The Asphalt Ledger – Part 2: The Induced Demand Machine — Why Every New Lane Builds the Case for the Next One

The Prediction That Was Always Wrong and Always Made Again

#

In 1999, the Texas Department of Transportation completed a widening of Katy Freeway — US Highway 290's western approach into Houston — adding managed lanes to a corridor that already carried approximately 198,000 vehicles per day. The project was justified on congestion relief grounds: an average 15–20% reduction in peak-hour travel times was projected, based on traffic models using an induced demand elasticity of approximately 0.1 — meaning a 10% capacity increase would generate approximately 1% more vehicle trips. The projection was, by this standard, optimistic but not unreasonable.

Traffic counts in 2001 recorded 218,000 vehicles per day. The travel time savings had dissipated. A second, larger widening was authorised in 2008, extending Katy Freeway to 26 lanes — including access and collector roads, the widest urban highway in the world. At a construction cost of approximately $2.8 billion, it was the largest highway investment in Texas DOT history to that date. 2011 traffic studies found that peak-hour travel times on the expanded freeway had increased over pre-expansion baselines in several segments. The additional capacity had generated additional demand — measured, this time, not over two years but within two years. The 2019 traffic count showed 295,000 vehicles per day on a corridor designed for a fraction of that volume. Katy Freeway is now the primary source of air quality violations in the Houston metro area.

The road was not widened by engineers who did not know about induced demand. The research on induced demand was available. It was excluded from the cost-benefit model that authorised the project — because the model was calibrated to produce an answer that justified construction.

The Elasticity That Infrastructure Models Refuse to Use

#

Induced demand is the phenomenon by which road capacity expansion generates additional vehicle travel — trips that would not have been made, or would have been made by different routes or modes, at pre-expansion congestion levels. Its existence is not contested in transportation research. Its magnitude is.

The Road Subsidy Multiplier's induced demand reconstruction term depends on the accuracy of the elasticity applied:

The induced demand reconstruction cost is calculated as: (additional VMT attributable to induced demand) × (pavement degradation cost per additional VMT) × (acceleration of maintenance schedule). This requires an elasticity estimate: the percentage change in VMT per percentage change in lane-miles.

The Research, the Model, and the Gap Between Them

#

What Three Decades of Transportation Research Show

#

The academic literature on induced demand converges on a long-run elasticity of vehicle-miles travelled with respect to lane-miles of approximately 0.75–1.0 for urban corridors. The landmark study is Duranton and Turner's 2011 paper in the American Economic Review, which used U.S. highway data from 1983 to 2003, combining three decades of lane-mile additions with county-level VMT measurements and controlling for population, income, and employment growth. Their central estimate for the long-run elasticity was 1.0: a 10% increase in lane-miles produced a 10% increase in VMT in the long run, consistent with what Duranton and Turner called "The Fundamental Law of Road Congestion." The finding implies that highway expansion is, in the long run, self-defeating as a congestion reduction strategy — an implication the study's authors stated explicitly.

Supporting evidence comes from Noland (2001), who estimated an elasticity of 0.6–0.7 for U.S. highway data 1984–1996; from Cervero (2003), who found elasticities of 0.3–0.6 for California state highways; and from Melo, Graham, and Noland (2016), whose meta-analysis of 55 studies found a mean short-run elasticity of 0.28 and long-run elasticity of 0.56–0.74. Across methodologies, datasets, and time periods, the weight of evidence places the induced demand elasticity for urban corridors in the range of 0.5–1.0, with the higher values applying to dense urban contexts where latent demand is greatest.

The Model That the DOT Uses

#

Official DOT traffic demand models — used to justify highway expansion projects in cost-benefit analyses that accompany environmental impact statements — apply induced demand elasticities of 0.0–0.2 in most states. The Federal Highway Administration's travel demand modelling guidance suggests accounting for "some" induced demand but provides no mandated elasticity value, leaving the choice to state DOT modellers. The Texas Transportation Institute's 2023 Urban Mobility Report — the standard reference for congestion cost analysis used in US highway project justifications — applies elasticities in the lower range of the academic distribution.

