MIS 752 · Lab 3 · Mikey Atwood

Where Drug-Enforcement Resources Go

A 30-second view of police spending on drug offenses: how concentrated it is, which offenses drive it, and why the average cost per case misleads. Built on 845,000 synthetic 2025 incidents.

81%

of police cost goes to the costliest 10% of incidents

0.79

Gini coefficient (0 = equal, 1 = one incident gets everything)

$1,350

median incident cost; the mean is $8,357 (6.2x higher)

$35,190

cost line for the top 10% of incidents

The question

A city's public safety director has five minutes with the city manager before the budget locks. The question: where are drug-enforcement resources actually going?

The answer is not "about $8,357 per case." That average describes almost no real case. Most incidents are simple possession arrests costing around $1,350; a small number of trafficking, manufacturing and distribution investigations cost tens of thousands each. When spending is this lopsided, the average misleads and the distribution is the story.

How concentrated is the spending?

The top 1% of incidents alone account for 12% of the year's cost. This year's example case, a trafficking investigation costing $51,151, sits at the 92th percentile: more expensive than 92% of all drug incidents.

Which offenses drive it?

Offense	Incidents	Share of incidents	Share of police cost	Average cost per incident
Possession	745,797	88.3%	13.2%	$1,250
Trafficking	49,468	5.9%	41.9%	$59,879
Manufacturing	24,788	2.9%	22.7%	$64,719
Distribution	24,947	3.0%	22.1%	$62,665

What a decision-maker should take from this

Volume is not cost. Possession arrests are 88% of the caseload. If the goal is to free up officer hours, they are the lever on workload; but they are only 13% of the dollars, so diverting them saves less money than the caseload suggests.

Track the tail, not the average. A budget built on the mean will be wrong for almost every case. The Gini coefficient and the top-10% share are numbers a department can track year over year.

What this cannot tell you yet. Whether costs follow crime severity or neighborhood circumstance (poverty, policing intensity, demographics) needs real incident data joined to census data by location. That is the natural next step with an agency's open data portal.

How this was built

The 845,000 incidents are synthetic. The offense mix (88.3% possession) follows the FBI drug-arrest share cited in the notebook (drugpolicyfacts.org); the cost per incident is an assumption: possession $500-$2,000, trafficking $20,000-$100,000, manufacturing $30,000-$100,000, distribution $25,000-$100,000. Because those ranges are assumed, the concentration shown here follows from them. The page demonstrates the method; real cost data would replace the ranges.

Gini, top-decile share and percentiles are computed over every individual incident. Summary tables are in this Space's data/ folder.