Mapping the Cocaine Supply Chain

The global supply of cocaine is at record levels after hitting a slump during the COVID-19 pandemic. Seizures of the drug and related arrests are also on the rise – at a higher rate than cocaine production – but despite that, 14,000 North Americans have died annually of cocaine overdose since 2017. That remained the case through 2021. This reflects an increase in harmful cocaine use due in part to fentanyl being introduced into the supply.

A team of UMD researchers is studying how to map and disrupt the cocaine supply chain. The research is supported by the National Science Foundation (NSF) and is led by the Smith School’s S. Raghu Raghavan.

The latest published work within the larger project uses data from the DEA’s System to Retrieve Information from Drug Evidence (STRIDE) to map cocaine trafficking networks in the U.S. “What we used was price information, basically the information about prices of any undercover purchases,” says Margrét V. Bjarnadóttir, associate professor at the University of Maryland Robert H. Smith School of Business. She co-authored the research with Greg Midgette, assistant professor of criminology and criminal justice at UMD, Siddharth Chandra, a Michigan State University professor and Pengfei He, a PhD candidate at Michigan State.

The STRIDE dataset reveals the states where police made undercover buys of cocaine as well as “how it was bought and how much was paid for it.” Bjarnadóttir says, “the key research question is what can we learn about illicit networks via this information?” STRIDE  doesn’t have anything in terms of the shape of the network, but “what it does give us information about, is pricing in different areas of the United States.”

It turns out, there’s much to be learned from the changing cost of the drug as it travels from the coca fields where it’s grown in Colombia – the world’s biggest producer of cocaine – to its destination in America. There are direct costs like the price of transport, “the gas for the truck, cost of a plane ride, etc.,” says Bjarnadóttir. But there are also indirect costs, like the risk involved with supplying cocaine. For example, the person moving the drugs might get arrested. So, she says, “if you move drugs from point A to point B you expect the price to go up.” That’s one way to learn the structure of the underlying supply chain, “because we know the prices move in the direction of increased costs.”

“We can further use price co-movement as a measure of how strongly connected two locations are.” Bjarnadóttir says if two locations are linked, then a shock in one location affects all connected locations up the supply chain, “two places are connected in the network if they move together.” This is how she and her colleagues constructed the cocaine supply chain.

They used optimized data combined from several measurements. “You have different data densities depending on location.” Bjarnadóttir uses California and North Dakota as examples. “There’s much more data from California than North Dakota because California has a larger population and more cocaine activity than North Dakota. But then we need to align the data to understand the co-movement between the two states. We do that using optimization.”

The cocaine supply chain mapping will give law enforcement something to use other than anecdotal evidence like that gained through arrests and drug seizures. Bjarnadóttir thinks the data-driven network she and her co-authors constructed, “will help police build drug intervention strategies. That’s where these kinds of methods can contribute.”

The Research, Analyzing Illegal Psychostimulant Trafficking Networks Using Noisy and Sparse Data, has been published in IISE Transactions, the flagship journal of the Institute of Industrial and Systems Engineers.