Interaction Between Earth and Societal Systems
The final session of the workshop before a plenary discussion was chaired by steering committee member Terry Chapin (University of Alaska), with presentations by Lisa Berkman (Harvard University) and Brian O’Neill (National Center for Atmospheric Research).
Lisa Berkman, Director,
Harvard Center for Population and Development Studies,
Harvard University
Lisa Berkman explained she would focus on public health issues, and include the perspectives of demographers, epidemiologists, and economists to discuss four statements:
- A framework is needed for determinants of population health—the Rose paradigm.
- Demographic distribution of populations matters more than average populations when considering the global future of population health.
- Social and economic determinants are important in shaping patterns of population health. Heterogeneity and inequality are both
-
important, for health as well as consumption (as discussed in previous talks).
- A governance structure is needed to promote sustainability.
Berkman argued that when considering the drivers for climate change and world health, responsibility rests with rich nations. While population is an important issue, consumption is equally important, and consumption is clearly the purview of the wealthier nations.
Berkman then described Geoffrey Rose’s contributions to population health research (see Box 7-1). She explained that under the Rose paradigm, everyone in a population changes in the same way, so in response to a stimulus, the whole population would be pushed in one direction. However, this tends not to be true; instead, the distribution can change its shape and size, which Rose did not consider and is outside his paradigm. Berkman used the example of body mass index (BMI) for women in developing nations (Razak et al., 2013). In Bangladesh and Bolivia, the number of women with low BMI (in other words, malnourished women) did not change appreciably over time. However, the number of women with high BMI increased. Berkman referred to this as a “double burden”—now these countries are experiencing difficulty with both underweight and over-
BOX 7-1
Understanding the Rose Paradigm
In his 1985 work Sick Individuals and Sick Populations, Geoffrey Rose studied blood pressure in two populations, a group of Kenyan nomads and a group of London civil servants. He found that the distribution curve was the same in both populations, but the average systolic value was about 10 points higher for the London population. The presumption is that over time and with such drivers as economic development, the Kenyan curve will creep to a higher and higher systolic average until it overlaps with the London population. Rose postulated that, if a mean moves in a certain direction, the deviating tail of the distribution will correspondingly move in that same direction. Rose stated, “Distributions of health-related characteristics move up and down as a whole: the frequency of ‘cases’ can be understood only in the context of the population’s characteristics…The population thus carries a collective responsibility for its own health and well-being, including that of its deviants” (Rose and Day, 1990, p. 1031). The paradigm infers that intervention strategies, such as sanitation, fluoridation, or seat belts, should push the whole population across its demographic distribution to be healthier in an equal way.
SOURCE: Data from Rose (1985) and Rose and Day (1990).
weight populations. Berkman pointed out that the ideal scenario would be one in which those with low BMI would gain weight and those with high BMI would lose weight. Even the ideal scenario is an asymmetric one, inconsistent with a Rose shift. Berkman said that this demonstrates how a single variable’s mean or average does not represent what is happening in a country. Looking only at average BMI in Bangladesh, for instance, would show an average increase, with no sensitivity to the fact that there is still an undernourished population. A more sensitive measure would provide information about both the center of the population and its extremes.
Berkman then examined social disparities within countries. She showed data that indicate that overall mortality rates have been dropping over time. Cancer rates in the United States from 1950–1990, however, show the drop is not consistent across socioeconomic status. Populations with the highest socioeconomic status show a drop in cancer-related mortality over time, but the lowest quintile shows an overall increase in mortality in the same time period. The reason, Berkman said, is simple: smoking. A reduction in smoking caused cancer rates to plummet in certain income classes. She referred to this as “exporting failure”: those who are the best off learn that a behavior is bad and they stop the behavior, but less well-off populations take up that behavior. Exporting failure takes place across income classes within a nation, and across nations—from the wealthiest nations to the poorer nations. Negative behaviors that have been exported include smoking, dietary behaviors that lead to obesity, and high rates of energy consumption. As a result, she said, inequality is widening across the world.
