Overview

U.S. export-control claims move through three perimeters—legal, measured, and economic—and three evidence levels—descriptive, causal, and mechanistic. A legal rule can define a transaction, while ECCN or EAR99 data measure observable export value; neither alone establishes firm harm. The article shows why causal inference needs a counterfactual, using the Russia–China comparison as an example, and why intensive-margin declines can matter even without firm exit. It also separates evidence of absence from missing evidence. The framework clarifies what customs records can support and what would require buyer-level data, complete licensing records, direct compliance measures, and longer-term firm outcomes.

Learning outcomes

Legal, measured, and economic perimeters describe different layers; none can substitute for the others.
A before-and-after decline is not causal evidence without a credible counterfactual; the Russia–China design sharpens that question but still relies on common trends.
A firm can remain an exporter while reducing shipment value or payroll, so intensive-margin outcomes matter alongside exit.

Keywords

Export controlsEvidenceCausalityCounterfactualPerimeterMechanisms

Checkpoint: From Legal Target to Firm Exposure

The U.S. export-control system is easiest to misunderstand when several different questions are answered with the same word: “controlled.” This checkpoint connects the legal decision, the shipment data, and the firm-level consequences so that readers can diagnose claims about export controls without jumping from a rule to a conclusion. The useful judgment is to ask which layer a statement describes, what evidence supports that layer, and what evidence is still needed before it can be connected to domestic harm.

One policy, three perimeters

The first layer is the legal perimeter. It consists of transactions to which the EAR or another applicable restriction can apply under particular facts. Those facts include the item’s technical characteristics, its classification, the destination, the end user, and the end use. The legal perimeter is conditional. It cannot be reconstructed reliably from a product name or a physical trade code alone.

The second layer is the measured perimeter. The Census study observes export shipments with an ECCN or EAR99 designation and uses those designations to estimate how much export value falls inside the observable control categories. That perimeter is large: the share of U.S. export value subject to the EAR rose from about 33 percent in 2010 to about 60 percent in 2024. But the statistic is a measurement of export value associated with the classifications, not a count of license applications and not a measure of the time each firm spends on compliance.

The third layer is the economic perimeter. This is the part of the firm whose behavior changes when the firm responds to a rule. It may include a restricted product, but it may also include a shared factory, an engineering team, a customer relationship, a distribution network, or domestic payroll. An economic effect can exist without the related product becoming legally controlled. The firm may change its activity because the restricted product was complementary to other products, because review work became more expensive, or because customers and managers became more cautious.

These perimeters overlap, but none is a perfect substitute for the others. A shipment may be inside the EAR while not requiring an explicit license. A firm may devote resources to classifying and screening that shipment even though no license is filed. A policy may then alter the firm’s wider decisions even though those other products remain outside the direct target. The movement from one layer to the next is the central analytical problem.

The distinction also protects against a misleading interpretation of the EAR99 trend. EAR99 shipments drove much of the rise in the measured perimeter, but EAR99 is not the same as prohibited, and it is not the same as entirely unregulated. It is a designation for an item subject to the EAR but not listed on the CCL. The outcome for a particular transaction still depends on destination, end user, end use, and other applicable conditions. A broad category can contain transactions with very different levels of regulatory intensity.

The geographic data require the same care. Most of the value of EAR-subject exports went to Group B destinations in 2024, while Group D accounted for a smaller share and Group E for a negligible share. That does not imply that Group B shipments are automatically simple or that Group D shipments all create the same burden.

Published by DeepMox

FAQ

What is the difference between a descriptive, causal, and mechanistic claim about export controls?

A descriptive claim reports a pattern, a causal claim attributes a change to a policy using a counterfactual, and a mechanistic claim explains how the effect could travel; each requires different evidence.

Why is a counterfactual needed to evaluate an export-control policy?

Because a post-policy decline can also reflect demand, exchange rates, political events, product changes, or firm decisions; a credible comparison helps attribute the difference to the policy.

Can a firm experience an export-control effect without exiting the export market?

Yes. Intensive-margin measures capture reductions in export value, production scale, or payroll among firms with positive activity, which exit measures can miss.

Previous9 minU.S. Export Controls: Measuring the Perimeter

Next9 minU.S. Export Controls: Who Bears the Exposure

U.S. Export Controls: From Rule to Evidence

Checkpoint: From Legal Target to Firm Exposure

One policy, three perimeters

FAQ