Quick answer: In July, without announcement, the Department of Revenue revised its estimate of who will pay Washington's 9.9% income tax from roughly 21,000 households to roughly 25,000, and raised projected first-year collections from $2.7 billion to $3.1 billion. The Department attributes about 4,000 of the additional households to income growth pushing filers who are currently below $1,000,000 above it by the time the tax takes effect. That is a threshold-drift finding, and it matters well beyond this one estimate: under sections 314 and 316 of ESSB 6346, the $1,000,000 standard deduction is not adjusted at all until October 2029, and thereafter only in odd-numbered years, using a national inflation measure. The number of households caught by this tax is designed to grow.
This post is part of our Complete Guide to Washington's New Income Tax.
01. What the Department actually revised
When Governor Ferguson signed ESSB 6346 in March 2026, the fiscal assumptions behind it were built on the November 2025 revenue forecast: roughly 21,000 filers, $2.7 billion in the first budget cycle, and $6.9 billion in the following biennium.
In July, the Department of Revenue produced new figures for the Office of Financial Management's fiscal impact statement on Initiative 645 — the measure to repeal the tax, which appears on the November 3, 2026 ballot. Those figures, derived from the June 2026 forecast rather than the November 2025 forecast, put first-year collections at $3.108 billion and the 2029–31 biennium at $7.947 billion to the state general fund — or $8.324 billion once the fair start for kids account is included. Across the full five fiscal years the statement covers, the total is $11.4 billion.
Two features of how this surfaced are worth noting.
First, the revised revenue numbers appear in a table inside the fiscal impact statement — Table 1, "Revenue loss, by account" — a document written for voters, rather than in any standalone Departmental release. The Department maintains an income tax page and a public explainer, The Facts on the Department's Costs to Implement Washington's Income Tax on Millionaires. Neither carries the revision. As of this writing, the Facts page still describes "potentially 21,000 taxpayers subject to it" — the superseded figure, on the Department's own site, while the Department was giving reporters 25,000.
Second, the revised household count of 25,000 appears nowhere in the fiscal impact statement. It emerged only in response to press questions. The only headcount the statement contains is a single passing clause — savings are assumed for tax administration duties for "approximately 30,000 taxpayers that will no longer be required" — a fifth number, in an administrative-savings context, unexplained and nowhere reconciled to the 25,000. The two may well be measuring different things; the statement does not say.
Neither point suggests anything improper. Fiscal impact statements are exactly where the Office of Financial Management publishes this material. But it does mean the most planning-relevant number in the revision was not published anywhere a practitioner would think to look, and the one headcount that was published does not match it.
One wrinkle remains unresolved. Operating budgets must rest on forecasts adopted by the Economic and Revenue Forecast Council, and the June 2026 forecast still used the old assumptions. September is the first forecast that could adopt the Department's higher figures. Dave Reich, the Council's chief economist, has said only that his office "will be taking a look at how they forecasted the tax," and that "our forecast may or may not match depending on our assessment of the method, risk, etc." For now the state is carrying two different numbers for the same tax.
02. The 4,000-household finding is the one that matters
The revenue increase is the headline, but it is the least durable part of the revision. Revenue estimates move with forecasts; the Department was explicit that had the June forecast been negative, these numbers would have gone down instead.
The household finding is more interesting, though not for the reason it might first appear. The Department's explanation is that dividend income, stock income, personal income, and wages all grew between the November and June forecasts, and that state analysts concluded roughly 4,000 households currently below the $1,000,000 threshold will be above it by the time the tax takes effect on January 1, 2028.
Be precise about what that does and does not show. It does not show that a new phenomenon appeared between November and June. The 21,000 figure already embedded an assumption about income growth between the estimate and the 2028 effective date; the revision reflects a higher assumed growth rate applied to the same fixed line, not the discovery of drift. Both the revenue number and the household number moved for the same reason, and both would have moved down had the June forecast been worse.
What it does show is the mechanism operating in plain view. On the Department's own account, those 4,000 households are not becoming wealthier in real terms. They cross the threshold because nominal income rises and the threshold does not. That is ordinary bracket creep, and it is unremarkable as economics.
