The goal is not to drive compensation to the bottom or to defend every existing compensation package. The goal is to “Goldilocks” public employee compensation: not too low, not too high, but just right.

The goal is to “Goldilocks” public employee compensation: not too low, not too high, but just right.

The Cost of Paying Too Little

Paying too little is not mainly a moral problem. It is a state-capacity problem. When governments cannot attract and retain qualified people for hard-to-fill roles, the result is not savings. The result is weaker service, slower enforcement, and lower-quality government.

One concrete example comes from New York City. In 2022, the New York City Comptroller reported severe understaffing in resident-facing roles across multiple agencies: the Department of Social Services’ Child Support Services division was staffed at only 415 people against 775 authorized positions, the City’s Cyber Command had a 36 percent vacancy rate, the Administration for Children’s Services’ Head Start and Daycare unit had a 35.2 percent vacancy rate, and the Department of Finance was short auditors and accountants.¹ State-level oversight has likewise documented continuing strain in city agencies, with elevated vacancies in Environmental Protection, Transportation, Health, and other service-delivery agencies.² For residents, understaffing in those kinds of agencies can mean slower benefit administration, weaker child-support enforcement, longer waits for services, and less consistent follow-through on basic government functions.

The same problem appears beyond New York. Recent research on U.S. local health departments found that non-competitive salaries are one of the main barriers to recruitment and retention.³ That is exactly what one would expect in a competitive labor market: if government employers pay too little for specialized work, strong candidates go elsewhere.

That said, vacancies do not automatically prove that government should simply hire more people. Better-governed places should also ask whether they are assigning public employees work that could be handled more efficiently through better institutional design. In housing and construction permitting, for example, licensed architects and engineers can often certify routine compliance, leaving public officials to focus on audits, field inspections, enforcement, complaints, and genuinely higher-risk cases.⁴ New York City already uses professional certification for some filings, and Chicago and Phoenix also have self-certification programs for qualifying projects.⁵ Staffing needs in this area should therefore be understood partly as a compensation issue and partly as a process-design issue.

The Pitfalls of Paying Too Much

Governments can also fail in the opposite direction. Compensation becomes too high, or too poorly structured, when earnings are driven less by labor-market scarcity and job difficulty than by political protection, weak oversight, or contract terms that are easy to game.

In the United States, this problem often appears not as obviously inflated base salary but as overtime abuse, leave manipulation, and pension spiking. Pension spiking is the practice of boosting late-career compensation in order to raise pension benefits tied to final-average salary.⁶ The core problem is not that every public pension is illegitimate. It is that badly designed systems can reward strategically timed compensation rather than long-run public value.

California audits have documented how these problems can grow under weak controls. In one overtime review, the California State Auditor found that a small number of employees in specific classifications accounted for strikingly high overtime costs, including nurses at Napa State Hospital whose average overtime exceeded their average regular pay.⁷ That is not just a budget warning sign. It is also a safety and fatigue problem. When systems rely too heavily on extreme overtime, employees are being encouraged to work hours that can degrade judgment and performance rather than supported by better staffing design.⁸

New York offers another example of how excessive overtime can become a structural fiscal problem rather than a short-term management tool. In 2025, New York State agency overtime reached $1.6 billion, with the Office of Corrections, the Office for People With Developmental Disabilities and the Office of Mental Health together accounting for most of the concentration described in the Comptroller’s report: those agencies comprised 21.7 percent of the workforce but generated 68.3 percent of overtime hours and 70 percent of overtime earnings logged by state agencies.⁹ When that happens year after year, the issue is no longer occasional flexibility. It signals a compensation and staffing system that is not being managed clearly or sustainably.

Overpay is therefore best understood as a jurisdiction-specific governance problem. Some places may still pay less than market for certain roles. Others may have compensation systems that are plainly too generous, too opaque, or too manipulable. Benchmarking should help residents distinguish between the two.

Finding the “Just Right” Zone

The practical test is not whether a compensation package looks high or low in the abstract. The test is whether the jurisdiction can recruit and retain competent people for important work without creating compensation systems that insiders can exploit.

If a city repeatedly cannot fill inspector, engineer, accountant, health, or social-service positions, compensation may be too low, work conditions too onerous, the hiring process may be too slow, or the role may be structured poorly. If a jurisdiction sees chronic overtime spikes, pension gaming, or politically connected hiring into unusually attractive positions, compensation or management may be too loose.

Good government requires both discipline and flexibility. It means paying enough where genuine scarcity exists, holding the line where roles are easy to fill, and redesigning processes so public employees are used where they add the most value.

Tools for Controlling Compensation Drift

Conduct Classification Reviews

Bureaucracies tend to inflate titles and compensation over time. A classification audit compares what workers actually do with the job descriptions on file. The U.S. Office of Personnel Management describes a desk audit as an interview conducted by a human-resources specialist to understand the duties and difficulty of a position.¹⁰ These reviews can uncover situations where roles no longer match the compensation attached to them or where titles have drifted away from actual responsibilities.

Benchmark Against the Labor Market

Compensation should not be set in a vacuum or defended by habit. Governments should benchmark compensation against comparable labor markets, but they should also benchmark outcomes: vacancy rates, overtime trends, turnover, service quality, and pension costs.¹¹ A place with moderate compensation but chronic vacancies may be underpaying. A place with large payrolls and runaway overtime may be mismanaging compensation in the opposite direction.

The real question is not whether compensation debates can be settled in the abstract. The real question is which jurisdictions are getting specific roles wrong, and what residents are receiving in return.

