When Leading Indicators Don't Agree

What the ABI and DMI Are Actually Telling You

The nonresidential construction market has two widely cited leading indicators. Both are released monthly. Both claim to forecast where building activity is headed roughly nine to twelve months out. In March, they appeared to diverge. One recovering toward neutral, the other pulling back from an elevated peak. Most readers see the headlines and move on. Few ask what these figures measure, or if the signals are as reliable as advertised.

The Architecture Billings Index (ABI), published monthly by the American Institute of Architects and Deltek, is a diffusion index. It is based on a survey of architecture firms asking whether their billings increased, decreased, or held steady compared to the prior month. The result is a score centered on 50. Above 50 means more firms reported growth than decline. Below 50 means the opposite. Since the ABI measures direction, not dollars, inflation does not distort it. Its limitations are different. It is a sentiment measure and subject to who responds, how large those firms are, and what sectors they serve.

The Dodge Momentum Index (DMI), published by Dodge Construction Network, works differently. It tracks the dollar value of nonresidential projects entering the early planning stages each month, drawing on Dodge's proprietary project database. That makes it transaction-based rather than survey-based, which sounds more concrete however, the DMI carries three methodological limitations rarely mentioned, and all of which are highly relevant today.

First, it is nominal. The DMI is reported in current dollars without adjustment for construction cost inflation. When costs rise, the index rises with them, even if the real volume of planned projects is flat or falling. Second, it is an intake measure with no attrition mechanism. When a project enters the planning database, it is counted. When a project stalls, is deferred, or is canceled entirely, it is not removed. In an environment where owner uncertainty is elevated and projects are dying on the vine at an above-average rate, the DMI has no way to reflect that erosion. The headline stays elevated regardless. Third, its sector composition is heavily distorted right now in ways the aggregate number conceals.

That distinction matters more now than it did five years ago.

What did each index show in the most recent data?

March ABI came in at 49.8, the closest the index has been to the 50-point neutral threshold since October 2024. Firm backlogs averaged 6.6 months, the highest since December 2023. New project inquiries rose. On the surface, it reads as a recovery.

March DMI came in at 250.5, up just 1.8% from February's downwardly revised reading of 246.2. That modest gain followed two consecutive months of decline from a December 2025 peak of 296.8. The DMI finished 2025 up 37% year over year, a figure that generated considerable attention in the industry press.

If the ABI is recovering, why aren't design contracts reflecting that?

This is the detail buried beneath the headline. Design contracts declined for the 24th consecutive month in March, and the pace of decline worsened from February. Billings stabilizing means architects keep existing projects moving. It does not mean owners are signing new work. The pipeline is being worked down, not replenished at an equivalent rate. For owners watching the ABI tick up toward 50, the recovery is real but narrow.

The DMI finished 2025 up 37%. Doesn't that signal a strong forward pipeline?

It signals something, but the three limitations described above are doing real work on that number. The nominal distortion alone is significant. A project valued at $50 million in 2022 might carry a $65 million price tag today for the same scope. When it enters the planning database, the DMI captures $65 million. The index goes up. Real planning volume did not.

Layer the attrition problem on top of that. In the current environment, with tariff uncertainty, elevated financing costs, and unresolved macroeconomic risk, more projects are entering the planning pipeline and quietly stalling than in a normal cycle. ConstructConnect's Project Stress Index shows abandoned projects down -1.5% y-y, but up 22.8% m-m. The DMI counts the intake. It does not count the exits. The 37% annual gain in 2025 is not fabricated, but a meaningful portion of it reflects cost inflation rather than genuine volume growth, and an unknown portion of the remaining pipeline may never convert to starts.

What is driving the DMI right now, and why does that matter for healthcare and institutional projects specifically?

The March DMI headline was propped up almost entirely by data center projects. Commercial planning rose 7% on the month, powered by that single sector. Institutional planning, which includes healthcare, education, and public buildings, fell 8.8% in March, following declines in January and February as well.

That sector-level breakdown is the most important figure in the report for owners in the healthcare and life sciences space. The aggregate DMI looks stable. The institutional planning pipeline is contracting. Those are not the same market, and they should not be read as such.

So what are these two indices saying together, and what should we do with that?

Read together, the ABI and DMI are describing a market in transition rather than expansion. Architects have capacity and are billing steadily but are not signing the new contracts that would sustain that activity twelve months out. The planning pipeline is narrowing in the institutional sector even as data centers keep the aggregate number elevated. An unknown share of what remains in the DMI pipeline may not survive the current uncertainty intact.

For healthcare and life sciences owners sitting on deferred projects, that combination points toward a specific window. Contractor and architect capacity in the institutional sector is available now in a way it has not been in several years. The leading indicators are not forecasting a boom. They are forecasting tightening, and the tightening has not arrived yet.

The question project teams should be asking is not whether the market is recovering. It is whether they want to be in the ground before it does.