Berkman closed with a discussion of governance requirements and the concept of legitimate coercion developed by Mansbridge (2013). She acknowledged that this area is not her own field of expertise, but said that it is an important topic to discuss. She pointed out that the government coerces its populations all the time. While illegitimate coercion can be recognized, what does legitimate coercion look like, she asked. How can large, highly interdependent structures produce sufficient legitimate coercion to solve collective action problems? She said that legitimate coercion can be normative (i.e., people believe it is right) or empirical (i.e., people see evidence that it is right). She said that climate change is an example where legitimate coercion may be critical. Berkman suggested that experts in collective action problems be brought into the discussion to see how the idea of governance structures could be addressed.
As a final thought, Berkman posited that health is socially patterned, and that social disparities exist in health. While these disparities are ubiquitous, they are also variable and have numerous possible health ramifications. The challenge, she said, is to find social and economic poli-
cies that influence large-scale health risks so that population health might be improved, disparities reduced, and sustainable societies and global environments created.
DEMOGRAPHY AND CLIMATE CHANGE: CURRENT UNDERSTANDING, FUTURE DIRECTIONS
Brian O’Neill, Scientist III,
National Center for Atmospheric Research (NCAR)
Brian O’Neill prefaced his presentation with a discussion of recent events in Boulder, Colorado, the location of NCAR. The area recently flooded; initial estimates say it was a 1-in-1,000-year rain and a 1-in-500-year flood—in other words, a rare and unexpected event. There were eight flood-related deaths in the area, along with an estimated $1 billion in damages. O’Neill said the region has begun a discussion of how to measure the impact of such an event on society. The losses of life and property were very important in the local community. However, overall life expectancy rates did not change in the area as a result of eight deaths. O’Neill said he raised this issue to show that, in a broader discussion of measuring the impact of climate change, using life expectancy as a primary metric is setting too high a bar. Something short of a catastrophic scenario is still part of the collective concern. The Colorado flooding also provided an example of differential vulnerability. In many cases, different neighborhoods received the same level of hazard but the results were markedly different: A mobile home community would be destroyed, while a high-income housing development would only experience wet basements. O’Neill pointed out that, when assessing the scope of the impact of climate change, it is important to account for the condition of different populations, in addition to simply the level of physical hazard.
O’Neill then described work on the relationship between demography and climate change, structured around the determinants of risk, which results from the interaction of a physical hazard (storm, flood, heat wave, etc.) and the exposure to and vulnerability of a population or ecosystems. Figure 7-1 shows a schematic of how aggregate risk is determined for natural disasters, and risks for climate change can be conceptualized in a similar manner.
O’Neill then summarized the Stochastic Impacts by Regression on Population, Affluence and Technology (STIRPAT) model. This model looks at historical data to assess the importance of demographic, economic, and technological factors on the emissions that cause climate change. It essentially converts the IPAT model (see summary of Turner’s talk, Chapter 2) into a statistical model, where P, A, and T are not assumed to be linearly
FIGURE 7-1 Schematically key concepts involved in disaster risk management and climate change adaptation, and the interaction of these with sustainable development.
SOURCE: O’Neill presentation, slide 2. Available: http://ipcc-wg2.gov/SREX/report/report-graphics/ch1-figures/ [February 2014].
independent variables. When estimated using national-level historical data, the STIRPAT model has shown that, on average, population is proportional to emissions, although this relationship varies across different sub-groups (for example, when population is grouped by income). Other demographic variables (such as age and urbanization) can also be statistically significant. Another approach to studying these relationships uses structural models that include causal relationships. For example, aging and urbanization in such models can lead to different levels of labor productivity and consumption behavior, which then lead to different levels of economic growth and production structure, which in turn influence emissions. Education can also influence emissions through its links to fertility (see Lutz’s discussion of maternal education, Chapter 6) as well as to its links to consumption, labor productivity, and innovation capacity. However, O’Neill said he finds the innovation variable to be rather complex, and he did not include it in his analysis, although other models do consider innovation capacity and technological change.
O’Neill then discussed the integrated Population-Economy-Technology-Science (iPETS) model, a global economic model that allows for heterogeneity within regions (similar to models discussed by Edmonds). Within regions, iPETS simulates household decisions about
consumption, industry decisions about production, overall land use, and market equilibrium—all factors that affect carbon emissions. The model is then given three country-specific population scenarios determined by the United Nations’ high-, medium-, and low-population projections. The results show large population effects in carbon emissions (O’Neill et al., 2010). O’Neill stated that to limit emissions to a 2-degree Celsius shift, current emissions would need to be cut 80 percent by 2050, a reduction that is not plausible through demographic changes alone.