Address Rigid Compensation Structures

In some parts of government, compensation is made less flexible by rigid salary schedules, job classifications, seniority rules, and bargaining structures that are not designed for fast labor-market adjustment. The clearest evidence comes from public education, where traditional salary schedules often pay almost entirely by years of experience and credentials rather than by subject-area scarcity or school-specific staffing difficulty.¹² That can make it hard to raise compensation for shortage areas without raising it across the board.

The point is not that collective bargaining itself is inherently bad. The narrower point is that some compensation systems are too rigid to match pay to actual recruitment conditions. When that happens, governments can end up undercompensating hard-to-fill roles while overcompensating easier-to-fill ones.

There are examples of jurisdictions trying to reform that problem. Denver’s negotiated ProComp system added differentiated compensation for hard-to-serve schools and hard-to-staff assignments within a unionized teacher-pay framework.¹³ In Georgia, a statewide bonus program for certified math and science teachers reduced attrition substantially, showing that targeted compensation changes can improve retention where shortages are real.¹⁴

Reduce Abuse and Future Liabilities

When collective agreements or personnel rules make it easy to inflate late-career earnings, convert overtime into pensionable compensation, or hide the true cost of staffing, taxpayers eventually pay for design failures that should have been fixed earlier. Employees should be paid clearly, fairly, and transparently for the work they actually do. Clear limits on pensionable earnings, tighter overtime controls, and better disclosure of unusually large compensation packages help make that possible while reducing opportunities for manipulation.

Paying a Premium for Public Safety

Public safety work can justify premium compensation. Police officers and firefighters face risks, scheduling burdens, and responsibilities that do not map neatly onto most civilian roles. Communities should be willing to pay enough to attract capable people for genuinely demanding public safety jobs.

But premium compensation is not a blank check. The existence of danger does not justify opaque overtime, inflated staffing hierarchies, or pension rules that reward manipulation over performance. The right question is not whether public safety workers deserve good compensation. They do. The right question is whether the compensation system is disciplined, transparent, and tied to actual public value.

Nepotism: A Persistent Threat

Few things destroy morale and public confidence faster than nepotism. When attractive public jobs are awarded on the basis of family ties or personal connections rather than merit, compensation becomes more than a budget issue. It becomes a governance issue. Would you want to be the supervisor of the mayor’s slacker nephew or the council president’s entitled mistress?

Would you want to be the supervisor of the mayor’s slacker nephew or the council president’s entitled mistress?

New York State’s own ethics rules recognize the risk. State law prohibits covered officials and employees from participating in decisions to hire, promote, discipline, or discharge relatives, and from awarding contracts to relatives.¹⁵ That is the right instinct but it only applies at the state level. If a job is attractive enough relative to other available options that insiders will fight to reserve it for friends and family, government needs stronger safeguards.

Nepotism can also compromise labor negotiations. It is hard to bargain at arm’s length when many of your friends and family are union members. Even well-meaning officials may find it difficult to separate private loyalties from public obligations, and the wider community can bear the cost.

Strict anti-nepotism rules, open competition, and transparent hiring are part of getting public compensation “just right.” Compensation systems that are much more attractive than comparable alternatives can invite patronage when hiring safeguards are weak. Compensation that is too low can leave agencies chronically understaffed. Both outcomes weaken government.

Are There Rare Exceptions?

There are rare cases where the rule may bend without breaking. If the mayor’s brother is a genuine financial savant, or the governor’s daughter is the kind of lawyer who could be billing ten times as much in private practice but wants to serve, blanket moralizing can miss the point. Public life has sometimes benefited from exactly that kind of high-talent, high-sacrifice service, including controversial cases like Robert F. Kennedy and non-family examples like Robert Rubin and John C. Whitehead, who left much better paid private-sector careers for government. But that is the exception that proves the rule. If an insider gets the job, the burden of proof should be brutal: clear legal authority, full transparency, real public scrutiny, and no doubt that the appointment reflects extraordinary merit rather than family convenience.

Conclusion

Public employee compensation is not a culture-war question. It is a governance question.

Some jurisdictions underpay the people they most need and get slower, weaker service as a result. Others tolerate compensation systems distorted by overtime abuse, pension manipulation, or political favoritism.

Better government requires a narrower and more practical standard. Pay enough to recruit and retain competent people for work that matters. Do not pay so much, or structure pay so poorly, that the system attracts rent-seeking, nepotism, or hidden long-term costs. And redesign public processes so government employees are doing the work that truly requires public authority rather than duplicating what licensed professionals can already certify.

That is the Goldilocks principle applied to government: not too little, not too much, and not in the wrong places.

If you have additional insights or examples of successful (or unsuccessful) efforts to balance public compensation, please share them at contact@benchmarkusa.org. Benchmarking experiences and lessons from around the country will help our communities learn from each other.

Price is what you pay.
Value is what you get.

Warren Buffett 2008 Letter to Berkshire Hathaway Shareholders

What do you actually pay for government at a specific address, and what do you get in return? BenchmarkUSA is built to make that answer clearer.

Price is the missing interface in local government

Most people have a rough sense that one place has “high taxes” and another place has “lower taxes.” But that is not the same as knowing the price of government at a real address.

In practice, people face overlapping layers: state, county, city/town/village, school district, and sometimes special districts. The tax burden is fragmented across those layers, and the quality or sustainability of services is even harder to evaluate.

That opacity weakens accountability. When price signals are hard to see, incentives get blurry. Households and businesses cannot easily compare tradeoffs, and governments receive weaker feedback from the people they serve.

“Under competition – and under no other economic order – the price system automatically records all the relevant data. Entrepreneurs, by watching the movement of comparatively few prices, as an engineer watches a few dials, can adjust their activities to those of their fellows.”

Friedrich A. Hayek, The Road to Serfdom (Reader’s Digest condensed edition).

