Why Do Some New Yorkers
Wait Longer for City Services?

These are two real complaints pulled from the city’s own records. Is the city treating rich and poor neighborhoods differently, or is something else going on?

To find out, we analyzed every closed 311 service request filed in New York City in 2024, over 3 million complaints across 177 neighborhoods. The data reveals something more specific than simple discrimination.

New York’s 177 neighborhoods, grouped by income into four quartiles, show a clear difference in 311 resolution times. The poorest quarter (Q1) waits a median of 10.8 hours for a 311 complaint to close. The wealthiest quarter (Q4) waits 3.9 hours. Nearly three times as long. Q2 and Q3 fall in between at around 7 hours, forming a staircase from the poorest to the wealthiest neighborhoods.

To look more closely at what is going on, we start with a heat map of the main variables. Two factors stand out immediately. Complaint type (0.84) and city agency (0.81) show the strongest relationships with resolution time. Filing channel is more moderate (0.35), while income, month, and population are all close to flat, which suggests that on their own they say little about how long a complaint will take to close. Complaint type and city agency also have a strong relationship with each other (0.87), and several other variables show intermediate overlaps, so much of what any single variable appears to explain may in fact be shared with the others.

To separate their contributions, we fit a sequence of nested regression models, adding one block of predictors at a time and watching how much each addition actually improves the fit.

As the plot above shows, income quartile, month, and filing channel together explain only 12.3% of the variance in resolution time. Adding city agency and complaint type lifts R² to 79.2%. Since agency, complaint type, and filing channel are correlated, we use a drop-one-block test on the full model to isolate each variable’s independent contribution. Removing complaint type drops R² by 12.6 percentage points (ΔR² = 0.126), while removing city agency drops it by 1.2 percentage points (ΔR² = 0.012), and removing filing channel drops it by just 0.05 percentage points (ΔR² = 0.0005). Complaint type is therefore almost the entire reason the gap appears in the model, with agency and channel adding little unique information.

If complaint type carries almost all of the gap, the next question is what those complaints actually are. The raw data contains roughly two hundred distinct complaint strings, which is too many to reason about directly, so we group them into four broad categories based on the nature of the underlying issue. Infrastructure covers plumbing, elevators, street conditions, electrical issues, and similar building- or street-system problems. Health & Safety includes no heat, rodent infestations, unsanitary conditions, water leaks, and other hazards that threaten habitability. Quality of Life covers noise, dirty conditions, graffiti, and similar neighborhood nuisances. Other is everything else, such as parking violations, blocked driveways, abandoned vehicles, etc.

Quality of Life and Other complaints close within a few hours regardless of income. Health & Safety and Infrastructure, however, take far longer. Health & Safety averages 61 to 71 hours across quartiles, while Infrastructure ranges from 66 hours in Q4 to 163 hours in Q1. In addition, Infrastructure splits naturally into interior housing repairs (plumbing, elevators, etc.) and public infrastructure (street conditions, sewers, etc.). Interior repairs take far longer to resolve, over 200 hours on average across quartiles, while public infrastructure is much faster at about 29 to 48 hours. As neighborhoods get poorer, interior repairs make up a larger share of infrastructure complaints, rising from about 28% in Q4 to about 68% in Q1.

The stacked composition chart indicates that the poorer the neighborhood, the larger the share of slow-resolving complaints. Health & Safety and interior housing repairs together account for 31.4% of Q1’s complaints, 19.2% in Q2, 13.8% in Q3, and just 9.6% in Q4.

We used Oaxaca-Blinder decomposition to indicate how much of the Q1-Q4 wait gap comes specifically from complaint composition. It builds a counterfactual where Q1 keeps its own within-type response speeds but takes Q4’s complaint mix, then computes Q1’s expected wait under that scenario.

Under that counterfactual, Q1’s expected wait drops to 1.7 hours, almost identical to Q4’s actual 1.5 hours. Complaint mix alone explains 97.8% of the gap between Q1 and Q4. The remaining 2.2% is the most any within-type speed difference could possibly account for.

But why do Q1 neighborhoods file so many more of these slow-resolving complaints?

Expanding the data across the calendar, the gap separates into two layers: a year-round baseline that persists every month, and a winter amplifier that temporarily makes the gap much larger.

