The Persistent Geography of Subnational Trade

The economy is often imagined as a fluid thing, a river of goods and services that constantly reshapes its course in response to new roads, cheaper shipping, or sudden crises. For decades, economists have assumed that the map of who buys from whom is highly flexible, shifting quickly whenever costs change or a new competitor appears. This idea underpins much of how we understand regional growth and how governments plan for the future. If trade routes are fluid, then a shock to one area should quickly ripple out, finding new paths and new partners. But what if the map is not a river at all? What if, beneath the surface of daily transactions, the connections between places are actually as solid and unchanging as bedrock?

A new study by Michael Lahr of Rutgers University challenges the long-held belief that trade patterns are fast-moving. By assembling a massive, consistent record of shipments between every U.S. state over thirty years, from 1993 to 2022, Lahr discovered that the geography of domestic trade barely moves. Despite the arrival of the internet, the collapse of the housing market, the Great Recession, and the global pandemic, the specific paths that goods take from one state to another have remained remarkably stable. The study suggests that the relationships between buyers and sellers are not easily broken; they are slow-moving features of the economy that resist change even when the world around them is in turmoil.

To reach this conclusion, the researcher had to overcome significant hurdles in the data itself. The United States conducts a major survey called the Commodity Flow Survey roughly every five years, which tracks the value and weight of goods moving between states. However, the data from these surveys is not always easy to compare over time. Some years, the government hides specific numbers to protect the privacy of companies, and in earlier decades, the data was stored in formats that are now difficult to access. Lahr spent considerable effort recovering the lost records from 1993 and 2007, treating the hidden numbers not as zeros but as missing information that still existed. He then stitched together a single, continuous timeline of trade flows covering seven different survey periods, creating the first complete picture of how the entire country has traded with itself over three decades.

The results of this reconstruction were striking. When the researcher compared the trade patterns of one five-year period to the next, the similarity was overwhelming. The ranking of which states traded most with each other stayed almost exactly the same, with a correlation score between 0.95 and 0.97 for every single transition. To put this in perspective, if you were to shuffle a deck of cards and then deal them again, the order would be completely different. But here, the order of trade partners remained nearly identical. Over the full thirty-year span, the correlation was still 0.93, a number so high it suggests the structure of trade is effectively frozen. Even more telling, the study found that at least 98.5 percent of the value of goods moving between states in 2022 was traveling along routes that were already active in 1993. The vast majority of the money spent on interstate goods in 2022 was flowing through the same channels that had been established three decades prior.

This stability held true even during the most turbulent economic times. The Great Recession, which caused massive job losses and factory closures, and the COVID-19 pandemic, which disrupted supply chains globally, left no visible mark on the map of trade relationships. The share of goods moving from one state to another did not rearrange itself in response to these shocks. The distance that goods travel also remained stubbornly consistent. The study measured how much trade drops off as the distance between states increases, a concept known as distance elasticity. In 1993, this drop-off was steep, and in 2022, it was almost exactly the same. Despite decades of improvements in trucks, logistics, and communication technology that should have made long-distance trade easier, the resistance to moving goods across long distances did not weaken.

The research also looked at why these patterns might be so fixed. It turns out that the type of goods being traded matters. Simple products that are bought and sold based on a posted price, like raw materials, show slightly more flexibility. However, complex goods that require a deep, specific relationship between a buyer and a supplier—such as specialized machinery or custom parts—show almost no movement at all. These relationships are built on trust, specific testing, and long-term contracts. Switching suppliers for these items is costly and difficult, so companies stick with their existing partners even when prices fluctuate or new competitors emerge. This suggests that the "friction" of trade is not just about the cost of shipping, but about the deep, sunk costs of maintaining a business relationship.

For the people who build economic models to predict how regions will grow or how a shock will spread, these findings are a major correction. Current models often assume that trade routes can be redrawn quickly to find the most efficient path. This study suggests that assumption is wrong. The map of trade is not a fluid network that reconfigures itself overnight; it is a slow-moving structure that carries its own history. When a region faces a shock, the impact will likely stay concentrated on the partners it has always traded with, rather than spreading out to new, distant partners. The economy adjusts not by finding new trade routes, but by changing prices, moving workers, or building new factories within the existing network. The study does not claim that trade never changes, but it proves that when it does, it changes at a glacial pace, carrying the geography of the past into the future.