The reason it deserves a practitioner's attention is not the revision. It is that ESSB 6346 does comparatively little to arrest the mechanism the revision happens to illustrate, and the drafting is more restrictive than the headline "indexed for inflation" description suggests. Whatever the Council adopts in September, sections 314 and 316 will keep producing this result.
03. What sections 314 and 316 actually provide
Section 314 sets the deduction: $1,000,000 per individual, or a combined $1,000,000 for spouses and state registered domestic partners regardless of whether they file jointly or separately. (The combined-deduction rule is a separate and substantial problem, addressed in our analysis of the marriage penalty.) Section 314 then states that the deduction "must be annually adjusted pursuant to section 316 of this act."
Section 316 does not deliver annual adjustment. It provides:
Beginning October 2029 and each October of an odd-numbered year thereafter, the department must adjust the standard deduction under section 314 of this act by multiplying the current standard deduction amount by one plus the percentage by which the most current consumer price index available on October 1st of the current year exceeds the consumer price index for the prior 12-month period, and rounding the result to the nearest $1,000.
Three consequences follow, and each cuts in the same direction.
The first tax year is unindexed. The tax applies beginning January 1, 2028. The first adjustment occurs in October 2029 and, under section 316(1), "takes effect for taxes due in the following calendar year" — that is, calendar 2030, which is the payment year for tax year 2029. Tax year 2028 therefore runs on a flat, unadjusted $1,000,000. Whatever nominal income growth occurs between now and the end of 2028 accrues entirely against a fixed threshold.
Adjustment is biennial, not annual. "[E]ach October of an odd-numbered year thereafter" produces adjustments in 2029, 2031, 2033, and so on, each effective for the following payment year. On the better reading, that means the deduction is adjusted for odd tax years and carries forward unchanged into even ones. Two things in the act cut against this reading and should be acknowledged. Section 314 states that the deduction "must be annually adjusted pursuant to section 316." And section 316(1) itself directs that "[t]he department must publish the adjusted standard deduction amount on its public website by October 31st of each year" — annual publication, which sits more naturally alongside annual adjustment.
Both are reconcilable with a biennial mechanism. Section 314's "annually" is a cross-reference to section 316 rather than an independent command, and section 316 is where the operation is actually specified. Annual publication is consistent with the Department restating an unchanged figure in even-numbered years, which taxpayers would need in any event. But the reconciliation is not so obvious that a contrary Departmental reading would be unreasonable, and the phrase that controls the outcome — "each October of an odd-numbered year" — governs when the Department adjusts, not when it publishes. Our view is that biennial adjustment is the better reading. It is not the only available one, and this is the most likely candidate in the act for a corrective amendment or a clarifying rule. Practitioners should not assume annual indexing merely because section 314 uses the word, and should not treat the question as settled either way until the Department speaks.
The measuring period does not cover the gap. The formula compares "the most current consumer price index available on October 1st of the current year" against "the consumer price index for the prior 12-month period" — approximately one year of inflation. Applied on a two-year cycle, a substantial portion of accumulated inflation is never captured, and the shortfall compounds.
A fourth point is smaller but worth flagging. Section 316(2) defines the index as "the consumer price index for all urban wage earners and clerical workers as calculated by the United States bureau of labor statistics" — with no geographic qualifier. Compare section 901(2)(e) of the same act, which for working families' tax credit purposes specifies "the average consumer price index for that 12-month period for the Seattle, Washington area." The legislature knew how to specify a regional index and did so elsewhere in the same bill. Whether the omission in section 316 was deliberate is unknowable from the text, but the practical effect is that a threshold applied disproportionately to Puget Sound incomes is adjusted by a national measure.
Finally, section 316(1) creates a one-way ratchet: "If an adjustment under this subsection (1) would reduce the standard deduction amount, the department must not adjust the amounts for use in the following year." That protects taxpayers in a deflationary year, which is the right result, and it is the only asymmetry in the mechanism that runs in the taxpayer's favor.
04. What the Department and the tax's defenders would say
The strongest response to all of this is that it proves very little. The Department stressed that these are estimates, that no collections data exists yet, and that the modeling simply reflects a better economic forecast. All of that is true and none of it is evasive. A revenue estimate that did not move with the forecast would be the anomaly.
