At the Natural Hazards Workshop this year, Daniel Aldrich gave a keynote talk about his findings from researching correlations between various variables and community recovery from disasters. In particular, he focused on his awesome empirical study of Kobe’s recovery from the 1995 earthquake. (While its not relevant to this post, he found that proxies of social capital had the greatest correlation.) In his keynote talk he used the term “community recovery” even though his independent variable was specifically population recovery. I wasn’t sure if he was using the terms interchangeably because he believes population recovery is a proxy for community recovery or if it was just easier to say “community recovery.”
So I wrote him and asked. He replied that “I emphasized population recovery because it is the most portable (across time and space) measurement of the health of a community. There has been a great deal of work by scholars like Yasui, Vale and Campanella, Davis and Weinstein, Edgington, and others who have emphasized that population recovery serves as a robust measure of recovery.”
Our conversation is indicative of the general struggle for those of us interested in community resilience to get our heads around what recovery means and how to evaluate it. Or, in this case, what it is that needs to “recover” for us to say that a community is “recovering.”
Suffice it to say, I’m not sold on the idea of aggregate population as the focus of community resilience and a meaningful means of evaluating a community’s recovery. As I wrote here, I think that ultimately community resilience is expressed by and needs to be measured as well-being. A community isn’t recovering if a community’s well-being isn’t recovering. This is true, in my opinion, regardless, of what infrastructure (built, social, or otherwise) has been restored. More specifically, simple process of population recovery, which does not measure who that population is comprised of, does not translate trends of infrastructure restoration to recovery trajectories of community well-being.
I would argue that community identity does serve to translate the dynamics of infrastructure to the dynamics of community well-being. There are many definition’s of identity, but I want to share one by the psychologist Erik Erikson who coined the term “identity crisis.” While Erikson developed the concept of “identity crisis” for the context of psycho-social development of individuals, I think his definitions can provide insight at the community scale. Of course, the concept of community is more complex than the concept of an individual–the whole being greater than the sum of the parts. At this point, its enough to say that I see a community as any group of people that somehow identify (formally or not) as “us.” This means that a community does not have to be spatially contiguous. A community can be spread across an entire county, as we saw after Hurricane Katrina. So, for example, I don’t think counting the number of people who now live in the Lower 9th Ward provides much insight into that community’s recovery. Though, yes, it does relate to the neighborhood’s recovery–a meaningful difference.
Erikson wrote that “[identity] is a subjective sense as well as an observable quality of… [internal] sameness and continuity, paired with some belief in the sameness and continuity of some shared world image. … [In achieving identity] we see emerge a unique unification of what is irreversibly given…with the open choices provided in available roles, occupational possibilities…” (Erikson, 1970, p. 51) That belief in the “sameness and continuity of some shared world image” gets at the “us” factor of community identity.
What happens if there is major disturbance to or within a community? What happens to the norms and means of a community no longer effectively promote the well-being of community members? Well, just like a middle-aged man who bought an impractical sport cars, community’s can have an identity crisis. For Erikson, a crisis in identity is a “…turning point, a crucial moment, when development must move one way or another, marshaling resources of growth, recovery, and further differentiation.” (Erikson, 1968, p. 16). (By the way, Erikson liked talking about identity crises related to childhood development, not the development of upper middle-class, balding men.)
Even if there is empirical work showing correlation between a community’s population and its health, as Aldrich suggested, I reject the usefulness of this connection. I also doubt the meaningfulness of any empirical correlation as far as disaster recovery goes. Conversely, I know that measures of community identity have been correlated with measures of well-being. Its not enough that population numbers return after a disaster to demonstrate the resilience of various communities’ well-being; there has to be some positive trend related to those communities’ identity. In fact, an upward trajectory of population in a particular location could indicate a negative trend of community identity compared to prior to a disaster. That population could be comprised of vastly different people, whereas the “us” that were previously there are struggling to maintain their identity and promote their own well-being.
