Benchmarking your agency against other jurisdictions

Reference article · 5 min read

A council member or a department head asks how the agency compares to similar jurisdictions. The question is reasonable, and the answer is harder than it looks. The figures easiest to find are the ones most likely to be defined differently in every jurisdiction reporting them. Staffing counts, cost per capita, and response times depend on what is counted, who counts it, and which services the agency provides at all.

A comparison that survives challenge does three things. It names the peer set and explains why those peers were chosen. It normalizes each figure against a denominator that matches the service. It documents the definition behind every number so a skeptical reader can check it. Settle first what decision the comparison supports, since a staffing request, a rate study, and a service level debate call for different metrics and different peers.

Select peers you can defend

Peer selection is the step most often skipped and most often attacked. Choose on characteristics that drive the cost or workload of the service in question, not on general similarity or geographic convenience.

• Functional responsibility. The peer must actually provide the service. A city policed by a county sheriff, or served water by a district, is not a peer for those functions.
• Scale and structure. Population, households, service connections, enrollment, or calls for service, whichever drives the work. Whether the service sits in an enterprise fund matters for cost comparisons.
• Operating conditions. Density, geography, seasonal population, age of infrastructure, and local labor market.

Write the criteria down before pulling data, and drop peers that publish nothing comparable. Use six to twelve peers rather than two or three, and keep the jurisdictions that flatter the agency along with those that do not. A peer set assembled after the numbers are known is indefensible.

Normalize for population, service area, and service level

Per capita is the default denominator and frequently the wrong one. Match the denominator to what generates the work.

• Residents for general government, parks, and libraries.
• Service population where daytime or seasonal population differs materially from the resident count.
• Units of infrastructure such as lane miles, accounts, meters, square footage maintained, or devices supported.
• Demand events such as calls for service, permits issued, cases opened, or tickets closed.

Then adjust for what differs across places and years: fiscal year alignment, inflation to a common dollar year, and regional wage differences. Separate operating from capital spending, and state whether pension contributions, other post-employment benefits, internal service charges, and grant funded positions are included. Service level is the hardest axis, since two agencies can spend the same per resident and deliver different hours, coverage, and response standards.

Know the public data sources

Most of what is needed is public, though often a year or more behind.

• Census Bureau. Population estimates and the American Community Survey for denominators; the annual surveys of government finances and of public employment and payroll for spending and staffing by function.
• Bureau of Labor Statistics. Occupational wage data for labor market context, and price indexes for converting to constant dollars.
• Agency financial reports. Each jurisdiction's annual comprehensive financial report and adopted budget, plus state comptroller or auditor portals that collect local data on a common schedule.
• Program specific collections. The National Transit Database, Highway Statistics for roads, national incident based crime and fire reporting, the Common Core of Data for schools, and drinking water reporting for utilities.
• Association programs. Professional associations run voluntary comparison programs using agreed definitions, narrower in coverage but more comparable.

Definitions and coverage change, so confirm current details at the source before publishing.

Understand why raw cost and staffing figures mislead

Cost and staffing look objective and are among the least comparable numbers published.

• Contracting differs. An agency that contracts dispatch, fleet, or IT shows fewer employees and higher contractual spending for identical service.
• Account structures differ. Overhead allocation, internal service funds, and the capital versus operating line shift large sums.
• Counting differs. Headcount, budgeted positions, filled positions, and full time equivalents are four different numbers.
• Boundaries differ. Consolidated city county governments, dependent districts, and authorities move whole functions off the books.
• Reporting practices differ. Much of this data is self reported by staff working from their own reading of the instructions.

None of this makes comparison useless. It makes undocumented comparison useless.

Present it so it survives review

Write a short methodology note that travels with every chart: peer criteria, source and vintage of each figure, the denominator, adjustments applied, and known limitations. Show the range and the middle of the peer distribution rather than a single average, since one outlier moves a mean. Where a figure is not comparable, say so and say what would make it comparable. Volunteering the weaknesses in your own analysis beats defending a clean number afterward.

What to ask vendors, and how to start

Judge benchmarking products and consulting studies on their sources, not their dashboards.

• Ask which fields come from public collections, which from surveys of participating agencies, and which are modeled.
• Ask for the data dictionary and the peer selection method in writing before award.
• Ask how far behind the most recent published year is, and how often data refreshes.
• Ask whether normalization is applied automatically and whether raw values stay visible.

In the contract, require the data dictionary and methodology as a deliverable, the right to reproduce results in public meeting materials, notice when methodology changes between refreshes, and export of the agency's data and the peer figures in an open format. Without a vendor, start with one service, one clear question, and a documented peer set.

HQT is a technology and software company serving state and local government agencies, with work spanning data analytics, reporting, and systems engineering.

Frequently asked questions