On the indexing design, a defender would point out that biennial adjustment is not unusual in state tax codes, that it tracks the state's biennial budget cycle, and that a $1,000,000 threshold indexed imperfectly still excludes the overwhelming majority of Washington households. The legislature's own findings in section 1 estimate the tax reaches "the wealthiest one-half of one percent" of households in the state. Some drift at that threshold, the argument runs, is a rounding error against the policy objective, and a taxpayer with $1,000,000 of adjusted gross income is not the sympathetic figure that bracket-creep arguments usually invoke.
There is also a fair criticism available from the other direction. Brian Heywood, who leads Let's Go Washington and sponsored the repeal initiative, argues the Department's model wrongly assumes a static population: "They think nobody is going to leave because of the tax. They're wrong." That critique has some force as a matter of modeling, though it cuts against the 25,000 figure being too high, not too low, and it is not obvious that migration effects would offset threshold drift within the first two years.
Our own view is narrower than any of these, and deliberately so. We take no position on whether the revenue estimate is right, and we accept that the revision itself is a forecasting event rather than a discovery. The claim here is only about the mechanism the revision puts on display: it is written into sections 314 and 316, it runs in one direction, and no forecast the Council adopts in September will change it. Whether 21,000 households or 25,000 pay in 2028, the threshold that determines which ones is frozen through that tax year and adjusted on a cycle that will not keep pace afterward.
05. What to do with this
Treat $1,000,000 as a receding threshold, not a fixed one. If you are modeling 2028 exposure for a client currently at $850,000 to $950,000 of federal adjusted gross income, the relevant question is not whether they are under the line today. It is whether nominal income growth — including raises, dividend growth, and portfolio income — carries them over a line that does not move at all before tax year 2029. Four thousand households are, on the Department's own analysis, in exactly that position.
Married couples should model the combined deduction, not the individual one. Two unmarried individuals each receive a $1,000,000 deduction. A married couple receives $1,000,000 between them, whether they file jointly or separately. For couples near the threshold this is the single largest driver of exposure, and it interacts badly with threshold drift.
Do not defer income into 2028 on the assumption the threshold will rise. It will not — not for tax year 2028, and on the better reading of section 316, not for tax year 2030 either. Cliff planning that assumes an indexed threshold is planning against the wrong statute.
Watch the September forecast, but do not plan around it. If the Economic and Revenue Forecast Council adopts the Department's figures, expect renewed attention to the tax's revenue potential in the run-up to the November 3 vote. If it declines to, expect the discrepancy itself to become an argument. Neither outcome changes a client's liability. What changes liability is Initiative 645 and the pending constitutional challenge — and until one of those resolves, the Before-2028 playbook is unchanged.
Update, August 19, 2026. This post has been corrected. The 2029–31 biennial figure originally appeared as "roughly $8.3 billion" without specifying the accounts included; Table 1 of the fiscal impact statement shows $7.947 billion to the state general fund and $8.324 billion including the fair start for kids account. The post originally stated that the revised household count appears nowhere in the fiscal impact statement; the statement contains no household count, but it does refer to "approximately 30,000 taxpayers" in an administrative-savings context, which is now noted. Source links have been added throughout.
Authority level. The reading of sections 314 and 316 above rests on the enrolled text of chapter 238, Laws of 2026; see also the Final Bill Report. No Department rule, interim guidance, or form has issued interpreting either section, and the Department's 2026 tax legislation page lists no income tax guidance. The 19-member advisory workgroup created by section 712 has not yet reported; its initial report is due December 15, 2026 and its final report December 15, 2027. Its meetings are public and carried on TVW, and are the most likely near-term source of any Departmental signal on the questions raised above. On the biennial-versus-annual question, two provisions cut against our reading — section 314's "annually adjusted" language and section 316(1)'s direction to publish an adjusted amount each October 31 — and while we think biennial adjustment is the better reading of the operative text, we would not characterize it as free from doubt. A contrary Departmental interpretation would be defensible, and this is a likely candidate for a clarifying rule or a corrective amendment. Treat the analysis here as substantial authority for planning purposes and revisit it when the Department begins rulemaking.
This post is for educational purposes only and is not legal or tax advice. Consult a qualified attorney about your specific situation.
