In new research, Matthias Breuer and Qingkai Dong examine how federal data collection can influence local spending. They examine road surveys, showing that roads included in federal samples are more likely to be funded and those that are not often face funding and safety declines, reflecting a need for improved measurement systems.
When Americans think about infrastructure, they may picture federally funded construction workers pouring concrete and asphalt, blocked off by traffic cones—visible evidence of the government at work. But the real power behind many local infrastructure decisions lies not in bulldozers or budgets, but in measurement.
Our new research shows that how the federal government measures public infrastructure ends up shaping where local governments spend. Using a long-running federal survey on highways, the Highway Performance Monitoring System (HPMS), we find that counties whose roads happen to be included in the federal sample receive more local spending, especially capital improvements. Conversely, when the federal government stops measuring certain roads, spending dries up and safety deteriorates.
The story that emerges is not only about roads, but also about how data itself governs public life. In an age of skepticism toward federal statistics, our evidence suggests that reducing information collection has real, measurable consequences for communities’ economic fortunes and safety.
The hidden power of federal measurement
Every year, the United States federal government collects thousands of data points about the nation’s economy, environment, and infrastructure. These surveys, run by agencies like the Census Bureau or the Federal Highway Administration, do far more than describe reality. They create it.
In 2021, over $2.8 trillion in federal funding was guided, at least in part, by data from federal surveys. Yet state and local governments also rely on these surveys to shape decision-making and spending. By defining what is measured and reported, these measurements define what is seen and, in turn, what is prioritized.
Nowhere is this dynamic clearer than in infrastructure. Roads are expensive, visible, and politically salient. The federal government’s HPMS, established in 1978 and implemented in 1980, was designed as a statistical system to monitor road conditions across the country. Each year, it collects basic data on roads and detailed information on a random sample of road sections.
That random sampling has an unintended but powerful side effect: some counties get measured more than others. Because the sampling was stratified by road type and traffic rather than geography, some counties ended up with more “sampled” roads simply by chance. And once the frame was fixed in 1980, those counties stayed more visible in federal data for decades.
The HPMS was never just a reporting tool. It became part of how the federal and state governments oversee local infrastructure, how states allocate funds to counties, and how counties justify their own budgets.
From random sampling to real dollars
Our study draws on two natural experiments built into HPMS’s history.
The first came with the survey’s launch in 1980, when a stratified random sample of road segments was selected. Because inclusion was unrelated to county characteristics, the share of sampled roads in each county created an almost random difference in federal attention.
The second came in 1993, when the Federal Highway Administration removed one road category, rural minor collectors (RMCs), from the survey to focus on more critical roads. That one-time decision effectively stopped detailed data collection for those roads, creating a sharp, unplanned reduction in federal measurement.
These two moments, a sudden expansion from prior uncoordinated state records and a sudden contraction of data collection, offer a rare opportunity to see how measurement alone affects local fiscal and infrastructure outcomes.
Roads that get measured, and what happens when they don’t
We find that counties with a higher share of federally sampled roads spent significantly more on highways after the HPMS began. The increase remains even after accounting for differences across states and years, suggesting it cannot be explained by changes in federal grants or state aid. The composition of spending is equally revealing. The additional funds go almost entirely to capital outlays, projects like road reconstruction and expansion, rather than routine maintenance.
The reverse is also true. In 1993, when the Federal Highway Administration dropped RMCs from the HPMS sample, the effects were immediate and negative. Counties that had previously been heavily sampled cut back on capital spending on highways, expanded their networks more slowly, saw reduced traffic volumes, and experienced higher road fatality rates. We did not find other major changes at the time—no new funding laws or regulatory shifts. What disappeared was the measurement itself, along with the attention and resources that followed it.
Taken together, these two natural experiments complete a simple but powerful picture. When Washington starts measuring, local governments invest; when Washington stops, investment fades.
Information and attention: the two channels of measurement
Why does measurement matter so much? Our evidence points to two complementary channels through which federal data collection influences local governance.
