How far can fare inspectors push NYC bus fare evasion down, and at what cost?
What your settings on the left produce.
What the MTA gets back each year, from fines collected and fares from riders who start paying, for every dollar it spends on the team.
The chance that any single fare evasion ends in a notice, given how many evasions your team can reach in a day.
Each panel keeps your settings but changes one thing, to see what it would take to reach your goal.
Each card keeps all your settings but one, and works out what that one would have to be.
Each line is a different total of paid rides per notice, from the ticketed rider and from spillover. Each line flattens out at a floor that no team size can get below, because agents can only ticket riders who are still evading.
The subway module will run on the same engine as the bus model, with different inputs.
This studio estimates how much NYC bus fare evasion the MTA's EAGLE team (its bus fare inspectors) could prevent, and what that would cost. Every number below comes from public MTA, New York State or press sources, or is marked as our own estimate. Research cutoff: November 4, 2026.
A rule we followed: whenever a fixed assumption had to be picked without data, we chose the value that makes enforcement look more effective or cheaper. These are tagged Favors enforcement. If the studio shows that enforcement falls short, it falls short even under generous assumptions. Where we kept a value that works the other way, because it is a real cost or the best available figure, it is tagged Works against enforcement with the reason.
| Assumption | Value | Source and notes |
|---|---|---|
| Bus fare evasion today | 47.4% | 2026 Q3, all NYC buses (local/limited 48.7%, SBS 46.7%, express 4.8%). MTA Bus Fare Evasion: Beginning 2019 (uv5h-dfhp). Method: boardings counted by automatic passenger counters, compared with paid fares (OMNY and MetroCard taps, scaled up for cash), then reduced by 1.7 for riders allowed to board without paying (the MTA says "1.7 percent"). In September 2025 the MTA stopped counting $0 split-fare transactions as paid rides and recalculated rates back to 2022 Q1, which raised them by about 1 point; the restated data was published Oct 24, 2025 (MTA dataset overview). Peak was 50.6% in 2024 Q2. This rate was measured while the EAGLE team was working. There is no reliable estimate of evasion without the team (it was rising before local-bus enforcement began in 2023), so the model treats 47.4% as the rate with today's team, and only notices beyond today's output lower it. High confidence. |
| Total bus riders per day (paying + evading) is fixed. The evasion rate only decides how many of them pay, so paid riders rise as evasion falls, up to the total. | 2,139,831 | Calculated from two MTA 2026 Q3 figures: 1,125,551 paid riders on an average day (xfre-bxip; average weekday is about 1.29M) ÷ (1 − 47.4%). MTA ridership counts only riders who pay; the dataset notes describe bus ridership as "a count of MetroCard and OMNY swipes and taps plus an estimate of cash fares paid." MTA Daily Ridership (sayj-mze2); MTA Bus Fare Evasion (uv5h-dfhp). 2026 Q3 is close to the 2026 average so far (about 1.12M across Q1 to Q3) but below 2025, which averaged 1.21M paid riders a day, about 7.6% more. A bigger rider base would mean more evaders and a bigger team needed. Favors enforcement |
| Scheduled shifts per agent per year | 260 | A 40-hour, 5-day week (exam notice 7602: "$63,066 per annum for a 40-hour work week"; agents work "various shifts, including nights, Saturdays, Sundays and holidays"). 52 × 5 = 260. Buses run 365 days, so at a 100% deployment rate about 71% of the team is on duty on an average day. The deployment-rate slider removes leave, training and other days off. |
| Supervisors per field agent | 1 per 6.46 | 52 Assistant Chiefs to 336 field agents on the September 2026 payroll, responsibility center 6119. MTA Employee Payroll (kcjb-nf3e). We assume the MTA keeps this ratio of supervisors to field agents at any team size; a much larger team might need fewer supervisors per agent. Works against enforcement |