O’Neill also examined how education affects the Human Development Index (a metric also discussed by Lutz, above). He found a fairly substantial increase in the Human Development Index with increased education level.
O’Neill then showed data related to the spatial population distribution in the United States. Projections of the U.S. population distribution in the year 2100 show a more concentrated picture of population than at present, with more people living along the coasts. For instance, it is projected that Florida will double its coastal population by 2100 even though coastal populations are more vulnerable to climate change effects. (O’Neill noted that this was the NCAR projection; an International Institute for Applied Systems Analysis [IIASA] model showed little change, while an Environmental Protection Agency [EPA] model showed a nearly threefold increase in coastal population.) O’Neill later clarified that these numbers did not include any projected changes to the location of the coastline itself due to rising sea levels. He said that the United States coastline is unlikely to change much by 2100, except under a high sea-level rise scenario. He noted that high sea-level rise is very likely to happen at some point and substantially affect the Florida coast, but the timing of that rise is very uncertain.
Turning to the data related to human exposure to heat, O’Neill noted the study looked at the number of days per person spent above 35 degrees Celsius. Increases in this heat exposure variable can be the result either of increased temperature or increased population in hot areas. Heat exposure days for the United States as a whole increased anywhere from 50–100 percent, depending upon the population projection used, with many regional effects. California, for instance, will experience many more heat exposure days in 2050, mostly attributable to the expected increase in population by that time.
O’Neill closed his presentation by summarizing his main point: The net effect of population growth is an increase in energy use and emissions. He argued that demographic characteristics such as urbanization and aging do matter, but their impact varies in different parts of the world. He said that education has a strong impact on well-being (i.e., the Human Development Index), although the impact on emissions is not
terribly significant. He briefly mentioned that the spatial distribution of population can strongly affect a population’s exposure to climate-related hazards, and development pathways have a strong influence on climate change risks. He noted he included some demographic interactions in his models, but other potentially important ones may still need to be included, such as the importance of women in the workforce, the details of urbanization, and lengthened time to retirement. It is important to include spatial variation in population densities as well as demographic characteristics (such as age, income, and education) that vary spatially, in seeking to explain population correlates of climate change and vulnerability. Finally, he suggested, it would be important to identify the most important determinants of vulnerability, because currently the relative importance of a population’s characteristics (such as income, education, spatial distribution, levels of inequality, and more) in responding to specific hazards is still unknown.
A participant postulated that the danger that climate change poses to human well-being is not well understood by the public. Public statements made by some people that climate change will cause millions of deaths are contradicted by other people who say climate change will cause no deaths. It becomes difficult for the public to understand the magnitude of this problem. O’Neill pointed out that the importance of climate change should not be calibrated from public statements of just anyone. He acknowledged that the science of climate change has not been communicated effectively enough, although he said the IPCC data contain the most reasonable assessment of current status and future projections. He pointed out that the risk of climate change is highest for unique and threatened systems such as fragile ecosystems and species, some Arctic cultures, and small island states. These are likely to be severely impacted even by a 2-degree increase in temperature. However, there are not likely to be large-scale health consequences or disasters in a 2-degree change, except in a truly extreme scenario.
A participant then asked about the patterns of changing BMI, asking why some of the developing world is following the U.S. pattern, while other countries are not. Berkman responded that many countries (notably India) have neglected those with the lowest socioeconomic status in favor of supporting the rising middle class. This is a country-level policy issue, and she suggested national policy is likely the reason for different patterns.
Another participant asked for clarification on urbanization and its relationship to emissions. O’Neill explained that urbanization leads to an
increase in labor productivity. While consumption patterns change and geographic location changes, the urbanization impact on emissions is dominated by labor productivity increases.
The difference between the tail and the bottom of a population distribution was questioned. Berkman suggested a concentration on high-risk populations (the bottom one-third of the socioeconomic status) rather than thinking about how to reduce inequality. She said it is likely more advantageous to change the shape of the distribution rather than move the distribution as a whole.