The resulting projections consistently overestimate congestion relief benefits and underestimate traffic growth. The TTI has documented, in its own mobility trend analyses, that projects projected to achieve 20–30% congestion reduction typically achieve 8–12% reductions within three years and revert to baseline or worse within seven. The institution whose data is used to justify expansion investment is the same institution that documents those expansions' consistent failure — in separate publications, using separate methodological frameworks, read by separate audiences.

This is not a research failure. It is an institutional design feature: the cost-benefit model that authorises construction and the performance monitoring that tracks outcomes exist in different bureaux, report to different budget authorities, and are subject to different political pressures. The cost-benefit model is a prerequisite for construction funding; the performance monitoring is optional disclosure. The perverse incentive structure ensures that constructions are authorised on optimistic projections that understate induced demand, and that the resulting underperformance is documented separately without triggering revision to the authorisation model.

The RSM Table for Documented Urban Expansions

#

Applying the RSM formula to documented urban highway expansion cases — using the Duranton-Turner elasticity of 0.75 as a conservative mid-point, rather than the 0.0–0.2 applied in official models — produces a range of Road Subsidy Multipliers that the DOT analyses authorising these projects did not compute.

I-405 Los Angeles widening (2014, $1.6B construction): Traffic volume on the corridor exceeded pre-widening levels within 18 months. Applying a 0.75 induced demand elasticity to the 10% capacity increase (lane-miles added), projected additional VMT over 20 years generates approximately $820 million in pavement degradation acceleration and early rehabilitation cost. 20-year lifecycle maintenance at standard Interstate rates adds approximately $1.4 billion. RSM = ($1.4B + $0.82B) ÷ $1.6B = 1.39. Every dollar of LA-405 construction generated $1.39 in 20-year public lifecycle obligation.

I-270 Missouri widening (2023, $695M construction, first phase): A lower-intensity corridor with less latent urban demand. Induced demand elasticity applied at 0.6 produces a lower additional VMT projection. 20-year RSM: approximately 2.4. The lower traffic density does not moderate RSM significantly because the rural/suburban pavement design is calibrated for lower loads, making induced traffic above design volume disproportionately damaging to pavement life.

I-35 Austin expansion (multiple phases 2018–2024, cumulative $3.2B): Austin's rapid population growth provides a context where both standard growth VMT and induced demand VMT compound simultaneously. RSM at 20 years: approximately 3.8. This is perhaps the most honest RSM case study available: a corridor where transportation planners knew the population growth trajectory would generate the demand regardless of induced elasticity, using expansion to meet growth demand while the induced demand term added materially to the maintenance obligation.

The Model That Would Change the Funding Decision

#

If official DOT cost-benefit models were required to apply a minimum induced demand elasticity of 0.5 — the conservative lower bound of the peer-reviewed literature — and to include the resulting RSM-adjusted lifecycle obligation in the fiscal impact analysis, the majority of urban highway expansion projects currently in planning pipelines would produce negative present-value assessments when the full public obligation is disclosed.

The Infrastructure Investment and Jobs Act of 2021, which committed approximately $550 billion over five years to US transportation infrastructure, did not require any project to disclose RSM. Environmental impact statements require traffic projections; they do not require those projections to use elasticities consistent with the academic literature on induced demand. The review process that scrutinises the environmental effects of a highway project does not scrutinise the fiscal model that projects its benefits.

The result is a systematic deployment of public capital into projects whose RSM exceeds 2.0 — whose total public lifecycle cost is more than double the construction price — while the authorising analysis records only the construction cost and a traffic projection calibrated to justify it. The next post examines why this calibration persists, who benefits from it, and what the RSM comparison between road and transit investment reveals about which infrastructure type the analysis framework is designed to favour.