For the past year, Scioto Analysis has been working with the Center for Climate integrity on a series of projects that attempt to measure the financial impact climate change will have on local governments. The first report on Ohio was released in July of last year, and just last month the newest report on Pennsylvania came out.
In total, the report projects that Pennsylvania’s local governments will have to spend over $15 billion by 2040 in order to adapt to climate change. That’s $15 billion to roughly maintain the same living conditions we have today in the face of climate change.
The report goes into a ton of detail about these costs, and there is even an excellent interactive tool our partners on this project, Resilient Analytics, put together in case you want to learn more about this project.
However, I wanted to dive deeper into one aspect of this report. How we defined the categories for our equity analysis.
Briefly, the equity analysis for this project involved identifying certain criteria that municipalities had to meet in order to get labeled as a certain equity category. For example, a municipality that has a poverty rate over 20% was labeled as high poverty.
Conceptually, we are trying to identify municipalities that are more vulnerable to climate change, that have less adaptive capacity, and that have been historically marginalized. The difficulty comes from the fact that we need to determine a specific cutoff point.
This problem comes up across all sorts of contexts. Think about how small the difference in wellbeing is for one person who earns $1 above the poverty threshold and a person who earns $1 below the poverty threshold. Small differences in experience but entirely different categories.
In most cases, we can turn to the standards set by others who have done research on this before. If there is a consensus among the community that studies these topics, then it is usually not our place to come to a different conclusion. Having consistency with past research on the same topic not only helps tie our work to the established literature, it also makes it easier to communicate the final results.
One way we defined our thresholds when there was not outside guidance was to try and make the new groups comparable to those we had already defined. In the case of our Pennsylvania analysis, we did this by making sure that roughly the same number of municipalities fell into each equity category.
This approach also ensured that we were getting enough of a sample size in each category to make reasonable inferences. Making sure that one group isn’t too small is very important to this type of analysis.
Whatever thresholds we choose to use, we need to make sure that they are well defined before we begin our analysis. It might be tempting to wait until the analysis is done and see what thresholds provide interesting results, but that is answering a different question. It would be dishonest to determine the thresholds after the fact and report comparisons between the two groups.