Money Actually Does Buy Happiness, Says Study

Wednesday, January 16, 2013

 

The late Notorious B.I.G. once rapped that with more money, comes more problems, which was just another way of repeating the oft-repeated notion that money can’t buy happiness.

 

But economists at the Wharton School at the University of Pennsylvania have refuted this idea, claiming in a new research paper that, indeed, money can purchase a happier state of being.

 

In “The New Stylized Facts about Income and Subjective Well-Being,” Daniel W. Sacks, Betsey Stevenson, and Justin Wolfers say people with higher incomes reported having higher states of well-being. That includes individuals all the way up the ladder to the top 10% of earners.

 

The researchers analyzed the incomes of 122 nations, along with the results of Gallup polls on citizens’ well-being that were conducted in those countries. In doing so, they found a direct correlation between richer countries and people who are happier.

 

They also found that as countries get richer, the happiness factor rises, leaving them to conclude there is no “happiness plateau.”

 

In the U.S., however, the researchers found that income inequality acts as “a tax on happiness.” Although a nation’s economic growth generally results in overall increased happiness, unequal access to that wealth undermines across-the-board experience of improved well-being. And so, although the U.S. economy has doubled in size since the early 1970s, self-reported happiness has declined.

-Noel Brinkerhhoff, Danny Biederman

 

To Learn More:

Yes, Money Does Buy Happiness: 6 Lessons from the Newest Research on Income and Well-Being (by Derek Thompson, The Atlantic)

The New Stylized Facts about Income and Subjective Well-Being (by D.W. Sacks, B. Stevenson, and J. Wolfers, PubMed, National Center for Biotechnology Information)

