The universal service contribution factor hit 38.8 percent as the assessable revenue base fell 45 percent in nine years

What is happening?

The federal universal service contribution factor for July through September 2026 is 38.8 percent, proposed by the FCC in Public Notice DA 26-546 released 2026-06-12 and deemed approved fourteen days later unless the Commission acts. That is the highest value in the quarterly series published by the fund administrator, which stood at 18.2 percent in the first quarter of 2016 — a rise of 2.13 times in ten and a half years, computed here from the two published quarters.

The number is a cap on a line item, not a share of a telephone bill. Under 47 CFR 54.712, restated in the notice, a carrier may not recover through a federal universal service line item an amount exceeding 38.8 percent of the interstate telecommunications charges on a customer bill, and the interstate portion of a wireless or internet-voice bill is itself a fraction of that bill, fixed by safe harbor or by a carrier traffic study rather than measured. What the fund actually takes is about 27.7 percent of projected collected assessable revenue — a requirement of USD 2,003.130 million against a projected base of USD 7,221.914 million for the quarter, both printed in the same notice.

The rise is a denominator story, and the three bodies that have examined it say so. Between 2015 and 2024 the annual contribution base fell from USD 60,290 million to USD 32,859 million, a drop of 45.5 percent computed from the published rows. Over the same nine years fund disbursements went from USD 8,372 million to USD 8,595 million, up 2.66 percent in nominal terms and down in real terms. The same filers reported total revenue that grew 12.11 percent across the period. What moved was the mix — assessable telecommunications revenue fell 49.29 percent while non-assessable non-telecommunications revenue rose 55.85 percent, all on the same annual form.

One figure in the current notice is missing from the notice itself. The Commission instructed the administrator to apply USD 125 million in unused funds against projected Schools and Libraries demand, and states that this offset reduces the third quarter factor below the expected factor based on the filings of the administrator. It never prints the expected factor. Recomputing it here from the formula and the unchanged inputs in the same four pages gives about 42.2 percent, roughly 3.4 percentage points above the number that was announced. No opened source states this.

Whose problem is this?

The body that computes the rate does not choose it. The rate is a residual of two quantities it does not control in the same quarter, and the notice that announces it becomes effective by default if nothing happens.

Where does this problem end?

The boundary matters here because nothing is broken and nothing is being evaded. What changed is which side of a statutory line the money sits on, and the line did not move.

What is the state now, and what should it be?

Now

Needs a new measurementthe target state: no opened document names a target. The FCC adopted a goal in 2011 to minimize the burden of the contribution factor on consumers and businesses, and the accountability office reported that the goal carries no measure meeting user-fee practice, that it recommended a fix in October 2020, and that the recommendation was still unimplemented as of May 2024. No source opened here states a ceiling on the factor, a level that would trigger review, or a projection for any future quarter.

How big is it?

132.3 million US households, as of July 2025 — the households reported with a telephone in unit, 97.7 percent of the 135.5 million US households counted in the same table. Any such household whose service carries interstate or international telecommunications charges is subject to the line item capped at the contribution factor.

This is deliberately a single figure rather than a band. The count of payers is not where the uncertainty lives. The uncertainty lives in how much of the levy each household actually bears, and the Commission states plainly that it does not collect the data that would settle it.

Where the range does belong is the amount. Total universal service contributions came to USD 5.37 per household per month in 2025 in constant 2025 dollars, of which the residential portion is estimated at USD 1.88 to USD 2.41 using the staff band of 35 to 45 percent residential. Annualized, that is USD 22.56 to USD 28.92 per household per year, computed here. A person count is not derivable from anything opened this round, because no opened source gives an average US household size.

One figure cuts against the obvious reading. That same per-household total was USD 8.67 in 2011 and USD 5.37 in 2025 in constant dollars, so the fund costs a household less in real terms than it did fifteen years ago while the rate more than doubled. The instability is not that households pay more. It is that a flat, shrinking-in-real-terms bill is being extracted at a doubling rate from a collapsing base.

Under what conditions does it arise?

1. The levy is defined by a legal category rather than an economic one. The statute binds every telecommunications carrier that provides interstate telecommunications services. What gets assessed is therefore whatever falls inside the statutory word telecommunications, not whatever earns revenue from carrying data.

2. The revenue crossed that line while the line stayed put. Same filers, same annual form: assessable telecommunications revenue fell from USD 214,477 million to USD 108,759 million between 2015 and 2024 while non-telecommunications revenue rose from USD 301,121 million to USD 469,302 million. The FCC examined whether providers were reclassifying interstate revenue as intrastate and concluded they were not — the shift is in the share of revenue reported as non-telecommunications.

