6 Ideas for Interpreting Drug Seizure Data
A perennial challenge when analyzing drug trafficking is that the most common reliable data comes from ventures that have failed, i.e. when arrests have occurred or contraband has been seized. There are thousands of drug seizure reports a year, but the question is: what can seizures tell you about the underlying trade?
If there is any certainty in the drug trade, it is that the product must go from the producer to the consumer. That simple fact drives the diversity and intensity of trafficking efforts. For drug trade analysis, the producer and the consumer are the anchor points – for example, the opium farmer, the heroin processor and the person injecting heroin. If you are unearthing a long-broken mosaic, the producer and the consumer are the corner pieces. The traffickers in between are scattered pieces, buried, diversely shaped, multi-colored and changing in the light. Seizures give you a fragment of a piece, but it is hard to know how much you can infer about the broader trafficking picture of which it is a part. The following are a few principles to consider.
Seizures imply the presence of law enforcement. No law enforcement, no seizures. So seizures reflect at least in part the focus of law enforcement resources. More subtly, they reflect the combination of four factors for law enforcement:
presence – motivation – capacity – corruption
This often leads to difficulty discerning whether increased seizures mean increased trafficking or more effective law enforcement. Nevertheless, whatever the combination of factors for a given location, growth in seizures should provide greater appreciation for the size of trafficking in that location.
The seizures themselves are valuable to some extent as reducing the amount of drugs flowing through a network. Rather than focusing on trafficking locations or methods, this encourages an interest in calculating market size and profitability, in order to estimate how important the volume of seizures could be relative to the size of supply.
Although traffickers show some ingenuity, they are also constrained by basic facts such as physical and human geography. For example, if enforcement risks are the same, trafficking is more likely to occur along a highway rather than over a high mountain pass. Similarly, if the only way to get a big shipment out of a country is by sea, then trafficking routes must reach the coast; if they are intended for concealment in long-distance commercial cargo, then they will be attracted to official ports. When such facts overlap in time and space with clusters of seizures, it may generate greater confidence that those seizures genuinely reflect a focus of trafficking.
Law enforcement can involve luck but it also involves predictable patterns of progress. For example, the arrest of one person increases the probability that one of their associates will also be arrested, perhaps because of information coming from interviews with suspects or analysis of their communications history. Instead of seizures being structured by physical geography and infrastructure, this creates a separate tendency for seizure patterns to follow social (trafficking) networks. It means that mapping seizures on a globe may not tell you their most important features or relationships between them. Instead, seizure points can be tagged by reference to a ‘cascade’ of connected interdictions, or be arranged by suspected membership of a single network.
This tendency also works in reverse: a network that is invisible is more likely to stay invisible. In that sense, successful concealment of one venture can pay off multiple times in the future with mitigated risks of detection and lower punishment if caught. The most important reason for effective concealment now may be more profit subsequently.
Over sizeable data sets, patterns in seizures are likely to reflect where trafficking has been more than where it is. This trend largely derives from the other factors above. For example, many attempts through a specific place will eventually generate a seizure, due to luck or information leakages. For another example, success in interdictions leads to greater interest and resources from law enforcement in a particular location, increasing the probability of later seizures (presuming traffickers do not respond – see below). However, both of these are lagging indicators and cannot on their own tell us the degree to which the pattern reflects current and future trafficking.
Instead, seizures in one place could be a reason to predict that trafficking is occurring or will occur somewhere else, because of two possibilities:
1. Traffickers will adapt as far as possible to failure and shift elsewhere.
2. Those organizing low-level couriers are letting some drugs be caught, either out of collusion with law enforcement or out of a sophisticated strategy to maintain law enforcement interest in a place other than their main trafficking routes.
To the degree that either of the above are true, seizures in a place now are a reason for a drug trade analyst to expect trafficking through other routes that are available for similar operations and methods.
Since drug markets usually adapt to seizures without collapsing, seizure data points must be telling you only a small part of the story. That should inspire a focus on generating qualitative information, ideally drawn from actors in the trade itself, for example through law enforcement intelligence efforts. Desk-based statistical analysis of a drug trade is useful for the most basic monitoring, but analysis of trafficking routes is an area of research in which field-based investigations are likely to discover more useful insights and leads.
That means, however, finding a way to combine unstructured reporting with structured seizure reports. In seizure data analysis, straightforward statistics using qualitative case information is rarely used to influence interpretations. For example, if seven out of ten cases with comments on trends in the size of drug consignments say they are getting bigger, to what extent can you extrapolate this to over-ride thousands of other seizures lacking any comments but which show a trend towards smaller seizures?
The difficulty of having more data
When data sets have only a few cases that include qualitative information, they can provide sparse additional knowledge on which to speculate about the rest. It is more difficult, however, when diverse qualitative information is reported for more than a small minority of cases. If possible, these need to be converted to structured data in order to support more meaningful quantitative exploration. If that is not possible or if qualitative information is used sporadically, there is danger that the ease of calculating percentages and making maps from basic data obscures the necessity of unearthing and poring over fragments from the field, in order to lock a piece more confidently into the mosaic.