The first is the information channel. The HPMS indeed provides reliable, comparable data on road quality and traffic demand. By reducing uncertainty about which roads most need improvement, it helps local officials allocate funds more effectively. Roads identified in (almost) poor condition or carrying heavy traffic receive larger spending increases. In this way, measurement enhances allocative efficiency by revealing needs that might otherwise remain invisible.
The second is the attention channel. Beyond supplying information, measurement changes visibility and incentives. Even roads in decent condition, once included in the HPMS sample, tend to attract more resources than similar roads outside it. Being “in the data” signals importance to elected officials eager to demonstrate results.
Over time, this reordering of attention can have far-reaching consequences. Because the HPMS sampling frame remained largely fixed, the same subset of roads kept receiving scrutiny and funding year after year, while other roads, equally in need but outside the sample, stayed neglected. This dynamic has quietly reshaped where public infrastructure dollars flow, even without any deliberate decision to redirect them. These differences have accumulated into real disparities in how measured and non-measured roads handle traffic capacity.
The political economy of data
As policymakers debate scaling back federal surveys or limiting data transparency, it is worth remembering that measurement is not a bureaucratic luxury but an investment in public capacity. Our findings reveal a broader tension in how governments use information: measurement can promote efficiency yet also distort public spending.
However, the solution is not to abandon measurement because of these imperfections, but to design it more thoughtfully. When surveys disappear, decision-makers lose visibility into infrastructure, and the decline in information can translate into lower investment and poorer outcomes. In contrast, well-constructed data systems can reduce bias, keep attention focused on genuine need, and ensure that information serves as a guide rather than a source of distortion. In this sense, designing measurement is an act of designing governance itself.
Designing measurement for governance
How can the design of measurement be improved?
First, rotate samples. Periodic resampling would spread bureaucratic attention more evenly and avoid long-term favoritism toward the same regions or assets. Importantly, adopting a rotating sample does not necessarily raise survey costs significantly, and many federal surveys already use periodic resampling to maintain representativeness.
Second, make performance reports focus on need-adjusted metrics, for example, the percentage of roads in poor condition. Dashboards that highlight which roads or communities face the greatest risks would create incentives to improve these outcomes and help direct resources toward the areas of greatest need.
Beyond roads
The logic extends far beyond infrastructure. Whenever governments measure something—school performance, air quality, or corporate emissions—they shape how people and organizations behave. Measurement changes incentives by changing what gets noticed.
The same holds true in the private sector. For environmental and social issues, the costs of pollution or poor labor conditions often spill over to society rather than staying within the firm. Because these effects are diffuse and external, companies have little financial incentive to monitor or reduce them unless required to do so. That is why measurement and disclosure standards, such as those for carbon emissions or workplace diversity, matter: they bring visibility to problems whose consequences fall largely on others.
Data as infrastructure
Data is often treated as a by-product of other investments. Our findings suggest something more fundamental: data are infrastructure. Just as roads connect communities, datasets connect decisions. And like physical infrastructure, informational infrastructure requires maintenance, investment, and periodic redesign.
The HPMS was created in an era of clipboards and roadside inspections. Today, sensors, satellites, and digital mapping tools could modernize it dramatically. Yet the core principle remains: what gets measured gets managed. If policymakers want better roads and safer bridges, they must invest not only in concrete and steel but also in the knowledge systems that guide them. Cutting surveys to save money is like turning off streetlights to save electricity: the savings are visible, but the costs, from accidents to inefficiencies and lost trust, are possibly far greater.
Federal data collection is not neutral administration; it is the foundation of government capacity. Investing in data strengthens the ability to govern effectively, allocate fairly, and plan for the future.
Author Disclosure: The author reports no conflicts of interest. You can read our disclosure policy here.
Articles represent the opinions of their writers, not necessarily those of the University of Chicago, the Booth School of Business, or its faculty.
Subscribe here for ProMarket’s weekly newsletter, Special Interest, to stay up to date on ProMarket’s coverage of the political economy and other content from the Stigler Center.