3. The named carve-outs are not the mechanism. In 2024 the four explicit deductions came to USD 1,436 million against a gross of USD 34,294 million, about 4 percent. Removing all four would move the factor by a few points. The exemption that matters never appears on a deduction line at all, because USD 469,302 million of revenue from the same filers is classified outside the category and so is never a candidate for the base. That is 13.7 times the entire assessable gross.

4. The rate is a residual, so it rises without anyone deciding. The factor is the ratio of projected quarterly program cost to projected collected assessable revenue. Nothing caps it, targets it, or triggers a review when it moves. Nobody has to decide to raise it, so nobody has to defend raising it.

5. The mechanism never runs out of money. The fund is a permanent, indefinite appropriation and always collects exactly what it requires, because the rate is solved for. There is no moment of breakage and therefore no forcing event.

What has been tried?

Thirty years of mechanism have produced one exercise of the permissive authority, one reform proceeding with no order, one mandated report that recommended more evaluation, one bill sitting in committee, and a discussion draft still pending.

What was found?

Why is it still unsolved?

Institutional exemption — the growing side of the industry sits outside the obligation by definition, and the definition is held in place by a statute and a court rather than by anyone who could be persuaded.

The first movement is that the carve-out is invisible because it is not written as one. Someone reading the base table sees four deductions and correctly judges them small. What that reader cannot see is the revenue that never reaches the table at all. Nearly half a trillion dollars of revenue from the same filers, on the same annual form, is classified as non-telecommunications and is therefore not a candidate for the base, not a deduction from it, and not a line anywhere in the accounting of the fund. An exemption written into a definition leaves no trace in the arithmetic it exempts you from.

The second movement is that the rate absorbs the shock silently. The factor is computed as a ratio and announced quarterly, and it becomes effective by default fourteen days after release. The fund always gets what it requires because the rate is whatever makes that true. So the collapse of the base produces no shortfall, no emergency, no vote, and no headline. It produces a slightly larger number in a four-page notice, and the current notice shows how much even that number can be managed — a one-time offset lowered the announced figure by roughly 3.4 points, and the notice says the announced figure is below the expected one without printing the expected one.

The third movement is that the exit is legally available and institutionally unaffordable. The mandatory limb of the statute cannot reach broadband, because a court held in January 2025 that broadband internet access is an information service and decided it without deference, which forecloses reinterpretation. The permissive limb is still open, and using it would require the Commission to find that the services meet the statutory definition and that the public interest supports the contribution. It has not used that power in twenty years, and in the same 2022 report where it diagnosed the problem it advised itself to avoid raising the cost of broadband and shifting the burden from corporations to consumers. Notice what that framing does. Adding a category to the base is scored as a cost increase, while concentrating the same fund on a base half the size is not scored at all. Congress holds the clean instrument, has had a bill in committee since May 2025, and has moved its own discussion draft from early 2026 to summer 2026.

What observation would mean it is solved?

Candidates — (a) the assessable share of total reported revenue from the same filers stops falling (b) the contribution factor falls and stays down across several consecutive quarters (c) the base is broadened by rule or by statute with a stated effective date.

(a) alone is slow and ambiguous. The ratio is computed from annual filings published with a lag approaching a year, so it cannot confirm anything about the current quarter. It can also stabilize for a reason that has nothing to do with the levy, if the non-assessable side simply stops growing. That would describe the technology market, not the arrangement.

(b) alone is the weakest. The factor is a residual of two moving quantities, so it falls whenever demand falls or a one-time credit is applied, with the base unchanged. The current notice is the demonstration: a single USD 125 million offset moved the announced number by roughly 3.4 points while nothing about the erosion changed. A run of lower quarters is fully consistent with a base that is still collapsing.

(c) alone counts paper. A statute that commands a rulemaking is not a rule, and the pending bill carries an exemption for edge providers below both a data-share threshold and a revenue threshold, plus a rule of construction disclaiming any new authority over broadband providers. A rule can also be narrowed in litigation, and the classification holding stands until Congress moves it. The three have to be read together, and (b) has to be read against what happened to demand that quarter.

What is it connected to?

Fills with researchstate universal service funds and their own contribution bases, the history of emergency broadband subsidy programs and how they were funded, spectrum auction receipts as an alternative revenue route, and comparable telecommunications levies in other countries. Relation type and evidence grade were not confirmed in this round.

