Technical Debt

Werner Cunninghamee

Infrastructure Finance Quarterly, Vol. 12, No. 3, Autumn 1992

Every bridge in America is borrowing against its own future, and almost none of them are keeping books.

I want to propose a framework for thinking about deferred maintenance in public infrastructure that I believe clarifies a problem the field has long discussed without adequately naming. I call the framework technical debt, by analogy to financial debt, and the analogy is more precise than it may first appear.

When a municipality defers a scheduled maintenance action — a bridge inspection, a pipe relining, a roof replacement — it does not eliminate the cost. It borrows against the future condition of the asset. The work that should have been done in 1990 must still be done; it must now be done in 1993, or 1995, or whenever the deferral ends. But the cost in 1995 will not be the same as the cost in 1990, because deterioration is not linear. A hairline crack in a bridge deck that costs eight thousand dollars to seal in the year it appears will cost fifty thousand dollars to repair three years later, when water infiltration has corroded the rebar and the crack has become a spall. The difference — forty-two thousand dollars — is interest on the technical debt.

This is the essential property that makes the debt metaphor apt rather than merely decorative. Financial debt accrues interest. So does deferred maintenance. The interest rate varies by asset type, climate, usage, and material — a steel bridge in a salt-spray environment accrues technical debt faster than a concrete culvert in the desert — but it is always positive. Deferred maintenance never gets cheaper. The question is only how much more expensive it gets, and how quickly.

The Taxonomy of Technical Debt

Not all technical debt is created equal, and the distinction matters for both engineering and finance.

Deliberate technical debt is deferred maintenance undertaken consciously, for a stated reason, with a plan for eventual repayment. A municipality that defers resurfacing a road scheduled for widening next year has taken on deliberate technical debt; the deferral is rational because the surface will be torn up anyway. A school district that patches a boiler rather than replacing it, because the replacement is budgeted for the following fiscal year, has made a similar calculation. The debt is real — the patch will cost more per month of operation than the replacement would have — but it is conscious, bounded, and tracked. This is, in financial terms, a well-managed line of credit.

Inadvertent technical debt is the dangerous kind. It accumulates not through decision but through neglect — the inspection that was skipped because the inspector was reassigned, the repair that was deferred because the paperwork was lost, the replacement that was postponed because the budget committee assumed it could wait another year without examining whether this assumption was true. Inadvertent technical debt is not borrowed; it is accrued, silently, in the gap between what an asset needs and what an institution provides.

The distinction matters because deliberate debt can be managed while inadvertent debt can only be discovered, usually at the worst possible moment. The Williamsburg Bridge crisis of 1988 is instructive. The bridge had been in service since 1903, and decades of deferred cable inspections had produced a technical debt so large that when the cable deterioration was finally assessed, the estimated repair cost — one hundred and fifty million dollars — exceeded the original construction cost adjusted for inflation. The bridge was not neglected out of malice. It was neglected out of the ordinary, incremental, institutional process by which next year’s budget is assumed to resemble this year’s, and this year’s budget is assumed to be adequate, and adequacy is assumed until the cables start to fray.

The Interest Rate

The interest rate on technical debt is determined by the physics and chemistry of deterioration, which are indifferent to budgets.

Consider the Riverside Elementary School in Dayton, Ohio — a composite example, though every detail is drawn from actual case files. The school was built in 1962 with a flat roof and a twenty-year membrane. The membrane was replaced on schedule in 1982. The next replacement was due in 2002. In 1998, the district deferred the replacement to fund an expansion at another school. By 2001, the membrane had begun to fail at the seams. Water infiltrated the roof deck. The deck, a lightweight concrete pour over steel joists, began to corrode at the joist seats. By 2003, the ceiling tiles in three classrooms had to be replaced due to water damage. By 2005, the structural engineer’s report identified joist deterioration requiring the replacement not merely of the membrane but of the entire roof assembly — deck, joists, and insulation — at a cost of $1.4 million, against the $180,000 that the membrane replacement would have cost in 2002.

The technical debt on the Riverside roof accrued at an effective annual interest rate of approximately 34%. This is not unusual. It is, if anything, conservative for water-infiltration scenarios, where the transition from surface deterioration to structural compromise can be abrupt and nonlinear. Corrosion does not proceed at a constant rate. It accelerates as protective coatings fail, as moisture pathways widen, as galvanic couples form between dissimilar metals. The interest compounds.

Bankruptcy

There is a point — and every infrastructure professional knows it, though few enjoy naming it — at which the accumulated technical debt on an asset exceeds its replacement value. At this point, the asset is technically bankrupt. Further maintenance spending is, in financial terms, throwing good money after bad. The rational decision is demolition and new construction, which is politically the most difficult decision a municipality can make, because a building that is still standing looks, to the layperson and the elected official, like a building that is still functional. The structural engineer’s report says otherwise. The structural engineer’s report is not on the ballot.

The Cuyahoga County parking garage that partially collapsed in 2014 had reached bankruptcy approximately six years before it collapsed. The maintenance records — such as they were — showed a consistent pattern of patch-over-patch repair, each patch cheaper than the systemic work it deferred, each patch adding another layer of complexity to the eventual reckoning. The total spent on patches between 2008 and 2014 was approximately $340,000. The replacement cost in 2008 would have been $2.1 million. The replacement cost after the collapse — including demolition, debris removal, liability, and new construction — was $7.8 million. The patches did not save money. They borrowed it, at rates that would embarrass a loan shark.

On Tracking What Is Owed

The infrastructure finance profession is sophisticated in its treatment of bonded indebtedness, pension obligations, and capital reserves. It is remarkably unsophisticated in its treatment of technical debt. Most municipalities do not maintain a ledger of deferred maintenance. They do not calculate the accruing interest. They do not mark the point of bankruptcy. They operate, in effect, like a household that tracks its mortgage payment but not the condition of its roof — and then acts surprised when the ceiling caves in.

I propose that every infrastructure asset under public management should carry a technical debt register, maintained by the responsible engineering authority and reported alongside the financial statements. The register would record each deferred maintenance action, the estimated cost at the time of deferral, the estimated cost at the current date (i.e., principal plus accrued interest), and the projected cost trajectory if the deferral continues. This register would not, by itself, compel maintenance. But it would make the cost of not maintaining visible, which is the prerequisite for any rational discussion of priorities.

The objection will be that this is expensive. I note that the Riverside roof cost $1.4 million because a $180,000 job was deferred. The register would have cost a few hours of an engineer’s time per year. The arithmetic is not complicated.

A Note on Metaphorical Borrowing

I am told that the “technical debt” framework has found admirers in the computing industry, where it is used as a metaphor for some kind of accumulated compromise in the construction of computer programmes.¹ I am flattered by the adoption, though I confess I do not entirely understand the analogy. Software, as I understand it, does not corrode. It does not suffer water infiltration. It is not subject to freeze-thaw cycling or galvanic reaction. The mechanisms by which deferred maintenance on a bridge becomes more expensive over time are physical and chemical; they are governed by the second law of thermodynamics, which applies to steel and concrete with a rigour I am not sure pertains to whatever it is that programmes are made of.

Still, if the framework is useful to them, they are welcome to it. I ask only that the original context not be forgotten. The bridges are real. The debt is real. The interest does not care whether anyone is paying attention.

Werner Cunninghamee is a senior analyst at the Pacific Northwest Infrastructure Group and a fellow of the American Society of Civil Engineers (Infrastructure Finance Division). He serves on the advisory board of the Oregon Department of Transportation’s Asset Management Programme.