The Broad Pattern: States with Higher Wage Growth Saw More Robot Exposure

This chapter interprets the Census study's first empirical view of the minimum-wage and robot-adoption relationship: a state-level comparison between changes in wage floors and changes in exposure to industrial robots. The point is to understand what the reported positive pattern actually measures, including its roughly 8-percent relative-to-mean calculation, and why a broad map can motivate a firm-level test without becoming a causal verdict.

Turning national robot growth into state exposure

The study does not count every robot installed in every state and then compare that count with each state's minimum wage. Instead, it constructs an exposure measure. The measure starts with national changes in robot deployment by industry. It then asks how much each state was already invested in those industries at the beginning of the period. A state with a large automotive or machinery sector receives more weight from national robot growth in those industries than a state with little presence in them.

That weighting is important because national technology change does not arrive evenly across the country. Industrial robots are concentrated in production settings where the tasks and supply chains support them. If the national automotive industry adopts more robots, states with a large initial automotive employment share will have greater predicted exposure even before researchers look at later state outcomes. The measure is therefore a way of translating national industry trends into a state-specific trajectory using the state's starting economic structure.

The authors make additional adjustments. They account for growth in industry value added, which helps distinguish a change in robot exposure from simple expansion in an industry. They also account for pre-existing robot stock and the baseline level of adoption. The resulting measure is intended to capture excess robot adoption associated with the interaction of national technology change and a state's initial industrial mix, rather than merely reproducing the fact that some states had more manufacturing to begin with.

The time windows are not identical. The robot-exposure measure covers a long difference from 2004 to 2016, while the minimum-wage comparison covers state changes from 2003 to 2015. The paper lags the minimum-wage measure by one year relative to robot exposure. That timing is consistent with a firm needing time to respond to a changed labor cost: a policy change may influence planning, procurement, installation, and adoption in subsequent periods rather than on the same day.

The wage measure also includes a practical institutional detail. States with no state minimum wage use the federal minimum wage, and some states have statutory floors below the federal rate. The comparison therefore reflects the wage floor that firms actually face under the study's construction, not a simple list of state legislative announcements.

Consider two hypothetical states to see why the exposure measure is not a direct machine count. State A begins the period with a large automotive and machinery base; State B has a much smaller presence in those industries.