Hayek’s key insight is directly relevant here: prices are a mechanism for communicating dispersed information. In local government, that mechanism is fragmented across institutions and hard for residents to observe.

BenchmarkUSA’s objective is to make those signals legible so incentive loops work better in practice. At a minimum, users should be able to see:

1. what they pay
2. what layer imposes each component
3. whether the system delivering those services appears fiscally resilient

What we are shipping now: Tax-Stack Calculator (work in progress)

The current implementation is NY-first and address-first.

For a given address, the calculator is being built to show:

• the jurisdiction stack that applies
• which taxes are source-backed and calculable now
• which components still require more inputs or source coverage
• explicit caveats when data is missing or method paths are still in progress

Current state (April 27, 2026):

• NY state + destination-based local sales-tax stack is loaded from official NY sources
• NY income-tax method routing is implemented in part, with remaining table/worksheet paths explicitly marked rather than hidden
• Household-income assumptions now support a source-labeled ACS/ZCTA default path for estimation scenarios
• The calculator is intentionally transparent about calculated vs input_required vs rate_missing states

Rollout plan over the next week

Near-term rollout is designed so people can test with real addresses immediately:

1. finish NY state coverage in the next few days
2. expand to VA/DC/MD next so Northern Virginia reviewers can test real addresses
3. expand to NJ/CT shortly thereafter
4. expand from specific addresses to more general municipalities

Rationale: Ship where users and reviewers can directly pressure-test the product. Get immediate feedback.

Beyond taxes: Fiscal Stress Assessment

Price alone is not enough. Two places can have similar tax levels while carrying very different long-term risks.

BenchmarkUSA already imported and validated New York Fiscal Stress Monitoring System (FSMS) coverage across NY entity types. That NY baseline is the model for building a broader, state-neutral Fiscal Stress Assessment framework for municipal entities across the U.S.

The next model layer needs to combine near-term and long-horizon signals, including:

• reserves/unassigned funds,
• debt-service pressure,
• delayed infrastructure maintenance and deferred capital investment,
• pension and other long-term obligation funding adequacy.

In other words: not only “what do you pay now?” but also “how sustainable is the system you are paying for?”

What users should get from this

For households, employers, site selectors, journalists, and local officials, the intended value is straightforward:

• clearer tax pricing at the address level,
• better comparability across overlapping jurisdictions,
• earlier visibility into fiscal fragility before it turns into service disruption or crisis.

Longer term, the objective is national coverage: an address-level tax-stack calculator and fiscal resilience grader for all U.S. addresses and political entities.

If you want to follow progress or share feedback, reach out at contact@benchmarkusa.org.

Why fiscal health matters to residents

When a local government’s fiscal stress score drops, real things happen:

Lower borrowing costs. Bond ratings rise, interest rates fall. Rockland County went from one notch above junk to Moody’s AAA — the difference between paying millions more in interest and having that money available for roads, parks, and services. Monroe County’s seven rating upgrades since 2021 mean every future capital project costs taxpayers less.

More room to invest. A government under fiscal stress is deciding which services to cut. A government in good fiscal health can make strategic investments — addressing environmental challenges, upgrading infrastructure, absorbing economic shocks. Fund balance isn’t money sitting idle; it’s the difference between reacting to crises and preventing them. The OSC’s own research on fiscal stress drivers shows how depleted reserves cascade into service cuts and deferred maintenance.

Tax relief becomes possible. Monroe County’s property tax rate is now the lowest since 1947. Rockland County proposed zero tax increase for 2026. Fiscal discipline doesn’t just stabilize budgets — it creates room to lower the burden on residents.

Accountability to the state. The OSC’s FSMS isn’t just a report card. For Nassau County, fiscal stress triggered a state control board (NIFA) with authority to reject budgets and impose its own. Getting out of stress means regaining local control over local decisions.

The turnarounds

Rockland County — From worst to first

Population: ~339,000. Peak stress: 87 (2012–2013). Current: 3.3 (2024).

In 2014, Rockland County had a $240 million debt load, a $138 million deficit, and the title of New York State’s most fiscally stressed county. Its credit rating was one notch above junk.

Under County Executive Ed Day, the county embarked on what became one of the most dramatic fiscal recoveries in state history. Budgeting practices were overhauled, reserves rebuilt, and the county’s $96 million deficit bond was paid off in 2024. Rockland is now the only New York county with a Moody’s AAA bond rating — their highest possible rating. From worst to first.

The FSMS data shows the trajectory clearly: 87 in 2012, 64 in 2017, 26 in 2019, 7 in 2021, 0 by 2022.

View Rockland County’s stress trajectory →

Monroe County — Seven rating upgrades in five years

Population: ~754,000 (home to Rochester). Peak stress: 82 (2012–2015). Current: 15.8 (2024).

Monroe County carried “Significant Fiscal Stress” for eight consecutive years (2012–2017). A 2018 state audit revealed the county was on the brink of insolvency, resorting to short-term borrowing to cover cash deficits, with fund balance at just 0.1% of total spending.

The turnaround under County Executive Adam Bello has been methodical: conservative budgeting, growing reserves, and a property tax rate now at its lowest since 1947. The county has received seven credit rating increases since 2021 and is now rated AA by all three major agencies — the first time since 2001.

Environmental stress has remained low throughout (under 17), suggesting the county’s problems were financial management rather than structural economic decline. That makes the recovery more notable — it was largely within the county’s control.

View Monroe County’s stress trajectory →

Niagara Falls — Recovery despite the odds

Population: ~48,000. Peak stress: 77.1 (2017). Current: 0.0 (2024).

Niagara Falls was designated “Significant Fiscal Stress” for five consecutive years (2017–2020), driven partly by a dispute between New York State and the Seneca Nation over casino revenue sharing. The city relied heavily on gaming revenue — $140 million since 2013 — and was exposed when payments stopped.

