The Center for Disease Control’s Behavioral Risk Factor Surveillance System, or BRFSS (http://www.cdc.gov/brfss/), is old -style big data: over 400,000 telephone surveys each year focused primarily on the physical health of America. Following the pioneering work by Andrew Oswald and Stephen Wu (http://ftp.iza.org/dp4600.pdf), economists have begun using the BRFSS for its questions on life satisfaction or “happiness.” This big health survey is the only publicly available data source with enough data to be able to estimate reliably average levels of life satisfaction at the metropolitan area level.
Metropolitan area-level life satisfaction in the BRFSS is related to a number of metropolitan variables, but some of these relationships are more stable than others (http://www.nber.org/papers/w20291). For example, more educated metropolitan areas appear to be happier, but that result largely disappears when we control fully for the income and education of the individual respondents. Life satisfaction is actually slightly higher in more unequal areas in the BRFSS, which is exactly the opposite of the strong negative link between happiness and inequality that shows up in the General Social Survey (http://www.nber.org/papers/w14419).
Segregated metropolitan areas are less happy; and African Americans are particularly unhappy in more segregated areas. This result is intuitive, and also deserves more study.
The single most robust fact, however, that comes out of the BRFSS is the strong link between urban decline and unhappiness. This result holds with or without individual controls. It is robust controlling for a wide range of other metropolitan area level variables. The relationship is concave. Places with fast growth are not much happier than places with modest growth, but there is a big happiness gap between places with modest growth and places with no growth.
Are some metropolitan areas happier primarily because they attract happier people? Are declining cities unhappy primarily because happier people tend to be the first to leave?
There is little evidence to support the hypothesis that differences in metropolitan area-level life satisfaction primarily reflect the selection of happier people into happier cities. There is one panel data set—the National Survey of Families and Households (NSFH)—that contains repeated questions about happiness for people who move across areas. When we estimate metropolitan area-level happiness effects using movers only (essentially estimating the change in happiness associated with moving into a particular area), these estimates line up well with the estimates that use movers and stayers. Moreover, there seems to be little evidence of happier people disproportionately leaving declining areas or unhappy people disproportionately moving into those areas.
If the correlation between unhappiness and urban decline isn’t just selective migration, there still remains a core question. Are declining cities unhappy because they are declining, or are they declining because they are unhappy? If decline causes unhappiness, then perhaps the federal government should do more to fight population loss in the Rust Belt. If people are understandably fleeing miserable places, then it would seem somewhat cruel for the government to bribe them to stay.
To address this key question, we went back in time. The General Social Survey has data on life satisfaction going back to the 1970s. The data enables us to ask whether the gap in happiness—between declining and growing cities—is rising or falling. Over the whole time period, we define decline based on population growth from 1950 to 2000. We then estimate whether this measure of decline is more strongly correlated with unhappiness in the 1970s, when urban decline was just beginning, or in the last decade, when urban decline had been progressing in many places for decades.
We find a strong pattern indicating that the link between decline and unhappiness has gotten weaker, not stronger, over time. The declining cities of the U.S. were particularly unhappy in the 1970s. They have gotten relatively less unhappy since then. These results do not suggest that the unhappiness declines have disappeared, but rather that they were more extreme in the past than they are today.
We supplemented the General Social Survey data with Gallup life satisfaction data from the 1940s. We lack detailed information on location in this data. We only know the size of a respondent’s city, not their particular metropolitan area. Moreover, almost all residents of large cities live in the Midwest and Northeast in the Gallup data. As a result, we cannot meaningfully distinguish between declining and growing cities. However, eight of the ten largest cities in the U.S. in 1950 have lost more than 20 percent of their population since then. We are therefore willing to take city size as a proxy for subsequent decline.
We find that in the 1940s, the residents of large cities reported substantially less happiness than residents of less dense areas. This evidence also supports the view that unhappiness preceded decline. Interestingly, there is no negative correlation between metropolitan area size and unhappiness in the modern data, although the residents of one large metropolitan area (New York City) are particularly prone to report low levels of life satisfaction. Cultural norms may also play a role in that particular data point.
A simple story can potentially make sense of these patterns in the data. Historically, America’s cities were not particularly pleasant places. Men worked in unpleasant factories or mines. Women toiled in sweatshops and feared for the health of their infants. The old metropolises were not built to provide pleasure to their ordinary citizens. Yet people put up with urban unhappiness because cities brought other advantages—most notably high incomes, which meant a brighter future for their children. Cities that declined between 1950 and 2010, like Detroit, paid their workers particularly well in 1950.
Over the last 60 years, the erstwhile advantages of those old manufacturing cities disappeared, as transportation costs plummeted. Factories relocated and people moved. There was a transition from areas in which firms had a productive advantage to places that were pleasant for people to live. Sunny living in Los Angeles became a new urban paradigm. Producer cities were replaced by consumer cities.
Yet some people still remain in the old industrial cities of the Midwest. They are still unhappy places, but people live there, perhaps because of family ties or low housing costs. Yet the happiness gap seems to be declining over time, which perhaps suggests a rough equilibration.
While this narrative fits in with standard stories about America’s urban evolution, there is also a subversive element in it that relates to the standard interpretation of life satisfaction or happiness. Many happiness researchers write as if people either should or do maximize happiness. That assumption would be violated if people accepted more unhappiness in return for higher wages or lower housing costs.
Yet the view that humans either do or should maximize happiness is an odd one. Why should one particular emotion be privileged over every other objective? Anyone who expects his or her happiness to automatically adjust upwards in response to important personal achievements is likely to be disappointed. Surely, there are more than a few happiness researchers who would accept a little less happiness in exchange for making an even more significant impact to social science.
The happiness gaps across space are difficult to reconcile with the view that humans pursue happiness as a monomania. A better interpretation is that happiness is just one objective among many, and that people—now and in the past—are willing to put up with a little less happiness in exchange for other goals.