Government Resource Allocation in Philadelphia
Philadelphia has one of the highest resource constraints in the United States. It is home to over 1.5 million residents, approximately 25% of who live under the poverty line. The COVID-19 pandemic has only worsened existing resource constraints, with the impacts disproportionately affecting low-income and minority residents.
I analyzed data published as part of Philly’s Open Data initiative in order to understand how equitable the Philadelphia government’s resource allocation has been. I look specifically at the allocation of educational and COVID-19 resources because those are two areas that have been salient recently given the pandemic and the lockdowns that have coincided with it. Additionally, I was interested in Philadelphia’s unique election climate, particularly its importance in deciding the outcome of Pennsylvania’s 20 swing electoral votes and its blue city/red state dynamic so I incorporated voter data in my baseline analysis.
STEP I: Baseline Precinct Analysis
To understand the relevance of elections in resource allocation, I started by analyzing the racial composition of each precinct using publicly available data regarding voter registration.
As shown in the scatter plot, Philadelphia’s precincts are extremely racially segregated. There are clusters of precincts nearing 0% and 80% POC voter registration showing that most precincts are either majority white or majority minority. Understanding the racial composition of voters in each precinct gave me a proxy for race in my future analysis. Using the racial composition of voters, instead of the overall racial composition of residents, was more beneficial to my analysis because voters elect the politicians who are allocating the resources and I was interested in determining if there was a political incentive behind resource allocation.
I also graphed voter turnout (in the 2018 Primary election) per precinct by race to see if there was a notable relationship between racial composition of a precinct and voter turnout.
The data showed that, on average, predominately white precincts (depicted by the blue dots on the scatter plot) had higher voter turnout than predominately non-white precincts. From a political standpoint, this creates a political incentive for politicians to concentrate resources in precincts with the higher voter turnout to increase their chances of reelection. Part II and III will examine if resource allocation is driven by this political incentive to please voters in high-turnout precincts (which, on average, are more likely to be white in Philadelphia).
STEP II: Analysis of Educational Resource Allocation
To analyze the resources allocated to schools in each precinct based on voter turnout, I used graduation rates as a proxy. I chose graduation rates due to a lack of publicly available data regarding school funding. Graduation rates provide an adequate proxy for funding due to the empirically-established causal relationship between increased funding and higher graduation rates.
Note: I use 25% white as the threshold for race in my analysis going forward for two reasons. First, when looking at the precincts that had public schools, there was only one that had more than 50% of a white electorate. This is largely due to the fact that many of the precincts that are over 50% white have private or charter schools and were excluded from this dataset. Second, Philadelphia overall is only 44% white. Thus, having an electorate that is 25% or more white is significant. I used this threshold in part III as well to keep the analysis constant and comparable between resources. I will refer to precincts with less than 25% of the electorate as white as “non-white precincts” and precincts with greater than 25% of electorate as “white precincts” going forward.
The scatterplot above depicts the relationship between graduation rate and racial composition per precinct. While there doesn’t appear to be a strong relationship, it is notable that all the blue precincts (with greater than 25% of registered voters identifying as white) have a graduation rate of at least 70% and most of them have a graduation rate higher than 80%. The mean graduation rate for non-white precincts is 80% and 86% for white precincts (the medians are nearly identical showing no significant skew).
All things considered, there does not seem to be a strong relationship between the racial composition of the electorate and graduation rates (which is being used as a proxy for the amount of funding its public high schools receive).
STEP III: Analysis of COVID-19 Resource Allocation
Due to supply shortages of COVID-19 testing kits and COVID-19 vaccines, many cities have come under fire for inequitable allocation of public resources. For example, in Los Angeles many of the vaccines allocated for high-risk populations were misallocated to wealthy white populations which sparked backlash last month. I wanted to see if a similar trend was playing out in Philadelphia.
I pulled data from Philly’s Open Data initiative that included the location of every free public testing facility to see if there was a relationship between the racial composition of the electorate in a precinct and the number of public testing facilities available. I specifically filtered for testing facilities that were free, open to the public, and did not require insurance to remove facilities that were not government funded from my analysis.
As depicted by the scatterplot above, on average, there appears to be more free testing facilities in non-white precincts. This indicates that government efforts to increase access to testing in the hardest hit communities (low-income communities and communities of color) has been somewhat effective. There are still precincts that are testing constrained (precincts with only one testing facility), but thankfully there are no precincts without testing facilities. This is an important step towards curbing the spread of COVID-19 among vulnerable populations.
Part B: Analysis of Vaccination Rates by Zip Code and Racial Composition
Vaccine ethics have been at the forefront of news pertaining to the vaccine rollout. As Philadelphia prepares to loosen vaccine eligibility on April 19th, I wanted to understand the current landscape of vaccination rates. To compare vaccination rates (which are reported by zip code) to racial composition by precinct, I averaged the racial compositions of all the precincts within a zip code (each precinct has a fairly similar number of residents) and graphed the two against each other.
I found the results of the frequency plot to be pleasantly surprising. Similar to the testing facilities analysis above, vaccination rates seem to be higher in zip codes with more non-white voters. The mean number of residents vaccinated is 22,757 higher for zip codes where less than 25% of voters are white (the median is 18,857). This data indicates that Philadelphia’s strict phase guidelines for vaccine rollout did benefit residents in non-white precincts.
However, there is one notable shortcoming in this analysis. I was unable to calculate the percent of residents vaccinated for each zip code (number of residents fully vaccinated/number of residents eligible for the vaccine) because it was not possible for me to determine how many residents per precinct are currently eligible for vaccination. Regardless, I think the aggregate data is still valuable because urban zip codes have roughly similar populations and the aggregate numbers still give us a good sense of where the vaccinates are being allocated.
Step IV: Discussion
Overall, the allocation of public resources in Philadelphia was more equitable than I predicted. When it comes to COVID-19 resources (free testing and vaccine access) the city appears to be making good on their promise to prioritize minority populations and hard hit zip codes. While there was some disparity that exists in educational resource allocation, the disparity is not as large as I thought it would be. Given that I was not able to control for variables like private funding, the disparity may not even be due to lack of government funding.
The Philadelphia Open Data initiative has increased transparency drastically by making all of the data that I used in this project free, easily accessible, and formatted for data analysis. I wanted to try to compare Philadelphia’s resource allocation to another city but was unable to find any other city that published data to the same extent as Philadelphia (many cities publish summary statistics which is not useful for data analysis because it does not show individual data points). I hope that other cities join Philly in making their data available to the public.
Sources of Data:
Data sets used:
- Polling Places
- Voter Turnout
- Voter Registration Counts
- School Performance: School Progress Report
- COVID-19 Vaccinations
- COVID-19 Testing Sites