Self-reported well-being as measured in surveys is the metric most often used by economists to measure happiness. This post examines the spatial distribution of happiness in U.S. metropolitan and rural areas. We’ll see which are the most and least happy areas in the country.
Furthermore, the post will look at the reasons of the geographical variation in happiness, its long-term trends, and finally some philosophizing about whether happiness and utility are the same thing.
In a recent working paper Glaser, Gottlieb and Ziv (2014) use data on self-reported well-being by metropolitan area to shed some light on the questions mentioned above. They use a survey which asked around 300,000 respondents between 2005 and 2010 about their life satisfaction among others.
First, let’s see a couple of interesting facts established by the analysis.
Variation across space. As expected, there is variation in happiness across space. The magnitude of this is relatively small, but this is just my opinion. Variation in happiness due to location stands at 4.2% the variation in happiness due to individual factors. Putting it in another perspective, a one standard deviation spatial movement is equivalent to one-third of the difference in happiness between high school and non-high school graduates; 1.8 times the difference between male-female happiness; and around half the difference between the happiness of those who earn $35-50,000 and those who earn $50-75,000 a year. The actual variation in happiness across space is shown in the map below.
You can find a complete ranking/list of U.S. metro and rural areas by happiness on Joshua Gottlieb’s website.
One weakness of the survey used is that it’s not centrally administered. Thus its implementation may vary somewhat from state to state. Controlling for state fixed effects reduces the size of the effect of location to only 1.7% of the individual variation in happiness. This is quite low, though still significantly different from 0. Moreover, this specification may be overcontrolling and restricting heterogeneity too much. So this is more like a lower bound in my interpretation.
Permanent vs. transitory effects. Here, the question of interest is whether the differences in happiness across space are due to temporary fluctuations or more long-term/permanent trends. The authors redo their analysis separately for each year in the sample (2005-10) and then compare the effects of spatial variation on self-reported well-being in each year. The idea is that any difference found is a transitory component (as these are subsequent years). They find that the transitory component is about 1.8% of individual variation in happiness, leaving the permanent component to be around 2.4% or about 60% of the spatial variance.
Urbanization and happiness. The authors find in their U.S. sample that urbanization is neither positively nor negatively associated with happiness. In a cross-country regression, they find that cities are generally happier but this result is driven by poor countries. It is important to know that in poor countries, urban areas – while by Western standards hellish – still provide many more opportunities than rural areas. This is quite likely the reason behind the result.
Second, the authors examine the relationship between urban decline and happiness. Some of the more interesting findings are discussed below.
Nonlinear relationship. The authors find that there is no robust linear relationship between population growth (as a proxy for urban decline) and self-reported well-being. They, however, find a very strong nonlinear relationship. It seems that for cities with below-median population growth, decline and happiness are strongly associated. For above-median cities, no significant relationship is found. This is illustrated nicely in the figure below. Note how at higher levels of population growth, there is no clear relationship, while at low levels of population growth there appears to be a clear positive pattern indicating that a higher level of population growth leads to higher happiness.
It seems therefore that urban decline does decrease happiness somewhat. But having an urban boom (as represented by fast population growth) will not make residents happier than just having a steady median-level population growth. The lesson is: if you want to maximize happiness it is enough to keep a city from going into a decline, there is no need for abnormal growth.
Urban disamenities and happiness. The next finding is that “urban disamenities” such as cold winters, precipitation, serious crimes per capita or pollution cannot significantly predict happiness. For crime for instance, this could be because crime is likely to be concentrated in certain subregions of a city, not affecting most residents. It is found that inequality does have a positive (albeit small) effect on happiness. So cities with higher inequality tend to be happier. I think this relationship is more likely to be nonlinear in reality, but this is just pure speculation. While one-by-one urban disamenities cannot explain happiness (except for inequality), controlling for all factors mentioned above decreases the effect of urban decline (as measured by population growth) on happiness by 40%. This means that a combination of all disamenities do explain some of the variation in self-reported well-being.
Happiness and decline in the long run. There are two main possibilities when it comes to the long-term relationship between happiness and decline. First, current unhappy cities have always been unhappy, but in the past they were compensated for this unhappiness somehow. For instance, Detroit used to have industry, and thus lots of employment opportunities. So people moved there despite being unhappy there. Once these opportunities dissipated, urban decline kicked in as an unhappy city that has little to offer cannot keep residents. Second, it is urban decline itself that caused currently unhappy cities’ lack of life satisfaction. According to this version, Detroit was happy when it had jobs, but then as the decline came it became an unhappy city.
The authors check trends in unhappiness and find support for the first hypothesis in their data. That is, cities like Detroit have always been unhappy. Matter of fact, they were even more unhappy in their heydays like the 70s (relative to other cities) than today. So it seems unhappiness in these areas is not recent, and it was not caused by urban decline per se.
Finally, the authors connect their results with economic theory. The main point worth mentioning is that according to a theory of equilibria in urban economics, if city A is happier than city B, then people will move there until it will be less happy, and perhaps city B more happy. This in the long run will lead to a happiness equilibrium so to speak. That is, all cities should be roughly equally happy. This is contradicted by the data as shown above.
To reconcile this mismatch, the authors propose that self-reported well-being is not the same as utility. That is, people/households maximize their utility when they make decisions. But this utility is not happiness. Happiness or life satisfaction is merely one component of it. If this were the case, then the results would not violate the aforementioned equilibrium theory.
There appears to be a long-standing philosophical/economic debate on whether utility and happiness are the same thing. Another interesting finding supporting the authors’ view is that parents of small children tend to be unhappy. But if people maximized happiness, then this would be terrible for our survival. I.e. it would be an evolutionary disadvantage. So parents must be compensated for child bearing via other things that they care about.
So a theoretical implication of this paper is that people sometimes sacrifice happiness in order to obtain other things such as various achievements. And therefore, we do not maximize our happiness, but rather our utility which is a different thing. But if we do not maximize happiness, then people in unhappy cities must be compensated somehow for their unhappiness. Otherwise, they’d just move away.
Indeed, the authors find that residents of unhappy cities are compensated by higher wages or lower housing costs. For instance, in the 1940s, unhappy cities were more productive and had higher wages. Similarly, around 2000 cities in decline had lower housing costs but wages comparable to those in happy cities.
To sum up, the paper documents a variation in happiness across U.S. metro and rural areas, establishes a relationship between unhappiness and urban decline suggesting that the former caused the latter and not the other way around; and lends credence to the theory that people are willing to sacrifice some happiness or life satisfaction for the right price.