| Supervisor salary | $108,158 | Assistant Chief rate of $51.80/hr × 2,088 hours, same payroll dataset. Benefits added at the same rate as agents. |
| Fine on a non-warning notice (current schedule) | $88.08 net | Second offense in 4 years is $100, with a $50 OMNY credit if paid on time; third and later is $150. Of fined notices in January to September 2025, about 72% were second offenses and 28% third or later (about 11,000 and 4,000; warnings were about 126,000), read by eye from the tier chart in the TAB report to MTA Finance Committee, Dec 15, 2025. All issuers, subway and bus; no split for buses alone or for the EAGLE team alone is published. Face value averages $114. The $50 credit is applied to 72% of paid $100 fines, the share of TAB payments made directly rather than by state tax-refund offset in the first half of 2025 (24,424 of 34,044; TAB report, July 22, 2025). No public data says how many pay on time, and the credit is counted as a full loss even though it is spent on fares. 0.72 × ($100 − 0.72 × $50) + 0.28 × $150 = $88.08. Medium confidence. Works against enforcement |
| Fair Fares half fines and waivers left out | not modeled | Fair Fares riders pay $50 and $75, and eligible riders who enroll before their hearing get the fine waived. No data on how many fined riders this covers. Favors enforcement |
| Tier mix stays at 2025 levels | 72% / 28% | 2025 was the first year of the four-year count, so few riders had reached the third tier. More should reach it over time, and the warning share should fall. Both would raise revenue per notice somewhat. |
| Warning share stays fixed as volume grows | slider value | A warning is only for a first offense in 4 years, so the more notices the team writes, the more of them go to people who already had one. The slider does not adjust for this. Our estimate: in 2025, 11% of fare-evasion notices (about 15,000 of 141,000, all issuers, January to September) went to people who had already received one (TAB report, Dec 15, 2025). If notices land at random, that repeat rate implies about 590,000 people receive them. At that size, today's team (about 238 notices a day) would see warnings settle around 76% once the four-year count fills, and a perfectly efficient team (about 7,065 a day) would see about 6%. So at today's pace the 89% default is close; at much higher volumes most notices would be fines, and fine revenue is understated. Rough estimate: the 11% covers subway and bus together, and notices don't land at random. Works against enforcement |
| Warnings deter as much as fines | 1 notice = 1 notice | The rides slider applies to every notice, warning or not. Favors enforcement |
| Fare recovered from each prevented evasion | $1.88 | Average fare actually collected per NYC Transit bus ride (2026 adopted budget: $617.5M farebox ÷ 328.6M rides). Lower than the $3.00 base fare because of discounts, fare caps and transfers. NYCT February Financial Plan 2026–2029. Medium confidence. |
| Every prevented evasion becomes a paid trip | 100% | Assumes no deterred evader skips the trip instead of paying. The MTA's own loss estimates assume some would. As a result, the most fare revenue the page can show is about $696M a year (every one of about 1.01 million daily evasions becomes a $1.88 paid ride). That is more than the MTA's ($315M, 2022) and the Citizens Budget Commission's ($568M, 2024) estimates of revenue lost to bus evasion. Favors enforcement |
| Deterrence stays the same however likely a notice becomes | fixed per notice | Each notice buys the same number of paid rides whether the chance of a notice is 1 in 1,000 or close to certain. In reality, a near-certain notice would likely deter more. This is why evasion has a floor that no team size can get below, and why the Odds panel can show close to a 100% chance while many evaders keep evading. Raise the rides sliders to see a stronger effect. Works against enforcement |
| Steady state, no rebound | long-run | The model reports a long-run average. Research on fare crackdowns finds evasion rebounds once inspections stop or move (Barabino, Carra & Currie, 2024 review, Journal of Public Transportation 26:100101). Favors enforcement |