Comments

robin 1 year ago
Let’s talk a closer look at the analysis of Sacks, Stevenson, and Wolfers (SSW). The precise version of the Easterlin hypothesis is “at a point in time both among and within nations, happiness varies directly with income, but over time, happiness does not increase when a country’s income increases” (Easterlin et al., 2010, PNAS). To me, a test of the Easterlin hypothesis is fairly easy: if I find a significantly positive relationship between well-being and a country’s income (measured as GDP per capita) over time, then the Easterlin hypothesis is falsified. In their latest analysis (http://www.sole-jole.org/12513.pdf), SSW analyze 7 data sets (Table 1). In 4 out of the 7 datasets they do not find a significant relationship between GDP per capita and subjective well-being with panel regressions (columns 2, 5, 6, 7; except the coefficient in column 6, row 3). So we see no significant evidence in 4 of 7 data sets, that happiness increases when a country’s income increases. My interpretation from these results would be that the Easterlin hypothesis can be rejected in 3 datasets, but it cannot be rejected in 4 datasets. In any case, this evidence is not robust enough for me to say that “economic growth and growth in well-being are clearly related” (as SSW do in their newest article in “Emotion”). Note that SSW argue that the Easterlin hypothesis should be tested in a slightly different way: if the coefficients from the within-country cross-section, the between-country cross-section, and the national time-series differ significantly, then the Easterlin hypothesis is falsified. I cannot say that SSW’s version of the test is wrong. But I am not convinced that the more straightforward way of testing the Easterlin as described above is wrong, either. When SSW use their version of testing the Easterlin hypothesis, they find that the between-country coefficient is not significantly different from the coefficient of the panel regressions in 6 of 7 data sets (the exception is column 3). However, I would definitely not subscribe to how they interpret their results in the latest “Emotion” article: “we find that those countries which enjoyed faster economic growth, on average experienced greater growth in well-being. This general conclusion holds across all the data sets we have studied”. In my opinion, this is simply not true. The result for some data sets is that there seems to be no significant relation between GDP per capita and well-being. SSW write in the “Emotion” article: “sensitive to the difference between a precise zero and a large but statistically insignificant number, we focus on quantitative comparisons, rather than statistical significance”. First, the quantitative comparisons in the 2011 working paper are all supported by significance tests. Second, if the economic significance (i.e., the magnitude of the coefficient) really matters more to SSW than the statistical significance, why are they not discussing the 4 negative coefficients in Table 1 at all? SSW state the stylized fact in the “Emotion” article that “there is no satiation point beyond which the relationship between income and well-being diminishes”. Using U.S. data, Kahneman and Deaton show in a 2010 PNAS article, that emotional well-being satiates after a certain income level. SSW know the work of Kahneman and Deaton and even cite it in their 2011 working paper. Ignoring it in the Emotion article seems not to be good scientific practice to me. SSW write in the “Emotion” article: “a regression of (standardized) well-being on income reveals that each doubling of income increases well-being by 0.34 standard deviations, identical to the cross-country gradient”. This assumes that the logarithm with base 2 is used. However, most statistics software uses the natural logarithm as default. This means that the results cannot be interpreted as a result of doubling income, but as a result of a 2.7-fold increase in income (since the base of the natural logarithm is e = 2.71…). This is a sizable difference, in my opinion. SSW do either not say which type of logarithm they are using, or they switch between types (see notes of Table 1 in the 2011 working paper). Anyway, I’m quite sure they use Stata as statistics software, which calculates the natural logarithm as default, and I would be very surprised if they would not use the natural logarithm throughout their analyses. Note also that some of the clustered standard errors in the work of SSW could be biased downward (i.e., too small, leading to a probably false inference about the existence of a relationship). This is because the theory behind the calculation of clustered standard errors works only if the number of countries (clusters) is large (approx. bigger than 50). See for example Jeffrey Wooldridge’s 2003 article in the American Economic Review. In Table 1 of SSW’s latest working paper the number of countries is below 50 in columns 3, 4, 6, and 7 of Table 1 in the 2011 working paper. Especially the significant result in column 3 should be cross-checked with methods that try to avoid the downward bias. SSW also try to reconcile their results with the latest defense of Easterlin et al. (PNAS, 2010). The last row of Table 3 finds the “strongest results”. But why is the limit 12 years and not 11 or 13? Such arbitrary choices sound suspicious, and I would like to see if this result is robust if other limits are chosen. From what SSW write in their latest “Emotion” article, one could think that Easterlin has not recognized that in the cross-section or in the between-country analysis higher income leads to higher well-being. This is simply not true. From his very first article in 1974, Easterlin has always said that higher income leads to higher well-being in the cross-section, but over time he sees no evidence for a relationship between happiness and national income. Thus, writing a sentence like “In short, cross-national comparisons show no Easterlin Paradox.” (p. 1184 in the ”Emotion” article) disregards that the paradox only exists if the cross-section result is put in contrast with the time-series result. Easterlin has clearly pointed to this issue in his 2010 PNAS article, and I wonder why SSW are imprecise in this regard. Note also that the latest two working papers of SSW have not yet been cross-checked by fellow scientists. Moreover, the latest version of the working paper I found is tagged as “preliminary and incomplete”. The analysis in SSW’s latest article in “Emotion” is partly based on their 2010 and 2011 working papers, which have not yet been published. Before somebody I can trust to be really competent in the field has not checked the work, I wouldn’t buy it as a “new and stylized fact”. This is not to say that SSW’s results are wrong, but saying that own unpublished work is a “fact” seems a bit bold in the world of science. The “panel of panels” (Table 2) seems an innovative idea to me, and the results suggest the interpretation that Easterlin hypothesis is not true on a global level. I would be curious to see what experts in the field think of the “panel of panels” approach. In any case, that might prove that on a global level, economic growth is positively related with higher well-being. Personally, I am more interested in what is true for my own country than in the general, global result. For the U.S., Stevenson and Wolfers have stated earlier that the U.S. seems to be an exception where the hypothesis cannot be rejected (2008 Brookings paper). SSW are more interested in the global result, and see the U.S. exception as “more of an interesting outlier than a key example”. I guess the reader has to make a personal decision if he/she thinks the U.S. exception is important or not. My bottom-line is: SSW, and especially Stevenson and Wolfers have proven in other previous work that they are talented and careful scientists. And I would not say that they are wrong in their income-happiness analysis. But, dear journalists, before you buy the “new facts” as facts, it could be worthwhile to be a little more patient until somebody has thoroughly scrutinized SSW’s recent work (especially the 2010 and 2011 working papers). Once the work went through a referee process and is published in a journal, you can just be a little more sure that economic growth really buys happiness. But maybe it’s not so important for journalists to wait, because if somebody finds out that SSW’s work was indeed preliminary and incomplete, then it would be worth another headline. Just my two cents worth.

Leave a comment

captcha