The recovery involved improved budgeting practices and reduced reliance on casino revenue. Fund balance grew from $1.1 million in 2021 to $21.3 million by 2024 — four consecutive years of operating surpluses. S&P raised the city’s bond rating from BBB+ to A- with a “positive” outlook.

What makes Niagara Falls remarkable is the environmental context. Environmental stress has held steady around 40 throughout the period — population decline, poverty, aging infrastructure. The city got its fiscal house in order despite persistent structural challenges, not because those challenges went away.

View Niagara Falls’ stress trajectory →

Nassau County — Fiscal improvement under state oversight

Population: ~1.4 million (Long Island). Peak stress: 73 (2014). Current: 3.3 (2024).

Nassau County’s fiscal story is complicated by NIFA — the Nassau Interim Finance Authority, a state control board created in 2000 with authority to monitor and, when necessary, override county financial decisions. Nassau was designated “Significant Fiscal Stress” by the OSC in 2014, 2017, and 2018.

The improvement since then has been real: scores dropped from 73 to 3.3, and the county has carried “No Designation” since 2022. But the story isn’t entirely one of local initiative — NIFA sent Nassau’s 2025 budget back for revision in late 2024, citing the use of prior-year surpluses to fund future expenses in violation of GAAP.

For a county of 1.4 million people, the second most populous in the state outside NYC, the tension between local governance and state oversight is itself a story worth following.

View Nassau County’s stress trajectory →

Town of Oyster Bay — Post-scandal recovery

Population: ~301,000. Peak stress: 76.7 (2017). Current: 0.0 (2024).

Oyster Bay’s fiscal stress is inseparable from its governance history. Former Town Supervisor John Venditto pleaded guilty to corruption charges in 2019 after a federal investigation revealed hidden side deals that exposed the town to over $20 million in concealed liabilities. The SEC charged both the town and Venditto with defrauding municipal bond investors. The town’s credit rating, once AAA, was slashed to BB+ — junk status, the only locality in New York to hold that distinction.

Under new leadership, the town eliminated its deficit and paid down $150 million in capital debt. Three consecutive years of zero fiscal stress (2022–2024) and zero environmental stress tell a recovery story, but also a cautionary tale about the fiscal damage that corruption inflicts on communities.

View Town of Oyster Bay’s stress trajectory →

Yonkers — The big city that turned around

Population: ~210,000 (4th largest city in New York). Peak stress: 67.5 (2020). Current: 5.0 (2024).

Yonkers oscillated between “Susceptible” and “Significant” for years, receiving a “Significant Fiscal Stress” designation in 2020 at 67.5. The reversal was sharp: 31 in 2021, 5 in 2022, 2 in 2023. Environmental stress dropped in parallel, from 37 to 13.

As the largest city on this list and the fourth largest in the state, Yonkers demonstrates that fiscal turnarounds aren’t limited to small jurisdictions with simple budgets. However, a 2024 OSC budget review noted the city’s reliance on $133.5 million in nonrecurring revenue — a reminder that sustained fiscal health requires ongoing discipline, not one-time fixes.

View Yonkers’ stress trajectory →

What these stories share

Turnarounds take 4–7 years. None happened overnight. Most involved consistent improvement over five or more budget cycles.

Environmental stress is stickier than fiscal stress. Niagara Falls brought fiscal stress to zero while environmental stress stayed at 40. You can fix your budget; you can’t quickly fix population decline or poverty rates. But being fiscally healthy gives you more options to address those structural problems.

The bond market rewards discipline. Rockland went from near-junk to AAA. Monroe got seven upgrades. Niagara Falls moved from BBB+ to A-. Every upgrade means lower interest rates on future borrowing — real money that can go to services instead of debt payments.

Fiscal stress is largely within local control. Monroe County’s low environmental stress during its fiscal crisis shows the problem — and the solution — was financial management. Oyster Bay’s corruption-driven collapse shows how governance failures create fiscal crises independent of economic conditions.

If you’re a local official, journalist, or researcher — or a resident who wants to understand your community’s fiscal trajectory — I’d love to hear from you at contact@nybenchmark.org.

All fiscal and environmental stress scores are from the NYS Comptroller’s Fiscal Stress Monitoring System. Population figures are from the U.S. Census Bureau American Community Survey. Pre-2017 FSMS scores have been normalized to the 100-point scale for trend consistency. Bond rating information is sourced from official county/city press releases and reporting by the Bond Buyer, WGRZ, Long Island Press, and Mid Hudson News.

Niagara Falls went from “Significant Fiscal Stress” (score ~75) in 2012 to near zero by 2024 — a textbook turnaround.

A single year doesn’t tell the story

The NYS Comptroller’s Fiscal Stress Monitoring System (FSMS) scores local governments on two dimensions: fiscal stress (fund balance, cash, operating deficits) and environmental stress (population trends, poverty, property values). Higher scores mean more stress — 0 is healthy, 100 is significant stress.

A scatter chart of the latest year gives you a useful snapshot. You can see which cities are under fiscal pressure and which face environmental headwinds. But one year is just a moment. Is a city at 45 fiscal stress getting better or worse? Did it spike from 20 last year, or slowly decline from 70?

That’s what the trajectory chart answers.

How to read a trajectory

Each trajectory chart plots fiscal stress (x-axis) against environmental stress (y-axis) with connected points showing the path a city has taken from 2012 to the present. The dots are color-coded by aid dependency — green means low reliance on state and federal aid, red means high.

Movement toward the bottom-left corner (lower scores on both axes) means improving health. Movement toward the top-right means worsening conditions. The arrowheads on each segment show the direction of change year over year.