A good case study illustrating the contrast between population recovery and the recovery of community identity in crisis after a disaster is given in Andrea Rees Davies‘ wonderful book Saving San Francisco. Davies weaves a story about socio-political forces deliberately and unwittingly exerted on communities’ identities. She describes and analyzes community identities with respect to gender, race, ethnicity, and class. I can’t possibly convey the complexity of Davies’ discourse; you should read her book. Instead, I ripped some demographic data from tables in her book to illustrate, in an oversimplified and incomplete way, the poor relationship between population trends and changes in community identity.
An excel file of this data is posted here. Github is weird and the link you click to download the file is “View Raw” or “Raw.” I threw together three charts from this data. You can grab the iPython Notebook for doing this analysis here.
The first chart at the top of this post shows the population of San Francisco neighborhoods in 1900, 1910, and 1920. Between 1900 and 1920 every neighborhood experienced net growth; the overall population grew exponentially in this time period. Between 1900 and 1910–the closest year that Davies presents population data for–only three neighborhoods experienced net loss: South of Market, North Beach, and Downtown. All three of these neighborhoods experienced growth the following decade, which more than made up for the prior decade’s decline. In fact, if you look at the second chart in this post below, which depicts the percent change in population, you see that the primary reason that San Francisco’s rate of population change increased between the two decades is because of the significant population growth in those three neighborhoods between 1910 and 1920.
However you slice it, San Francisco’s population recovered from the 1906 earthquake and fire. Given the geographic importance and adolescence of the city at this point in history this is far, far from surprising. Basically, the event could not overcome the pre-disaster forces already at play on San Francisco–at least with respect to overall population growth.
Okay, so the population recovered, but what about the identity of various communities in San Francisco? The third graph in this post gives some cursory insight to this question. The chart shows changes in employment, which can be seen as a proxy for class, between 1900, 1910, and 1920. (Davies includes much more insightful data on things like race and gender, but I was too lazy to rip that data.) For each neighborhood, the left bar represents 1900, the center bar represents 1910, and the right bar represents 1920. Overall, San Francisco became more white collar each decade. The overall employment rate in 1920 was less than in 1900, though not hugely so.
When you look at the neighborhood scale, things become more complicated. The biggest example is Chinatown. The makeup of the neighborhood changed significantly. The population of Chinatown didn’t change much (in spite of efforts to the contrary from several people and organizations, according to Davies). However, unemployment went way up, there were fewer white collar workers, and, according to data cited in Davies, by 1920 there were no non-Asians in Chinatown.
Clearly, the identity of Chinatown changed after the earthquake. One can argue whether this trajectory was already irrevocably in motion prior to the earthquake. But it is much more difficult to argue that increasing unemployment results in positive trajectories of community well-being and by extension that, as far as this data shows, the Chinatown community recovered (by 1920). Further, because the identity of Chinatown changed, the identity of San Francisco necessarily changed as well. As Davies writes, “Chinatown was defined by its difference from the rest of San Francisco. But the truth was that Chinatown had always been an integral part of the city’s identity.” (p. 25).
Lastly, we can take a look at the relationship between total population change and the change of class identity (with employment demographics as proxy). Looking at the fourth figure, if identity wasn’t changing, you’d expect the points relating total population change to change in working class population to trend roughly horizontally around zero. Clearly that’s not the case. Of course, regardless of there being a disaster, a flat trend may not be expected or manifested. The point is simply that one can’t assume total population can give much insight into a community’s identity and the well-being founded on this identity.
I don’t think that Daniel Aldrich is claiming that overall population has to recover for us to say a community has recovered. At least I hope not. But I do think that his point of “portability” is one thing that gets in the way of our understanding and ability to apply the concept of community resilience. In large part, whats driving our understanding at our current state of the art is simply what is easily quantifiable or what data is easily available. This is okay to an extent, but we can’t confuse the availability of numbers for representing resilience with the theoretical variables and mechanics of it, such as community identity.