Infrastructure Costs Are a Procurement Story

Americans have long intuited that we pay too much for infrastructure. Only recently, however, have researchers documented just how high those costs are: American urban rail projects cost far more than comparable projects in other rich countries, and the cost of building American interstates has risen several times over since the 1960s.

There are many possible explanations, and several are probably true at once. They divide roughly into three categories. The first concerns what we pay for inputs: labor costs, debated for decades, and more recently materials prices driven up by tariffs. The second concerns what we get from those inputs: Goolsbee and Syverson find that by 2020 labor productivity in construction had fallen below its 1950 level. The third concerns who gets a say. Brooks and Liscow show that the historical rise in highway costs tracked the expansion of “citizen voice,” which gave citizens more room to shape, delay, or block projects.

All these explanations focus on the process of building infrastructure. But the price a government pays reflects both what the project costs the construction firm and the markup the firm charges above that cost. Those markups are a function of market structure — how many firms there are, and how they compete. In the US, most infrastructure is acquired through public procurement, where the government makes purchases from private firms competing for contracts. Infrastructure costs are thus part of a larger question about procurement.

And procurement is large. Across all levels of government, it is about one-tenth of U.S. economic activity, roughly $3 trillion a year. Most of that spending occurs at the state and local government level, which accounted for 67 percent of the total in 2023. Yet these are also the levels we can observe the least — national data on those markets is close to nonexistent. That leaves two key questions: how much competition is there, and is it an important factor in how much the government pays?

In a working paper, I study these questions in the market for highway construction. Highways, like most infrastructure, are procured at the state level, so the market is fragmented across state governments, and the underlying data is difficult to compile. To build a highway, a state government puts each project up for auction, and the lowest bid wins. I assemble, for the first time, highway auctions from all contiguous U.S. states over roughly two decades, which gives me precise data on who bids and how much. Using this data, I find that competition is limited: auctions with just one or two bidders account for about a third of projects, and the share of auctions receiving three or fewer bids rose from 47 percent in 2010 to 63 percent in 2024.

How much does this matter for prices? The obvious approach is to check whether projects with more bidders are cheaper. But that comparison is misleading: bigger jobs attract more bidders, so the raw correlation between the number of bidders and prices need not reflect the effect of competition. I take two alternative approaches. The first uses a bit of randomness in where competition shows up. Every so often an out-of-state firm starts bidding in a new state, and projects near its entry point suddenly face stiffer competition. Comparing these projects with otherwise similar ones helps isolate the effect of competition. The second approach leans on theory. Given the structure of these auctions, a firm’s optimal bid and markup above cost can be worked out. Assuming firms have roughly figured out the best strategies by experience, the model can be estimated to quantify markups and simulate prices under greater competition. The two approaches rest on different assumptions but produce similar estimates: one additional bidder lowers the government’s cost of procuring infrastructure by about 10 percent.

If auction markups are high, why don’t more firms enter the market and compete them away? My analysis suggests the answer is that entry is costly. The profits from facing only one rival may not cover the bureaucratic cost of learning to work with the procurement agency. The geographic pattern of bidding bears this out. Projects attract more bidders when other projects sit nearby, so that firms can spread a fixed entry cost across several opportunities. The effect stops at state lines. An in-state cluster of work twenty miles away draws new firms in; the same cluster across a border draws none. Entering a new state, it seems, carries fixed costs high enough to deter firms that would otherwise compete.

Firms say much the same thing. Requirements and prequalification differ state by state, and the average highway bid runs 164 pages of forms. Contractors describe the burden as extensive, especially when state and federal rules diverge. Both simplifying and aligning those processes could make entry cheaper and increase competition. However, simpler rules are not the same as less government. In fact, Liscow et al. find that when experienced engineers leave state transportation agencies, project costs rise by about six times those engineers’ wages. Investing in capable staff who write and evaluate these contracts is complementary to easier entry, not in tension with it.

At the same time, procurement rules exist for a reason. The Lincoln Memorial Reflecting Pool is a recent case in point, drawing news coverage over no-bid contracts awarded under an urgency exception to firms with little federal work experience and possible political ties. Here the rules were bypassed, and the result was clearly non-competitive. Yet the ordinary process fares little better. The National Park Service, which manages the Reflecting Pool, struggles to attract bidders at all. Reviewing 28 construction contracts funded by the Great American Outdoors Act, the Interior Inspector General found that 13 drew only a single proposal.

Taken together, the relationship between regulation and competition may look like an inverse U. Too few safeguards, and procurement invites favoritism and low-quality work. But too many rules — or rules that differ needlessly across states and agencies — can make bidding so costly that few firms participate. We occasionally reach the first extreme by overriding the system altogether. More often, we may be closer to the second: enough rules to deter potential bidders, leaving governments with little competition and higher prices.