Some patterns you’ll see:

• Steady improvement: a clear path from upper-right toward lower-left, year after year
• Volatile swings: large jumps in one year followed by corrections
• Environmental lock-in: fiscal scores improving while environmental scores stay stubbornly high (a city can fix its budget but can’t easily reverse population decline)
• Filing gaps: missing years appear as breaks in the trajectory, common for cities that filed late or skipped a year

What we’re seeing

A few patterns stand out across the dataset:

Environmental stress is stickier than fiscal stress. Many cities have meaningfully improved their fiscal scores over the past decade — better fund balances, healthier cash positions. But environmental scores (driven by demographics, poverty rates, property values) move slowly. A city can right-size its budget in a year or two; reversing population loss takes a generation.

Aid dependency correlates with trajectory direction. Cities with high aid dependency (red dots) tend to cluster in the upper-right quadrant and show less improvement over time. Cities that generate more of their own revenue (green dots) have more fiscal flexibility to course-correct.

Niagara Falls: a turnaround story. As shown in the trajectory above, Niagara Falls had a fiscal stress score near 75 in 2012 — deep in “Significant Fiscal Stress” territory. Over the next decade, the city steadily improved its fiscal position, reaching a score near 0 by 2024. The trajectory tells the story clearly: a long, consistent march from the right side of the chart to the left.

The 2017 methodology change is visible. FSMS switched from a weighted fraction system (0-1.0) to a direct 100-point scale in 2017. We normalize pre-2017 scores to the 0-100 scale for visual continuity, but you may notice a cluster of trajectory shifts around 2016-2017 that reflect the scoring change rather than actual fiscal movement.

Try it yourself

Explore the full dataset — cities, school districts, counties, towns, and villages — on the stress analysis page. Click any dot to see its trajectory, zoom and pan to focus on specific time periods, download charts as PNGs, and expand the data table and indicator breakdown to see exactly which stress indicators drove each year’s score.

Data sources

All scores come from the OSC FSMS, covering fiscal years 2012-2024 for municipalities and 2013-2025 for school districts. Aid dependency is calculated from OSC Annual Financial Report data (state aid + federal aid as a percentage of total operating revenue). See our methodology page for details on data processing, normalization, and known limitations.

The missing 40%

New York City is home to 8.3 million people — roughly 40% of New York State’s population. It runs the largest municipal government in the country, with a $117 billion budget that dwarfs every other city in the state combined.

And until today, it wasn’t in our dataset.

The reason is structural: NYC has its own Comptroller and does not file Annual Financial Reports with the NYS Office of the State Comptroller (OSC). Every other city in the state — all 61 of them — files with OSC. NYC operates under an entirely separate reporting system.

This isn’t a data gap we could fix by asking nicely. It required building a separate import pipeline from scratch.

Why ACFR, not budget data

NYC publishes financial data through multiple channels: the Checkbook NYC transparency portal, the annual budget, and the Annual Comprehensive Financial Report (ACFR). We evaluated all three.

Budget data was tempting — it’s well-structured and available through an API. But a municipal budget is a plan, not a record. It reflects political priorities and revenue forecasts at the time of adoption. What actually happens over the next twelve months can look very different. Departments overspend or underspend. Revenue comes in above or below projections. Emergency spending appears that nobody budgeted for. Some governments track closely to their budgets; others treat them more like rough guidelines.

For NYC specifically, the gaps are large. Revenue projections can miss by 4-9%, and expenditure estimates can differ by billions. The Citizens Budget Commission, the Independent Budget Office, and the Comptroller’s own analysis all document this pattern. If you’re comparing cities to each other — which is the entire point of benchmarking — you need to compare what they actually spent and collected, not what they hoped to.

The ACFR is audited actuals. An independent auditor has verified the numbers. This is the same standard we use for every other city in the state (via OSC filings), and it’s the standard that credit rating agencies, the Government Finance Officers Association, and academic researchers rely on. Since FY 2022, the NYC Comptroller has published downloadable Excel files containing the ACFR’s statistical tables. The Ten Year Trend tables give us FY 2016-2025 from a single download.

Maintaining audited-actuals-only across all 62 cities is worth the extra work. Mixing budget data for one city with ACFR data for the rest would undermine every comparison on the site.

What we imported

From the FY 2025 ACFR statistical tables:

• Expenditures by functional category (General Government, Public Safety, Education, etc.) and by agency (~90 agencies including Police, Fire, DOE, CUNY, Sanitation, and more)
• Revenue by source (real estate tax, income tax, sales tax, federal grants, state grants, charges for services, etc.)
• Fund balances by GASB 54 classification (Restricted, Committed, Assigned, Unassigned) for General Fund and all governmental funds

In total: 77 metrics, 10 fiscal years (FY 2016-2025), 763 observations.

We mapped NYC’s functional categories to OSC’s level_1_category names (e.g., NYC’s “Public Safety and Judicial” becomes “Public Safety”) so that cross-city comparisons work. Agency-level detail goes into level_2_category, giving NYC richer drill-down data than most other cities have.

One important caveat: the Department of Education’s $34.1 billion appears as a city expenditure line item under “Education.” But unlike upstate cities where school districts are separate entities with their own per-pupil metrics, NYC’s DOE is part of the city government. We capture the spending total but don’t yet break it down as a school district with enrollment-normalized metrics. That’s a future project — and a significant one, given that NYC’s education budget alone is larger than the total budget of most US states.

The 98 agencies we haven’t mapped yet

NYC’s ACFR lists 134 individual agencies across 13 functional categories. We mapped 36 of them to individual level_2_category entries — the major departments like Police, Fire, DOE, Sanitation, and so on. The remaining 98 are captured in their category totals (no spending is lost at the aggregate level) but aren’t broken out individually yet.

