The Russia Test

The U.S. export-control question becomes causal in the 2014 Russia episode. This chapter reconstructs what the policy changed, which firms were compared, and what the first result establishes before wider outcomes are interpreted. The case matters because a simple graph showing that Russia-bound exports fell after 2014 would mix the policy with every other change affecting Russia; the comparison is designed to make the destination-specific effect more credible.

A defined policy aimed at an oil-sector activity

In February 2014, Russia invaded and began occupying Crimea. The United States and its Western allies responded with sanctions and trade restrictions. The episode studied here is narrower than the full history of those measures. It focuses on the Russian Industry Sector Sanctions in Federal Register notice 79 FR 45675, which became effective on August 6, 2014.

The notice added eight ECCNs and 52 Schedule B codes to the set of exports to Russia subject to licensing requirements with a presumption of denial. The policy concerned items that could be used directly or indirectly for the exploration or production of oil in Russian deep-water projects—defined in the paper as deeper than 500 feet—Arctic offshore projects, or shale projects. The covered examples included drilling equipment and parts, oil-field pumps, subsea and Arctic-capable marine equipment, seismic equipment, and related software and technical items.

“Presumption of denial” is important language. The measure was not simply a statement that every listed object had become impossible to export under every circumstance. It changed the licensing posture for a defined destination and industrial activity. That is exactly the kind of policy that can have a direct target while leaving open the empirical question of what firms do with their other products and markets.

The case should not be generalized to all U.S. export controls without qualification. The paper deliberately does not use the much larger post-2022 Russia measures as its main event. Those controls were broader and changed through multiple rounds of new requirements and prohibitions, making it difficult to isolate the effect of one policy change. The 2014 episode is useful partly because its timing and target are more contained.

The researchers identify treated firms from their pre-policy behavior. A firm enters the treated group if it exported at least one of the targeted products to Russia in 2012 or 2013. Approximately 550 firms met that condition. The pre-treatment period matters: it identifies firms that actually had exposure to the destination-specific restriction before the rule changed, rather than labeling every exporter in the industry as equally affected.

The control group is built from firms that exported the same set of targeted products to China during the same period. China was another Group D destination in the paper’s country classification. Approximately 1,100 firms made up this comparison group, after excluding firms that also belonged to the Russia treatment group. The researchers therefore compare exporters with the same product exposure but different pre-policy destination exposure.

This is not a claim that Russia and China were identical markets. A useful control does not need to be a perfect twin. It needs to provide a reasonable picture of how firms with the relevant product capabilities might have changed in the absence of the Russia-specific restriction. Holding the product set constant helps remove one obvious source of difference: the firms were not being compared simply because one sold oil equipment and the other sold unrelated goods.

Timing receives similar attention.