What these sources do not say

• The number the notice says it suppressed. The current notice states that a USD 125 million offset put the factor below the expected level and then never prints the expected level, although every input needed to compute it is on the same four pages. Recomputed here it is about 42.2 percent. No opened source states it.
• What share of the levy households actually bear. The table that translates the fund into a household figure carries its own note saying the Commission does not currently collect data that would allow the residential amount to be calculated accurately. What it publishes instead is a staff belief — 45 to 55 percent residential for 2011 through 2017, then 35 to 45 percent since 2021. The band moved by ten points and nothing in the report says why or on what evidence either band rested.
• Three years are simply missing from that table. It has rows for 2011 through 2017 and then for 2021 through 2025. The source note enumerates provenance year by year, ends at 2017, resumes at 2021, and never accounts for the interval. Those are the years that contain the pandemic and the largest single-quarter jumps in the history of the factor.
• Any target at all. No opened document — from the Commission, the administrator, the accountability office or the research service — states a maximum acceptable factor, a level that would trigger review, or a projection for any future quarter. The goal adopted in 2011 was to minimize the burden, and the recommendation to give that goal a usable measure was still open four years after it was made.
• How often the arrangement already breaks. The notice concedes that because the factor exceeds 12 percent, a contributor under the limited international revenue exemption can be required to contribute more than its entire interstate end-user telecommunications revenue, and offers a case-by-case waiver. It does not say how many contributors are in that position, how many have petitioned, or how many waivers were granted.
• How much of the base is a guess. The base includes estimated revenue for 202 providers that did not file, out of 3,397. The dollar magnitude of that estimate is never disclosed, so the share of the USD 7.221914 billion base that is estimated rather than reported is unknown to any reader.
• That the published base is the one used. The footnote to the annual base table lists six reasons the published figure differs from the amounts actually used and concludes that actual contribution bases have been based on different amounts than those shown. The report never publishes those amounts.
• Anything about the base in the newest proceeding. The universal service notice released three weeks before this round proposes audit sampling, a pay-and-dispute model, an administrative budget cap and a smaller administrator board. The contribution base is not among the reported proposals.

See the evidence

Every quantitative claim above rests on a document that was opened in full. The public notice, the base filing of the administrator, the annual monitoring report, the 2022 report to Congress, the accountability office report, the Sixth Circuit opinion and the printed bill were all retrieved as primary documents and read, which is unusual for a document at this evidence tier and is the reason the numbers here carry six decimal places where the source does. The trade and advocacy sources were used for timeline and characterization only. Where sources overlap they agree: the quarterly factor series published by the administrator matches the series in the monitoring report quarter for quarter with no discrepancy found, and the diagnosis that base erosion rather than spending drives the factor is stated separately by the Commission in 2022, the accountability office in 2024 and the research service in 2026. Where they disagree the disagreement is left standing rather than resolved. The annual base for 2024 is USD 32,859 million from annual filings while the four quarterly bases reported for 2025 sum to USD 32,316 million from quarterly projections, and the footnote to the first table says the amounts actually used differ from those shown without publishing them. The fund for 2025 appears as USD 8,614.46 million required in one table and USD 8.45 billion disbursed in the research service brief, and no opened source states the relationship between a requirement and a disbursement. The pre-2016 trajectory is given as about 4 percent in 1998 by one source and about 6 percent in 2000 by another, and no single authoritative pre-2016 series was opened. Two figures in this document are computed here rather than reported anywhere — the counterfactual factor of about 42.2 percent and the percentage changes derived from published rows — and each is labeled as such where it appears. This is a Path A output, so observation_refs is empty and provenance_mode: press-derived.

People affected

Estimated range 132,300,000–132,300,000 As of 2025-07

Sensitivity The real interval is on magnitude, not on the count. Total universal service contributions came to USD 5.37 per household per month in 2025 in constant 2025 dollars, of which the residential portion is estimated at USD 1.88 to USD 2.41 using the FCC staff band of 35 to 45 percent residential — USD 22.56 to USD 28.92 per household per year, computed here. That band is a staff belief and not a measurement: the note to the same table states the Commission does not currently collect data that would allow the residential amount to be calculated accurately, and the band moved from 45 to 55 percent for 2011 through 2017 to 35 to 45 percent since 2021 with no stated reason or evidence. What this count fails to capture: the 55 to 65 percent of contributions borne by businesses, which no opened source attributes to any counted population; implicit pass-through, since the FCC surcharge figure for 2024 excludes surcharges collected through higher prices rather than as a line item; and the overlap with fund beneficiaries, since a Lifeline household is a payer on any non-Lifeline line and no source quantifies the intersection. The limit in the opposite direction: this count can overstate, because a household whose voice service rides on a product outside the telecommunications category appears here as a telephone household while carrying no assessed interstate charge at all, and that group is precisely the population the contribution base has been losing. A person count is not derivable, because no source opened this round gives an average US household size.

Regional breakdown No source opened this round breaks the contribution base, the contribution factor, or the household telephone count down by state or region. The monitoring report tables used here are national totals, and the quarterly filing of the administrator reports a single national base. Splitting the national household figure by state population share would be apportionment, which this project rejects, and would in any case be wrong here because the assessed interstate portion of a bill is set per carrier by safe harbor or traffic study rather than by geography.