Some of these unmapped agencies are uniquely NYC:

These aren’t data gaps — they’re structural differences in how NYC governs compared to every other city in the state. Borough Presidents, Community Boards, and a citywide DA system with five offices don’t exist anywhere else. Mapping them to OSC-compatible categories requires careful judgment about where they fit, and in some cases they simply don’t have a clean equivalent.

Parsing the Excel files

The technical challenge wasn’t the data itself — it was extracting it from Excel files that were clearly designed for human reading, not machine processing.

The ACFR statistical tables are formatted for print. Agency names spill across multiple rows because they’re too long for a single cell. Category headers like “General Government:” sit on their own row with no numeric data. Some categories span multiple pages and repeat with “(cont.)” appended. The fund balance file has decorative filler text between section headers.

A few specific headaches:

Multi-row agency names. “260 Department of Youth and Community Development” spans three rows in the spreadsheet — the name on the first two rows (text only, no numbers), the data on the third. The importer accumulates text-only rows and prepends them when it hits a data row.

Category headers vs. name fragments. Both are text-only rows with no numbers. The difference: category headers end with a colon. But some category names also span two rows (“Parks, Recreation, and / Cultural Activities:”), so the importer has to distinguish between a legitimate two-row header and filler text that accumulated before the real header.

Filler text accumulation. The fund balance file has rows like “Capital projects,” “Debt service,” “Noncurrent mortgage loans” sitting between “Reserved for:” and “Committed for:” — descriptive labels with no data. If the importer naively combines all accumulated text with the next colon-terminated header, you get nonsense. The fix: only combine if exactly one text row was accumulated (a legitimate two-row header). Three or more accumulated rows get discarded as filler.

We verified the parser’s output against exact dollar amounts from the published ACFR PDF. Police Department: $6,610,389,000. Total Expenditures: $109,610,157,000. Real Estate Tax: $34,756,900,000. Every number matches to the dollar.

The fund balance surprise

This was the most interesting finding. When you look up a typical NY city’s fiscal health, one of the key metrics is Fund Balance % — Unassigned General Fund Balance (account code A917) as a percentage of expenditures. This tells you how many months of reserves a city has.

NYC’s General Fund has no unassigned fund balance. Zero.

This doesn’t mean NYC is broke. It means NYC classifies its reserves differently:

• Restricted: $2.0 billion (legally restricted for specific purposes)
• Committed: $1.97 billion (the Revenue Stabilization Fund — NYC’s rainy day fund, committed by City Council resolution)

Under GASB 54, “Committed” is one step below “Assigned” and two steps below “Unassigned.” NYC’s policy is to park its reserves in the Committed classification, which has a higher bar for release (requires City Council action). This is arguably more fiscally conservative than leaving reserves as “Unassigned.”

We’ve updated our Fund Balance % metric to use Available Fund Balance — the sum of Committed, Assigned, and Unassigned balances. This is what credit rating agencies (Moody’s, S&P) and the GFOA actually look at. Under this definition, NYC’s $1.97B Committed balance shows up correctly. The old metric (Unassigned only) would have shown 0% — a classification artifact, not a sign of financial weakness. Buffalo had the same problem in some years after moving reserves to their Committed rainy day fund.

We’ve also added domain notes in our MCP server to flag NYC’s unique fund balance structure for AI tools that query the data.

Key numbers (FY 2025)

For context, NYC’s police budget alone ($6.6B) is larger than the entire budget of any other city in the state. Per-capita and percentage metrics are the only meaningful way to compare NYC with its peers.

What’s next

• MCP server — Our MCP server lets AI tools like Claude query NYC and all other city data directly, with domain-aware caveats that flag comparability issues like fund balance classification automatically.
• Agency-level drill-downs — The remaining 98 agencies, including District Attorneys, Borough Presidents, and Community Boards, will get individual mappings in a future update.
• Historical extension — The FY 2022 ACFR ZIP extends some metrics back to FY 2005. A future import will add that historical depth.
• DOE as a school district — Breaking out the Department of Education’s $34.1B with per-pupil and enrollment-normalized metrics, comparable to our upstate school district data.

NYC is live now at benchmarkusa.org/entities/nyc.

I started this project in New York because I live here. Later, I wanted to know: how many other states have the same kind of data?

When I started seven weeks ago, my plan was to manually download PDF audit reports from every city, county, town, village, district, and authority in the state, then either hand-enter the financial data myself or secure funding to have credible freelancers help me. I spent the first several weeks doing exactly that — pulling ACFRs for cities like Yonkers and New Rochelle and seeding my database by hand.

Then I got lucky. I discovered that New York’s Office of the State Comptroller had already collected all of this data and made it available for bulk download in machine-readable form. Weeks of planned manual labor replaced by a few import scripts. That lucky break raised an obvious follow-up question: which other states have done the same thing?

The Fifty-State Review

I conducted a review of all fifty states to understand how each one collects, standardizes, and publishes local government financial information. I used Google Gemini’s Deep Research to systematically review state comptroller sites, statutory reporting requirements, and public data portals. The tooling dramatically accelerated discovery, but the classifications, interpretations, and conclusions are my own.

What I found is a sharp structural divide in how different states consider, collect and make available local government financial data.

The Real Transparency Divide

The key distinction is not whether data is “public” — virtually all government financial data is technically public. The question is whether it’s structured.

In states like New York, Washington, Ohio, Indiana, and the other green states on the map above, local governments submit standardized financial data into centralized systems. These platforms enforce uniform charts of accounts, validation rules, and consistent definitions. Spending, debt, and fiscal health can be compared across jurisdictions with minimal manual work.

