Rural broadband went wrong at handover, not at appropriation, and the agency that signed the award should carry the account

Two explanations dominate the rural broadband debate. One is that there was too little money. The other is that the money moved too slowly. My position is that the money bought the wrong thing. It bought construction, and a connection is not construction. It is a line that works, at a price a household will pay, kept working after the crews have gone. A programme that pays against construction milestones buys construction, however fast the money moves.

That distinction is not pedantic, because the two things have different owners. A build has a completion date and a contractor who is finished on it. A service has no completion date and needs somebody still present in year eleven, splicing a cut, replacing a failed unit, answering a household that has stopped paying. Appropriations were written for the first. Accountability was written for the first as well, and that is the defect.

The interval was not free

Between the appropriations and the work there was an interval of years, and the interval had a price. The producer price index for all commodities averaged 227.3 in 2021 and 260.3 in 2025, a rise of 33.0 points, or 14.5 per cent. Measured from 2020, when the index stood at 194.4, the rise is 65.9 points, or 33.9 per cent. The index is a level on a 1982 base of 100 and is not a construction cost index; it isolates neither conduit nor fibre nor a bucket truck. What it establishes is that a sum fixed in nominal dollars in 2021 commanded less work by 2025. The path was not one way. The index peaked at 264.5 in 2022, fell in each of the two years after, and at 260.3 has still not regained that peak.

Those dollars were also bidding against everything else being built. Nonresidential construction spending ran at an annual rate of 845 billion dollars in 2021 and 1,309 billion in 2025, a rise of 464 billion, or 54.9 per cent.The series is stated at a seasonally adjusted annual rate, so each point is a yearly rate rather than one month's outlay. Across the same four years the ten-year Treasury constant maturity rate moved from an average of 1.45 per cent to 4.29, a rise of 2.84 points in the price of time. No subgrantee borrows at that rate, and what one actually paid to raise its match is not recorded in this brief. The direction of the move is.

None of that was hidden. Input prices, a construction boom and the cost of money are published daily or weekly. An appropriation denominated in nominal dollars, awarded years before the work, places the risk of all three on whoever signs the last contract. In practice that party is a small rural provider or a co-operative, which is the party least able to carry the risk and the one with the least standing to refuse it.

Who actually makes a connection

The work is done by a named and countable group of people. In the occupational wage estimates for 2025, West Virginia reports 1,270 telecommunications line installers and repairers out of 701,610 jobs in the state, which the file states as 1.811 for every thousand. Their annual median wage is 89,690 dollars against a state median of 45,300, a gap of 44,390 dollars, or 98.0 per cent. They are 3.6 per cent of the state's 35,400 installation, maintenance and repair workers, whose own median is 49,740.

That row carries a location quotient of 2.88, meaning West Virginia's share of jobs in the occupation is nearly three times the national share. West Virginia is also the only jurisdiction in which the occupation appears in this file at all, because the file keeps the twenty-two major groups and the twelve occupations each state is most concentrated in and drops the rest. The absence elsewhere is a property of the file, not of those states. Nor does the file record which industries employ these workers, or that any of them has touched a subsidised build.

The narrow point survives every one of those caveats. A wage at almost double the state median is what scarcity looks like when it is priced, and a state short of these people is not made less short by an appropriation. More to the argument, the same 1,270 are the maintenance crew. Build a network across that state and it is these hands that keep it alive, for as long as it is alive. An award that pays for their time once, during construction, has bought the cheap half of the asset.

The delay argument does not reach it

The standard account blames delay: environmental review, pole attachments, permission to cross a railway, state offices without the staff to run a subgrant round. Every item on that list is real and each has added months. The delay is not in dispute here. What is in dispute is what curing it would have produced.

Speed acts on a design. Remove every avoidable month from a programme that pays against construction milestones and the result is construction, delivered earlier, with the same silence about year eleven. Faster is better than slower and it is not the same as right. The delay account is also convenient, because it locates the failure in procedure, where no single official signed anything, rather than in the terms of the award, where somebody did.

Where the account should sit

Accountability belongs with the state agency that signed the award, and with a named officer inside it. Not with the federal appropriator, which cannot see an individual route. Not with the subgrantee alone, which did not write the terms it was offered. The signing agency chose the applicant, set the milestones and holds the money back. Spreading responsibility across three levels of government has meant that no level has had to answer for the distance between a dollar obligated and a household online.

Settle that account against a working line with a subscriber on it, at a posted price, at a stated date after completion, and retain a share of the award until the test is met. Price the duty to keep serving for a fixed term into the award itself, so that a bidder intending to walk away in year eight bids higher than one who does not, and the difference is visible when the decision is made rather than discovered afterwards. Name the successor in the award document: who operates the plant if the awardee fails, and on what terms it transfers.

Every one of those terms requires a number that does not exist. This brief holds no series recording awards, obligations, locations reached, lines installed, subscribers or prices paid, and the files it works from contain nothing from which such a figure could be built. No chart accompanies this piece, because the chart the argument needs has not been published by anyone. Connections delivered against dollars obligated, reported by award and by state on a fixed schedule, would let a reader test this argument and let a legislature test the programme. The absence is not a gap in the data. It is the accountability failure itself, in the place where it is easiest to see.

My position, then, is that the rural broadband programmes were written as construction purchases when they should have been written as purchases of service, and that the account for the difference belongs to the agency that signed each award. Pay for the line when it carries traffic and somebody is paying for it. Publish the count. Until both hold, the programmes will go on delivering the thing they were drafted to buy, which is plant, and the household at the end of the road will go on being described as reached.