know that increases in income do create considerable happiness for a time and decreases in income cause temporary decreases in happiness. Might the higher income groups be over-populated with people who have had a transitory increase in income (about which they are happy) and the lower income groups with people who have had a transitory decrease in income?
    In other words (ignoring the role of income relative to a peer group) might happiness be determined by transitory, rather than permanent, income?

    Will Wilkinson writes:

    Adam, I doubt there's nearly enough income mobility, up and down, to get your hypothesis off the ground.

    Constant writes:

    The interest in reported happiness leaves me puzzled. I almost get the impression from writers about this subject that they think people making choices should be, and are, aiming to increase their long term reportable level of happiness. I get the impression that they equate reportable happiness with being better off. However, that seems to me to be wrongheaded. Happiness and unhappiness seem to me to be states of emotion which have a regulatory function, analogous to the function of the sense of balance. If we start tilting to the left or to the right, then our sense of balance makes us feel unbalanced and we right ourselves. There is, then, a normal state of being in balance. Similarly, there is a normal balance between happiness and unhappiness which the human animal needs at some point to return to, because happiness and unhappiness perform a regulatory function and in order to perform it well they can't continually be out of balance. We can't continually be super-delighted, nor can we continually be super-depressed, without adversely impacting the regulatory function of happiness and unhappiness.

    Even though we pursue happiness, in the end, for our own well-being, the level of happiness must return to normal. The honeymoon must end. And it does. The body provides us with rewards, which give us a temporary happiness. When we run or each hot chili, endorphins are released in our body and we get a feeling of well-being, of happiness. But the feeling fades. Our body for some reason, probably for a good reason, does not permanently reward us. People who artificially prolong the reward by means of opium or other drugs, ultimately pay a price.

    People who are too sad for a prolonged period become a danger to themselves and are in need of help. Similarly, people who are too happy for a prolonged period of time may also have problems, though prolonged sadness seems to be a much more common condition than prolonged happiness. Apparently the body needs balance.

    So, in short, preference, even strong preference, for A over B does not mean that someone who has A will, for more than a brief period, be reportably happier than he would have been with B. In the long term, he might be exactly equally happy either with A or with B. This poses a puzzle only if we believe that the point of having A is that it will increase our happiness. But a person can prefer, strongly prefer, A to B regardless of whether A makes him reportably happier over the long term. A can make him much better off without making him one little bit happier. Happiness is not the ultimate goal.

    Rather, it's the other way around: first of all, we get our goals, and then, when we reach them, we feel happy about it (briefly). Happiness is not the purpose of the goal; rather, the goal is the purpose of the happiness. Happiness is like a party that we put on to celebrate reaching an important goal. The purpose of the goal is not to have the party - we could have a party any time we like. The purpose of the party is to celebrate the goal. Happiness is like a little party in the head.

    Suppose you want something. You want it really, really bad. Then, finally, you get it. What happens in your head is that you're filled with so much joy that maybe you'll dance.

    Well, was the point of it all to feel that joy and do that dance? Not typically, it wasn't.

    Steve writes:

    Uhm, if you make enough money to be secure in your home, your health and your material posessions, wouldn't that allow for money=happiness. I know if I lived day-to-day, never knowing if my income was steady, or knowing that if something of major financial investment were to occur randomly and that I would lose my home, health, or significant material possessions I would certainly live in a less happy state than I currently do.

    the saying that money does not give way to happiness refers to excess wealth, wealth beyond security in home and health.

    gary lammert writes:

    As of 9 August 2005, the growth fractals for the composite American equities, which represent the sizeable fraction of global equity worth, has - with reasonable probability - been completed with respect to a maximal valuation secondary to the highs of March 2000. Telltale at this near global US equity summit point is the very contrasting new 35 year low for venerable Delta Airlines which closed Tuesday below 2 for the first time in 3 and 1/2 decades, down over 94 percent from its high of over 70, 5 years ago.

    Bank stocks are sinking, having received and continuing to receive broadsides from both the Fed's increasing interest rates and the narrowing short term-long term spreads. The interest rate dependent real estate stocks, e.g., HGX, are breaking lower trend lines. It is over.

    The final blow-off for the Nikkei and Euro stocks very likely occurred today with exhaustion gaps to yet another multiyear high. The Euro equity finale is contained in the third fractal of the 12/31/31 day sequence. The last 31 day sequence, which parallels US equity composites:

    6/15/10 to the top (a Fibonacci number 1.6 x the base of 6) with today day 15 of 15 for a terminal sequence of 6/15/15 days.

    An averaged Fibonacci growth sequence dating from August 2004 was completed today for the US equities at a lower short term high. Market sentiment is, by contrarian standards, appropriately at its most optimistic apogee. Fractally, contained within the valuation summits, the great American equity market composites have already provided evidence of a primary decay pattern. Expect the unexpected. Expect nonlinearity.

    -Gary Lammert "

    Robert writes:

    In other words (ignoring the role of income relative to a peer group) might happiness be determined by transitory, rather than permanent, income?

    Maybe, but on what time scale. One might make a good case for happiness based on one's standard of living comparing favorably to one's parents' standard of living (when they were your age), but that seems like a longer timescale than you were suggesting.

    Paul N writes:

    The key point in this discussion, to me, is one that Will already made: Bryan is mistaking correlation (richer people in a group are happier, which is true) with causation (getting more money makes you happier, which is not true).

    I find the idea that people should naturally maximize happiness completely bizarre. Let's think about natural selection - would we expect evolution to have favored increasingly happy people? Uniform happiness is about the most dangerous characteristic for a population to have, from an evolutionary point of view.

    Comments for this entry have been closed
    Return to top