Unfortunately, in much of the country, financial disclosure still means posting standalone audit PDFs. Try comparing per-capita police spending across 200 municipalities when each one publishes a different 300-page PDF with its own chart of accounts. It’s functionally opaque.

And then there are the states that have no centralized local reporting infrastructure at all.

The green states have centralized, machine-readable systems. Yellow states have central repositories but the data is often PDF-only or limited in export options. Red states lack centralized local reporting infrastructure entirely.

145 Million Americans Can See How Their Tax Dollars Are Spent

The good news is that those twelve green states are home to over 145 million people — nearly 44% of the country according to the 2020 Census. That’s how many Americans currently live in states where structured local government financial data exists and bulk comparative analysis is possible. The other 56% are largely in the dark.

Those local governments collectively spent approximately $1.18 trillion in FY 2022, according to the U.S. Census Bureau’s Annual Survey of State and Local Government Finances — a massive share of public spending that, in these twelve states, can actually be tracked and compared.

The Best State Databases

see below for more links and details about the best state local government financial dtabases. I’ve ordered them roughly by how useful they are for the kind of bulk comparative analysis I’m doing:

Florida: The Gold Standard

Florida deserves special mention. In 2018, the state passed HB 1073 mandating that all local governments file their financial reports in XBRL — the same structured data standard the SEC requires for corporate filings. The LOGERx system went live in 2022 and remains the first and only state-level XBRL mandate for local government financials in the country. Most other green states, including New York, publish bulk CSV or offer interactive dashboards, which is excellent for analysis but doesn’t carry the semantic richness of XBRL. If other states follow Florida’s lead, a truly comprehensive national database of local government finances becomes possible.

California: A Cautionary Tale

California illustrates both the promise and the fragility of transparency infrastructure. The state’s raw data is solid — the State Controller’s ByTheNumbers portal publishes 13 million+ fields of structured financial data for cities, counties, and special districts going back to 2002. Up until very recently they had a NYBenchmark-like analytical layer that made that highlighted at risk governments and made the data more accessible to non-experts. The State Auditor’s High-Risk Local Government Dashboard ranked all 471 California cities on fiscal health using ten financial indicators — exactly the kind of tool that lets residents and journalists ask hard questions about how their local government is managing money. Then in October 2023, following a change in leadership at the auditor’s office, the dashboard was quietly discontinued and scrubbed from the website. No announcement, no archive, no explanation.

This shouldn’t surprise anyone. Structured financial data invites scrutiny, and scrutiny creates pressure. There will always be elected officials and government employees who resent having their spending decisions compared to their neighbors’. That resentment is a feature, not a bug — it means the data is doing its job. But it also means transparency infrastructure is perpetually at risk of being defunded, deprioritized, or quietly shut down by the very people it’s designed to hold accountable.

The problem is compounded by the collapse of local news across the country and around the world. When California killed its dashboard, there was barely anyone left to notice. The California Policy Center, a good-government nonprofit, has since built a replacement — but civic projects come and go, and the underlying data should never depend on one organization’s funding or interest. The episode illustrates why transparency infrastructure needs to be grounded in statutory mandate — like Florida’s XBRL requirement — rather than left to the discretion of whoever happens to hold office. Dashboards built on political goodwill can disappear the moment that goodwill changes.

The variation across these systems is significant — in format, granularity, whether participation is mandatory, and how easy they make bulk access.

Where This Goes Next

There’s no reason this project should stay limited to New York. Every green state on the map above has the data infrastructure to support the same kind of cross-jurisdictional benchmarking I’m doing now. Washington, Ohio, Indiana, Michigan — these states are ready for the same treatment, and expanding to cover them is a medium-term goal.

For the yellow and red states, I hope this analysis is useful in a different way. The gap between “we publish audit PDFs” and “we have a centralized, structured reporting system” is the gap between nominal transparency and functional transparency. States that close that gap don’t just help researchers and civic tech projects — they give their own residents, journalists, and policymakers the ability to ask basic comparative questions about how their tax dollars are being spent. Every state should aspire to be green on this map.

The absence of structured data is the main reason civic technology projects so often stall outside a handful of states. Identifying where high-quality data already exists is the first step toward scalable, cross-jurisdictional accountability.

Live New York benchmarking work based on this analysis is available at benchmarkusa.org.

Why It Matters

My hope is that this project will highlight the practices of communities that are particularly well run, delivering world class government services at a reasonable price, but also draw attention to those governments that are particularly mismanaged. I’ve got faith that voters want good government and are willing to vote for politicians who manage and hire in such a way that delivers it. I also believe that voters often have a hard time knowing exactly what is going on in their governments. When governments and school districts fail to get their paperwork into the OSC, you have to wonder if they know themselves what is going on inside. For more on how we use OSC data and what we measure, see our methodology.

Chronic vs. Occasional Non-Filers

In any given year, there are governments and school districts with otherwise ok finances that miss a year. We can imagine a retirement or maternity leave leaving a community with the temporary loss of capacity to get that paperwork in. Those communities will recover and do fine. It’s the chronic non-filers that we’ve got to worry about.

What the Data Shows

The NY Benchmarking Project Database highlights the cities that haven’t gotten their OSC Annual Financial Report paperwork into the OSC and differentiates those that just miss a year here and there from those few that are chronic non-filers. Currently, just three cities are chronic non-filers: Mount Vernon, Ithaca, and Rensselaer.

What’s Next

As we add more data – the counties, towns, villages, districts, and authorities – we’ll highlight the well run as well as those in stress and not filing, too.

But governmental finance is complex, often influenced by the exigencies – political, financial, etc. – of the moment rather than by any interest in clarity and standardization. New Rochelle, for instance, has a school system that is politically (elected school board) and financially (levies its own taxes) independent. My city, in contrast, has a strong mayor who appoints the school board and the city pays for all of the school district’s expenses. I need to subtract out education-related expenses if I want to compare my city with New Rochelle.

