Is there a correlation between the median household income level of an NYC neighborhood and its buildings’ carbon footprint?

Introduction

We often like to blame the rich (or the poor, depending on what group we feel more kinship to) for the city’s problems. This is no different when it comes to contribution to climate change. But are the rich or poor really more to blame for greenhouse gas emissions, at least on the part of buildings they inhabit?

Local Law 84 requires large buildings in NYC to report their energy and water usage annually for benchmarking purposes. The buildings compliant to the law are 50,000 sq ft or larger and in the case of city-owned buildings, 10,000 sq ft. or larger. Looking at the data reported for calendar year 2019, I wanted to explore the spatial distribution of the greenhouse gas (GHG) emissions associated with these large buildings and see if there are any correlations with census data (ACS5 2014–2018) on median household income. I also wanted to see if the outcome changes if we focus on buildings with specific use types (i.e. multifamily housing, office, schools, etc). I use greenhouse gas intensity (kg CO2e per sq. ft.) as a measure of a building’s climate efficiency. For example, high performance all-electric buildings would have a lower GHG intensity than old, leaky, fossil fuel burning buildings.

Methods and Results

First, I looked at the distribution of the buildings by greenhouse gas intensity across the boroughs and found that there is no significant difference between them (the IQR ranges overlap). The boxplot below represents 25,608 buildings and excludes 618 outliers for visualization purposes. We can conclude that while the median household income across the boroughs is certainly different ranging from $38,085 in the Bronx to $82,459 in Manhattan(ACS5 2014–2018), the greenhouse gas emission intensity of large buildings is not.

Before drilling down to the ZIP code level and pulling in the census data, let’s take a deeper look at the makeup of this dataset. I wanted to understand what building use types were responsible for the majority of the greenhouse gas emissions. From the pie chart below we an see that just 5 building use types (multifamily housing, office, K-12 school, college/university, and hospital) are responsible for almost 85% of the total emissions. The Other category includes everything from warehouses and wastewater treatment plants to prisons and restaurants and hotels. The single largest source of emissions is multifamily housing units. Multifamily housing in NYC is responsible for the annual emission of 9.86 million metric tons of CO2e which is the equivalent of over 54,000 railroad cars worth of coal burned (EPA GHG Equivalences Calculator).

Now we know that Multifamily housing in particular is of interest. We can see if there is any correlation between the income level in a certain geographic area and the greenhouse gas intensity of the multifamily housing stock. When grouped by ZIP code, from the below scatter plot we can see that there is really no relationship between household income and greenhouse gas emissions intensity.

You might wonder if this relationship looks any different if we include all the building use types. As you see below, there is still no relationship.

The conclusion we can draw from here is the GHG efficiency of buildings is roughly equal across ZIP codes when plotted against household income. But what about overall emissions, you might ask. Do wealthier ZIP codes have higher overall emissions? Maybe on account of having more high emission buildings, such as hospitals or offices?

You can see that there a slight positive relationship between median household income and overall GHG emissions. There are many ZIP codes with low overall emissions irrespective of median income. It looks like a few ZIP codes with higher emissions in the middle-income areas are responsible for the slight positive relationship. This could be attributed to the presence of college campuses and hospitals.

Conclusion and Limitations

The conclusions we can draw from this analysis are limited by the robustness of the data. The data is based on human entry and some greenhouse gas emissions might have been calculated differently depending on the building’s management. The original dataset included 26,815 unique buildings, but the removal of rows that didn’t include data for greenhouse gas emissions left us with only 25,608 buildings. Of course this is only a subsection of all the buildings in NYC. Further, we might get a more revealing relationship between income levels and building-level greenhouse gas emissions if we used a true spatial unit such as PUMAs or census tracts for the basis of comparison instead of the more convenient ZIP code.

In spite of these limitations, we can conclude that there is no strong relationship between the carbon footprint of large buildings and the median household income of the ZIP code they are located in. As residents of NYC, we must all take responsibility for our contribution to climate change, can’t blame it all on the wealthy.

For more information on how the technical analysis was performed for this post, please follow this link to the R Studio code.