Q4 is the fastest quartile in almost every month of the year. As the line plot shows, even in June, when all four quartiles converge into a narrow band between 2.8 and 6.4 hours, Q4’s 2.8 hours still sits below the other three. This baseline gap exists because Q1, Q2, and Q3 file far more Infrastructure and Health & Safety complaints, such as plumbing failures, elevator outages, rodent infestations, and water leaks. Those types take days or weeks to resolve regardless of season, so they keep Q1 through Q3 medians higher than Q4’s even in the calmest months. The baseline is inequitable year-round.

On top of that baseline, winter adds a dramatic spike. In January, Q1 waits 25.3 hours while Q4 still waits only 3.1. Q2 and Q3 fall in between at around 12 to 13 hours. The gap between Q1 and Q4 reaches over 22 hours in January, shrinks through spring, bottoms out in June at just 1 hour, bumps up slightly in July and August as summer heat stresses aging plumbing, and then climbs back through autumn into another winter peak. The staircase that is so visible in winter nearly flattens in summer.

So what drives the winter spike? One complaint type dominates the answer. HEAT/HOT WATER.

In December, Q1 neighborhoods filed 88.2 HEAT/HOT WATER complaints per 10,000 residents, more than four times Q4’s rate of 21.6, with Q2 and Q3 in between. In summer, heating complaints drop to near zero across all quartiles and the bars converge. Starting in October they climb again, with Q1 rising the fastest. Using heating months (October–March) versus non-heating months (April–September), this single complaint type accounts for 15.5% of all Q1 complaints in heating months, but only 2.6% in non-heating months.

To quantify this interaction, we fit a simple model on the 48 month-by-quartile cells: median resolution time ~ month + income quartile + HEAT/HOT WATER share + (income quartile × HEAT/HOT WATER share). A month-only model explains 42.3% of variation (R² = 0.423), while adding HEAT/HOT WATER share and the interaction raises fit to 94.2% (R² = 0.942). The winter spike is therefore a mix effect concentrated in lower-income neighborhoods, not a standalone calendar effect.

311 is essentially a city-level request and complaint intake system. It records issues reported by residents and routes them to the appropriate responsible parties. As the diagram below shows, complaints reach the city through two different paths depending on what kind of problem they involve.

For housing-related problems, 311 is not the first step but an escalation mechanism. Tenants are expected to contact their landlord first, and only when the landlord delays, refuses, or fails to act do they file a 311 complaint. The case is then routed to HPD, which may inspect the building, issue violations, and apply legal or administrative pressure on the landlord. 311 does not repair anything directly. Instead, it turns an ignored private request into a documented public matter the city can enforce.

For public-space problems, no private party owns the issue, so residents report directly to 311. The system then acts as a dispatcher, automatically routing each case to the relevant city agency for handling or enforcement.

This asymmetry is what makes the housing pathway so unevenly distributed across neighborhoods. In low-rent buildings, owners often calculate that the cost of replacing a failing boiler or repairing chronic plumbing leaks exceeds the marginal rent they can extract from the unit, so they defer the work, sometimes indefinitely. Tenants with little bargaining power have no way to force the repair privately, and when the landlord stops answering, 311 becomes the only remaining lever. HEAT/HOT WATER, plumbing, and pest complaints pile up in the dataset as a result. Matthew Desmond’s fieldwork in “Evicted” (2016) documents this dynamic in detail, and NYU’s Furman Center has shown repeatedly that the most serious housing deficiencies cluster in the city’s poorest ZIP codes.

Higher-income neighborhoods rarely generate the same volume of housing tickets, but not because their buildings never break. When a leak appears in an owner-occupied apartment, the owner simply calls a plumber. When it appears in a well-maintained rental, a responsive landlord fixes it to protect the rent roll and the building’s value. The problem is resolved privately, before 311 is ever dialed, so it never enters the dataset at all. Public-space complaints from wealthier areas still flow in as usual, while the housing branch carries noticeably lighter traffic. The higher complaint volume in poorer neighborhoods is, at its root, a record of failed private maintenance, where tenants have been left with no option but to escalate to the city.

Now that you’ve seen the citywide patterns, find your own corner of the map. The choropleth below colors every NYC neighborhood by median 311 resolution time. Darker means longer waits. Hover over any area to see the full picture, including income quartile, top complaint types, and how long residents typically wait for help.

Or type your ZIP code below to jump straight to your neighborhood and see where it falls in the city’s service gap.