| Costs left out | $0 | Overtime (median about $1K per agent in 2025), vehicles, handheld validators, training, NYPD support, and Transit Adjudication Bureau processing (about $5.5M in 2024). Favors enforcement |
| The EAGLE team's actual 2025 output | ≈195 / day | About 53,300 EAGLE team notices from January to September 2025 (273 days), from the chart of notices by issuing agency in the TAB report, Dec 15, 2025. Used for the "today" comparison and the "Match 2025 output" button. Medium confidence: read from a chart. |
| Today's team output | ≈238 / day | 336 field agents × (195 notices a day ÷ 275 agents in 2025): the 2025 rate per agent on payroll, applied to the September 2026 team. The page opens on this, and only notices above it lower evasion. If your settings write fewer, evasion stays at 47.4%, because the model can't estimate how much it would rise. |
| Field agents during 2025 | ≈275 | Average monthly count of Special Inspectors and Fare Enforcement Agents on the payroll, January to September 2025 (range 260 to 292). Our count from kcjb-nf3e, responsibility center 6119. This is headcount on payroll, not agents deployed. Used only by the "Match 2025 output" button. |
| Base fare | $3.00 | Since January 4, 2026 (was $2.90). Used only to compare with the expected penalty per evasion. |
| Slider | Baseline | Source and notes |
|---|---|---|
| Field agents | 336 | The slider stops at 4,000, about 12× today. Answers the model works out, like the team needed to reach the target, can go past it. 262 Special Inspectors + 74 Fare Enforcement Agents, September 2026, responsibility center 6119 (the unit that holds the "Dir Fare Enforcement Eagle Team" title). kcjb-nf3e. NY1 reported "350 of these unarmed agents" on Oct 22, 2026. History: about 140 (Aug 2023), more than 250 (Sept 2024), about 285 field agents (Dec 2025). High confidence. |
| Agents needed per notice | 2 | Our judgment. Agents are unarmed civilians and work in groups; a 2023 state Comptroller audit (2022-F-23) describes surges run by "a group of six MTA Special Inspectors." Not sourced. |
| Deployment rate | 82% | Your framing: 275 ÷ 336, average 2025 field headcount over September 2026 field headcount. It implies about 47 days a year of vacation, sick leave, training and other time off out of 260 scheduled shifts (18% of 260). No public source reports how many agents are deployed per shift. Not sourced. |
| Field-time utilization | 3.4% | The default is the EAGLE team's 2025 rate, so the page opens on today's team at today's pace. The "Perfectly efficient: 100%" button sets every field minute to producing a notice at the chosen pace. That can't happen in practice: agents always wait between buses, because buses and riders never arrive in an endless stream. The rest of field time goes to waiting, moving between stops, and checking riders who paid. Agents sometimes board buses and write notices while the bus is moving. The 2025 rate. In 2025 the EAGLE team wrote about 195 notices a day from about 275 agents on payroll. With 82% deployment and 5 shifts a week, about 161 agents were on duty on an average day, so each wrote about 1.2 notices per shift. At 2 agents × 5 minutes per notice, that's about 12 minutes of a 6-hour field shift: 3.4%. The dashboard recalculates this at your settings. At a 7-hour shift it's 2.9%; at 15 minutes per notice it's 10%. The "Match 2025 output" button sets utilization to that rate at your other settings. See the worked example below. |
| Hours in the field per shift | 6 | Our estimate for an 8-hour shift (exam notice 7602 specifies a 40-hour week). No public source gives field hours. Not sourced. |
| Minutes to write one notice | 5 | Your "perfectly efficient" case. Press accounts (Streetsblog) describe 5 to 15 minutes per stopped bus. No MTA figure is public. |
| Rides a ticketed rider pays for afterward | 1 ride | Your baseline: the notice buys one paid ride from the person who got it. Time labels assume 2 bus rides a weekday (10 a week, about 43 a month, 520 a year); that conversion is for reading only and doesn't change any result. The only MTA figure on deterrence: "Paid boardings increased 7% at EAGLE intervention stops" ("Update to the MTA Operating Committees," Apr 28, 2025), with no duration or spillover reported. |
