2026 employer tax source ledger

What has changed in this dataset

Version 2026.09.21, covering 50 states and the District of Columbia, effective January 1, 2026. Every correction below names what changed, what it cost an employer, and the official document it came from.

• September 21, 2026 · correction · NJ, DC, VT, IL, MA

  Five cited sources were dead. On a site whose selling point is explicit sourcing, that is the cheapest way to lose a reader.

  Opening all 78 URLs the dataset cites found five returning 404: the New Jersey employer page (used for both its unemployment rate and its TDI employer contribution), the DC Paid Family Leave employer page, the Vermont Child Care Contribution page, the Illinois employer page and the Massachusetts employer contribution page. Two of the five had been added the previous day from an audit agent's output without anyone opening the URL. All five now point at live official pages, each confirmed to serve the figure it is cited for -- the Massachusetts replacement shows the same 2.42% new-employer rate the dataset records. A checker now opens every cited source on demand. It deliberately does not treat a 403 as dead: nineteen state agencies refuse a script while serving the page perfectly to a browser, and calling those broken would bury the real failures in noise.

  No rate changed. Five sources that could not be checked by a reader now can be.

  Source: Each replacement is cited on the state page it belongs to, published September 21, 2026, read September 21, 2026.

• September 20, 2026 · disclosure · OR, PA

  Oregon's two transit payroll taxes are employer-paid and were missing. Locality costs are now listed, never summed.

  The Oregon Department of Revenue states the transit tax is imposed directly on the employer: TriMet at 0.8237% and Lane Transit at 0.80% of payroll for work performed inside each district, no wage cap, about $494 and $480 a year on a $60,000 salary. Both were absent. Adding them naively would have been its own error: an employer is inside one district or the other, never both, and most Oregon employers are in neither, so summing all three of Oregon's costs would have published $1,214 that nobody pays. Every omitted cost now declares whether it is statewide or limited to a locality; only statewide costs are summed. Pennsylvania's Pittsburgh tax was reclassified the same way, and its source moved from a forms index that never stated the rate to the city's own payroll tax regulations, which do. The Denver and St. Louis employer taxes were left out: both city sites refused connection from here and neither figure could be verified against a primary source.

  Oregon declares two more costs, both marked as district-only. No headline total rose: Oregon's statewide figure stays $240 and Pennsylvania's is zero with one Pittsburgh-only cost listed beside it.

  Source: Oregon Department of Revenue: A guide to TriMet and Lane Transit payroll taxes, published November 18, 2025, read September 20, 2026.

• September 20, 2026 · correction · ALL

  The workers compensation multiplier was the one per-state cost with no source. It was invented, and for nineteen states it was wrong by more than a quarter.

  Every state carried a bare number between 0.92 and 1.22, clustered on 1.0, with no sourceUrl while every unemployment record had one. It multiplied a base rate into a dollar figure the reader saw, exported to CSV and printed to PDF. Measured against Oregon DCBS's biennial premium rate ranking -- the standard public cross-state comparison, which publishes each state's manual rates as a percent of the study median -- the mean relative error was 25%, nineteen of 51 states were off by more than a quarter, and the rank correlation was 0.40, so even the ordering was barely better than noise. Hawaii carried 1.18 against a real 2.31. North Dakota carried 1.02 against 0.45. The District of Columbia carried 1.12, above the national average, against a real 0.67, well below it. All 51 now use the published figure and each record carries the index rate, the national rank and the source. The study is the 2024 edition published June 2025, the most recent available; the dataset says so rather than implying a 2026 figure.

  Workers compensation estimates changed in all 51 jurisdictions. Hawaii roughly doubled; North Dakota, Arkansas and West Virginia roughly halved.

  Source: Oregon DCBS: Workers' Compensation Premium Rate Ranking, Table 1, published June 1, 2025, read September 20, 2026.

• September 19, 2026 · disclosure · CO, DC, DE, HI, MA, ME, MN, NJ, NM, NV, NY, OR, PA, VT, WA

  Fifteen states have mandatory employer costs this calculator does not charge. They are now named instead of left silent.

  An audit of all 51 jurisdictions confirmed 17 mandatory, employer-paid costs the model had no field for: state paid-leave employer shares, payroll taxes, per-head assessments, and in Hawaii a statutory health premium of roughly $7,385 a year per employee. They are disclosed, not added. Sixteen of the seventeen depend on a fact the calculator never asks for -- headcount, city, quarterly payroll -- so charging every employer would overcharge the ones the rule never reaches. Each is published with its size at a $60,000 salary, the condition that triggers it, and its official source.

  No total changed. Fifteen state pages and the calculator now name what they leave out; the other 36 jurisdictions were swept and are clean.

  Source: Per-cost official sources are published beside each entry on the state pages, published September 19, 2026, read September 19, 2026.

• September 19, 2026 · correction · NY

  New York's unemployment wage base was a figure no published year ever used, and its rate was only part of what the employer pays.

  The dataset carried a $12,900 New York taxable wage base. The 2025 base was $12,800 and the 2026 base is $17,600, so $12,900 matched neither. Checked against an independent 51-jurisdiction list, New York was the only wrong wage base in the dataset -- the other 50 agree. The rate was wrong in a second, separate way: 3.4% is New York's normal contribution rate, a component, while the employer actually remits 4.025% plus the 0.075% Re-employment Service Fund, a total of 4.1%. Both figures were on the page the dataset already cited. The experienced range was corrected to the 1.7%-9.5% that same page publishes.

  A New York employee's unemployment line rose from $438.60 to $721.60, a 65% understatement corrected. No other state changed.

  Source: New York State Department of Labor: Unemployment Insurance Rate Information, published January 1, 2026, read September 19, 2026.

• September 19, 2026 · correction · CA

  California employers were quoted the headline 0.6% FUTA rate. They do not pay it.

  California has owed a Title XII federal unemployment advance since 2021, so its employers lose part of the 5.4% FUTA credit. The dataset had no concept of a credit reduction at all: every calculator multiplied the $7,000 wage base by a flat 0.006. The 2026 planning reduction is 1.5%, giving an effective rate of 2.1%. The U.S. Department of Labor also publishes an estimated 3.8% BCR add-on that would take the rate to 5.9%; California has received add-on relief every year since 2014, so the add-on is recorded and published but not applied. Nothing is final until November 10, 2026.

  A California employee's FUTA line rose from $42 to $147. Every other jurisdiction is unchanged.

  Source: U.S. Department of Labor, Division of Fiscal and Actuarial Services: FUTA Credit Reductions, published September 11, 2026, read September 19, 2026.

This record starts on 2026-09-19. Changes made before that date were not logged, so the absence of an earlier entry means nothing was recorded, not that nothing changed.

How to use this ledger responsibly

EmployerCost normalizes public information for planning. Experience ratings, industry schedules, assessments, successor rules, and employer-specific notices can change the payable rate. When an assigned notice exists, use it.

See the calculation methodology, the SUTA guide, or open a state-specific guide.