We have all had that moment of panic standing in front of the waste bin holding our lunch trash, facing three choices: ‘paper’, ‘plastic’, ‘landfill’, unsure of which bucket to use. If you’re like us, at this moment you have probably also wondered: How much of this trash that we so carefully sort actually gets recycled? If you’re in New York City the answer is, unfortunately, not very much. NYC produces 33 million tons of waste a year, more than any other city in the world, and only 17% of that waste gets recycled. That is far below the US average of 32% a year.

In 2013, Mayor Bill de Blasio released One New York: The Plan for a Strong and Just City. Under this plan, the city committed to a goal of sending zero waste to landfills by 2030. Since the inception of OneNYC, the city has made incremental steps towards its zero waste goal. While the total waste produced annually has since marginally reduced, the amount of waste recycled has remained relatively constant.

The cornerstone of NYC’s recycling efforts is its curbside program, which collects paper, metal, glass and plastic. There are 2,300 curbside recycling bins in the city yet only 50 percent of the contents of these bins gets recycled.

Well, just whose fault are our poor recycling rates? The city’s, ours or New Jersey’s? The truth is, it doesn’t matter. Increasing our city’s recycling rates has large scale benefits and it simply does not get the cool reputation that it deserves. In 2020, recycling and reuse activities in the US generated 681,000 jobs, $37.8 billion in wages and $5.5 billion in tax revenues. This equates to 1.17 jobs for every 1,000 tons of materials recycled. In terms of environmental sustainability, existing recycling rates in the US saved 193 million metric tons of carbon dioxide equivalent in 2018 - that’s the same as taking 42 million cars off the road in a year!

We need to be smarter about our recycling and waste initiatives, and one way to do that is to use data to examine community recycling behaviors. This approach could help us provide recommendations to city officials on how they can better allocate resources to improve citywide recycling rates.

Having to sieve through 33 million tonnes of waste to collect data, though a joy for the many rats in our city, is probably an unsettling task for most data scientists. Thankfully, the New York Sanitation Department (DSNY) has published monthly waste and recycling data since 2016. Since DSNY provides free, regularly scheduled curbside trash collection to every residence and public building in the city, they are able to measure the tonnage of waste produced in each community district. Unfortunately, if you run a commercial or industrial business, you do have to hire your own private haulers to manage your waste or risk having some furry four-legged visitors!

Besides measuring a simple breakdown of trash, DSNY also calculates two useful recycling metrics that tell us about the health of recycling in the city at any given time. The diversion rate calculates the amount of materials put in recycling bins as a fraction of total waste. In other words it tells us how much of the city’s waste is sent to recycling plants. As many of us would not like to admit, sometimes we accidentally (or not-so-accidentally) discard non-recyclable materials into the recycling bin as opposed to the landfill bin. This information is accounted for by the capture rate. The capture rate is the amount of materials actually recycled as a percentage of the total material sent to the recycling plant.

We were interested to see if these rates were uniform across NYC or, if different, were they correlated with factors such as population density, assessment value, income, educational attainment and land use. So, we took a look at the 59 community districts across all five boroughs.

Looking at this data we see that over the four year period from 2016 - 2020 average monthly diversion rates by Community District vary significantly, from a 6.8% diversion rate in Community District 1 in the Bronx (which includes Mott Haven, Melrose, and Port Morris neighborhoods) to 29.6% diversion rate in Community District 6 in Brooklyn (which includes Red Hook, Carroll Gardens, Park Slope, Gowanus, Cobble Hill and Columbia Street Waterfront District neighborhoods).

A similar spread is seen in capture rates. Again, Community District 1 in the Bronx has an average capture rate of just 22% compared to 68.8% in Queens Community District 11 (which includes the neighborhoods of Bayside, Douglaston–Little Neck, Auburndale, East Flushing, Oakland Gardens and Hollis Hills)

We thought there might be four variables that influence community recycling behaviors: income, education, population density, and land value. We were interested in these four features because we believed that they might provide insight into the reasons behind differing recycling rates across community districts. While we thought that income and education would be positive predictors of recycling rates, we wondered if population density and land value might negatively predict recycling rates.

Indeed, our analysis shows that there is a weak but statistically significant negative relationship between population density and diversion rates. This relationship is even more pronounced with capture rates, where the negative relationship is over 5 times stronger.

As expected, income and educational attainment are all positive predictors of both diversion and capture rates. Land value, on the other hand, is only a positive predictor for diversion rate; it is not a statistically significant factor for capture rates.

Most interestingly is that, when controlling for income, education, and land value, population density is no longer a statistically significant factor for diversion rates but it is still statistically significantly negatively associated with capture rates. To put it more clearly, while population density may not impact how much waste a community recycles, it does correlate with a decrease in how well the community recycles.

We often equate denser communities - and cities specifically - with more sustainable living. However, these results show that when it comes to waste management this may not always be true. It is not hard to imagine that as people live in closer quarters and higher highrises, the systems we have in place to efficiently separate recycles from trash may be overwhelmed. Perhaps a renewed vision for recycling infrastructure in cities, such as a pneumatic tube network (like the one on Roosevelt Island) for recyclables.

Furthermore, when splitting community districts into high income (more than 32% make more than $100,000) and low income (less than 32% make $100,000) and controlling for education and land value, population density is actually a positive predictor of diversion rates for high income community districts and a negative predictor of diversion rates for low income communities. What this tells us is that targeted recycling initiatives would have the highest impact by focusing on high density, low income neighborhoods.

Recently, in a bid to weather a budget shortfall due to the COVID-19 pandemic, Mayor Bill de Blasio and the City Council cut $106 million from the Department of Sanitation’s budget. With limited resources available, it becomes even more important to be able to make data-driven policy decisions to improve recycling rates. It is not sufficient to have more recycling bins - they should be placed in specific community districts in the Manhattan, Bronx and Brooklyn. It is not sufficient to have more outreach programs - they should perhaps be targeted towards high density, low income neighborhoods.

If you found this article interesting and are driven to continue as a data pioneer for the recycling industry, we would recommend looking into the following areas for extensions of this work:

(1) Commercial recycling rates This dataset only captures residential recycling rates. In dense commercial office areas like midtown, we would expect commercial waste to account for a significant portion of the neighborhood’s overall waste. Therefore developing strategies to collect this information could provide meaningful new insight.

(2) Location of recycling assets While we did not explore the spatial relationship between existing recycling assets (recycling bins, recycling centers, collection days etc.) and neighborhoods recycling rates, this could provide interesting suggestions for future policy decisions relating to placement of these assets.