| Other evaders who start paying, per notice (spillover) | 0 | Your baseline: no spillover. Covers riders who see a notice or hear about it later. Only evaders can add paid rides; about half of bus riders are evading. No public source measures lasting spillover. The MTA's +7% paid boardings was measured at the stop while agents were there. |
| Rides each of those evaders pays for | 1 ride | Your baseline. Has no effect while spillover is 0. Total paid rides per notice = ticketed rider's rides + other evaders × rides each. |
| Notices that are warnings | 89% | "Approximately 89% of the fare evasion offenses resulted in warnings" in 2025 to date, all issuers. TAB report, Dec 15, 2025. It was 85% in the first half of 2025. High confidence. |
| Fines actually collected | 63% | Uses the 2024 yield of $63.32 per $100 notice. TAB "yield per NOV" was $74.99 (2023) and $63.32 (2024) on a $100 fine, including state tax-refund offsets (TAB report, July 22, 2025). A 2016 state Comptroller audit found about 56% collected (OSC). Not published separately for NYCT and MTA Bus. Medium confidence. |
| Base salary | $68,000 | Median Special Inspector hourly rate in Sept 2026 payroll ($32.58) × 2,088 hours. Posting range: $60,933 in year 1 to $93,740 in year 6. Fare Enforcement Agent: $63,066 to $97,021 (exam notice 7602). High confidence. |
| Benefits as % of salary | 92% | NYC Transit 2026 adopted budget: $3,911.8M in health, pension, retiree health (OPEB) and other fringe on $4,260.3M of payroll = 91.8%. 75% without retiree health; 87% from 2025 actuals. NYCT February Financial Plan 2026–2029. High confidence (agency-wide, not specific to the EAGLE team). Includes retiree health, which is a real cost to the MTA; without it the rate would be 75%. Works against enforcement |
| Target evasion rate | 5% | A policy goal for comparison, not an MTA target. |
Everything is per average day, then multiplied by 365 for yearly totals.
Total riders B = 1,125,551 paid ÷ (1 − 47.4%) = 2,139,831 (fixed) Evasions today E₀ = B × 47.4% = 1,014,280 (measured while today's team was working) Today's team output S₀ = 336 agents × (195 ÷ 275) = 238 notices a day (2025 rate per payroll agent) Agents on duty = field agents × deployment rate × 260 ÷ 365 (on an average calendar day) Notice capacity C = (agents on duty ÷ agents per notice) × (field hours × 60 ÷ minutes per notice) × utilization The default utilization (3.4%) makes 336 agents write exactly S₀. Once utilization is set to the 2025 rate, minutes and agents per notice cancel out; they matter only when utilization is set above it. Deterrence D: each notice leads to D paid rides that would otherwise have been evasions. D = rides the ticketed rider pays for + other evaders who start paying × rides each pays for. Only notices above today's output lower evasion, because there is no reliable "no team" rate: extra notices X = C − S₀ (if C is below S₀, evasion stays at 47.4%) Agents can only ticket riders who are still evading, so notices written S ≤ remaining evasions E₁. If C ≤ (E₀ + D × S₀) ÷ (1 + D): S = C, E₁ = E₀ − D × X Otherwise: E₁ = (E₀ + D × S₀) ÷ (1 + D), S = E₁ (every remaining evader is ticketed) New evasion rate = E₁ ÷ B; paid riders = B − E₁ Chance of a notice = S ÷ E₁ per evasion Lowest possible rate at D = (E₀ + D × S₀) ÷ (1 + D) ÷ B Rides needed for target t, any team = (E₀ − t × B) ÷ (t × B − S₀) Rides needed with your team = max((E₀ − t × B) ÷ X, rides needed with any team) Team needed for target t = (S₀ + (E₀ − t × B) ÷ D) ÷ notices per agent per day Cost = agents × salary × (1 + benefits) + (agents ÷ 6.46) × $108,158 × (1 + benefits) [6.46 = 336 ÷ 52] Fines = S × 365 × (1 − warning share) × net fine × collection rate net fine = 0.72 × ($100 − 0.72 × $50 credit) + 0.28 × $150 = $88.08 Fares = S × D × 365 × $1.88 (every notice, today's included, is credited with D paid rides)Every number below can be checked with a calculator. Settings: 336 agents, 82% deployment, 6 field hours, 5 minutes per notice, 2 agents per notice, 100% utilization (the "Perfectly efficient" button), 1 ride per notice and no spillover (D = 1 + 0 × 1 = 1), 89% warnings, 63% collection. The page opens at the 2025 pace instead (Step 2).