Custodial Pass-Throughs

But it doesn’t stop there. My initial data showed that White Plains was “spending” more per capita than any other city in NY – over $8k, almost double the next city. I looked more closely and noticed that White Plains and several other cities in Westchester and Nassau counties collect taxes on behalf of the county, school districts, and special districts, then pass those dollars along. That’s not spending but I was counting it as such. Once I pulled those custodial pass-throughs out, White Plains’ per-capita spending dropped to a more reasonable level, ranking it #4 instead of #1.

Interfund Transfers

I also found a second problem: interfund transfers. The data I’m working with comes from the Office of the New York State Comptroller (OSC), which reports finances at the fund level, not the consolidated government-wide level that cities publish in their ACFRs. When a city moves money from its General Fund to its Debt Service Fund, that shows up as an “expenditure” in one fund and “revenue” in another. Sum across all funds without eliminating these internal transfers and you double-count. Across all cities and years in the database, that’s $10.5 billion in phantom spending.

The Plattsburgh Anomaly

A particularly weird manifestation of this: I highlight the percent of total annual spending that a community spends on debt service, which seems like a good metric for how well a community has been living within its means. Plattsburgh, NY was showing debt service at 156% of annual spending – a mathematical impossibility. The problem was that my code was only counting General Fund expenditures in the denominator while including debt service from all funds in the numerator. Plattsburgh books all of its debt in non-General Fund accounts, so the denominator was artificially small. Once I included all funds (minus the custodial pass-throughs and interfund transfers), the number dropped to a still-high-but-real 38.7%.

Time for an Audit

I’m glad I was able to find and fix these outliers but it raises the question: what other errors are lurking in my data? I don’t want to be like the Beardstown Ladies Investment Club who claimed in their 1995 bestseller on investment advice that they earned a 23.4% annual return since inception. Three books later an audit showed their actual return was 9.1%, well below the S&P 500 returns for the time period. Time for an internal audit to see if my numbers match what the cities report in their own ACFRs rather than my interpretation of what the OSC is collecting and publishing on the cities’ behalf. See details of what I’ll be looking for in the app’s AUDIT.md file.

I’ll update what I find here. Fingers crossed that my numbers aren’t too far off – and if they are, better to find out now than after someone makes a decision based on bad data.

This project starts from a simple contention:

Government at the local, county, and state level can be more effective, more efficient, and more affordable if it is more transparent and if we systematically compare outcomes across jurisdictions.

The New York Benchmarking Project exists to make that comparison possible.

Why Benchmarking?

Benchmarking is the practice of comparing organizations performing similar functions in order to identify best practices, outliers, and inefficiencies. It is routine in private industry, medicine, logistics, and engineering. It is far less routine in local government, where fragmentation, nonstandard reporting, and political incentives often obscure comparison.

Yet New York and states around the country contain a natural laboratory. Here we’ve got:

• 62 cities, ranging from very small to very large
• 62 counties (a numerical coincidence, not a structural symmetry)
• Hundreds of towns, villages, and special districts
• Largely shared legal frameworks
• Common state oversight and reporting requirements

If one city delivers public services at lower cost with equal or better outcomes, that fact should be visible. If another city spends far more for worse results, that too should be visible.

Benchmarking does not dictate policy. It creates the conditions for informed judgment.

The Problem: Transparency but Fragmentation

New York and our Office of the State Comptroller is actually a national leader in collecting and disseminating municipal financial data: Much of the data is available but it’s not standardized or designed to compare one jurisdiction with another. The OSC also doesn’t provide demographic information like population, education, median home values and crime rates. Those details come from other databases like the US Census office and the Department of Justice.

As a result, even engaged citizens, journalists, and policymakers struggle to answer basic questions such as:

• How much does this city spend per resident on core services?
• How do employee compensation and benefits compare to peer cities?
• Are debt levels and long-term liabilities out of line with similar communities?

Voice, Exit, and Decline

When local governance performs poorly, residents have three broad responses: Exit, Voice, and Loyalty.

New York has increasingly seen the former.

Since the early 20th century, New York’s share of the national population has declined relative to other states. This shift is reflected in the state’s representation in the U.S. House of Representatives, which has steadily decreased as population growth has moved elsewhere (see Figures 1 and 2). While many factors contribute to this trend, governance quality and cost of living influence whether people choose to stay, invest, and raise families.

Figure 1. New York’s share of seats in the U.S. House of Representatives relative to other states, illustrating a long-term decline in national representation.

Figure 2. Absolute number of U.S. House seats held by New York since the House was standardized at 435 members.

This project is premised on a belief that decline is not inevitable. Better information enables better choices—and better governance.

What This Project Will Do

The New York Benchmarking Project aims to:

1. Collect and document financial and governance data starting with the 62 cities in New York State and moving on to its counties, towns, villages, districts, and authorities.
2. Publish comparable metrics that allow meaningful benchmarking across jurisdictions.
3. Explain methods and assumptions transparently, so conclusions can be challenged and improved.
4. Support civic education, public discussion, and evidence-based reform.

All work is intended to be transparent and verifiable. All claims are intended to be traceable.

What This Project Will Not Do

• Advocate for a specific political party or ideology
• Replace democratic decision-making
• Reduce complex policy choices to a single metric

Data informs judgment; it does not substitute for it.

Where This Is Going

This site will publish:

• Essays explaining the motivation and methodology
• Notes on data sources and limitations
• Early findings and comparisons as they emerge

A public data explorer is in development and will be made available as the dataset matures.

The work will be slow, careful, and incremental by design.

New York deserves government that works as well as its people do. Benchmarking is one step toward that goal.