STEP 1: How many notices can the team write on a typical day? Agents who show up on a scheduled day 336 × 82% = 275.5 Spread over 7 days (they work 5) 275.5 × 260 ÷ 365 = 196.3 agents on duty Notices per agent per shift 6 hr × 60 ÷ 5 min = 72 if writing alone Two agents per notice 72 ÷ 2 = 36 per agent per shift Utilization × 100% = 36 Team capacity C = 196.3 × 36 = 7,065 a day STEP 2: Today's team at the 2025 pace (the page's default) Agents on duty in 2025 275 × 82% × 260 ÷ 365 = 160.6 Notices per agent per shift 195 ÷ 160.6 = 1.2 Agent-minutes per notice 2 agents × 5 min = 10 Field time spent writing 1.2 × 10 = 12 min of 360 = 3.4% Today's team at 3.4% S₀ = 336 × (195 ÷ 275) = 238 a day, so evasion stays 47.4% STEP 3: What perfect efficiency does to evasion (1 ride per notice) Total riders B = 1,125,551 ÷ (1 − 47.4%) = 2,139,831 Evasions today E₀ = 2,139,831 × 47.4% = 1,014,280 Extra notices X = 7,065 − 238 = 6,827 Evasions left E₁ = 1,014,280 − 1 × 6,827 = 1,007,453 New evasion rate = 1,007,453 ÷ 2,139,831 = 47.08% Chance an evasion gets a notice = 7,065 ÷ 1,007,453 = 0.70%, about 1 in 143 STEP 4: Why 5% is out of reach at 1 ride Target evasions Eₜ = 5% × 2,139,831 = 106,992 Lowest possible rate = (1,014,280 + 238) ÷ 2 ÷ 2,139,831 = 23.7% Rides needed with any team = (1,014,280 − 106,992) ÷ (106,992 − 238) = 8.5 Rides needed with this team = (1,014,280 − 106,992) ÷ 6,827 = 133, about 3 months STEP 5: Money per notice Net fine when paid = 0.72 × ($100 − 0.72 × $50) + 0.28 × $150 = $88.08 Average brought in per notice = (1 − 89%) × $88.08 × 63% = $6.10 Fare recovered per notice = 1 ride × $1.88 = $1.88 Annual cost = 336 × $68,000 × 1.92 + 52 × $108,158 × 1.92 = $54.7M Cost per notice = $54.7M ÷ (7,065 × 365) = $21.20 Money back per $1 spent = ($6.10 + $1.88) ÷ $21.20 = about 38¢ At the 2025 pace, the same team writes 238 a day: $629 per notice and about 1¢ back per $1.The model's results depend more on how many notices each agent can write than on how strongly a notice deters. If agents wrote a notice every 5 minutes for their whole field shift, about 1,000 agents could reach 5% evasion, assuming each notice kept a rider paying for a month. At the EAGLE team's actual 2025 pace, the same goal would take about 30,000 agents.
That gap isn't a sign the team is doing its job badly. It comes from how buses work. Agents have to wait for the next bus to arrive. Not every bus that stops has evaders on it. Checking every rider and writing notices takes time while buses keep to their schedules. The main ways to raise output are making sure agents check every bus that stops and work every hour of the shift they're paid for, and both have limits. No team will write notices every minute of its shift.