How the numbers actually get calculated in Oregon
The basic premise of any crime cost model is deceptively simple: take the number of reported incidents in a given category, multiply by an average cost-per-crime, and sum across categories. What happens after that is where things get messy. Oregon uses a modified version of the methodology originally developed by the National Council of Crime and Delinquency, but adjusted for state-level wage data, healthcare costs, and criminal justice system expenditures specific to the Pacific Northwest. The result is a set of annual reports that many people cite, few actually read past the executive summary, and even fewer understand how the figures were derived. OCCIF stands for Oregon Crime Cost Forecasting, which is the state's formal methodology for estimating the economic impact of different categories of criminal activity. It is maintained by the Oregon Criminal Justice Council in coordination with the Office of Economic Analysis. The model produces annual estimates covering property crime, violent crime, drug offenses, and white-collar crimes, with separate calculations for direct costs (medical care, victim losses, law enforcement response) and indirect costs (lost productivity, long-term healthcare, reduced quality of life). The most recent cycle produced a total statewide cost estimate of approximately $14.2 billion across all reported categories, though the breakdown between categories shifts noticeably year to year based on changes in reporting rates and incident counts. The actual calculation involves several data sources layered on top of each other. Uniform Crime Reporting data from the FBI feeds the baseline incident counts. Oregon-specific average restitution payments come from the Department of Corrections case management system. Medical costs are derived from Oregon Health Authority hospital discharge data and adjusted using state-specific cost-per-case averages rather than national benchmarks, which is one of the things that makes the Oregon model meaningfully different from running a NCCD calculator on raw national data. Lost earnings use Oregon Bureau of Labor Statistics wage indexes by county and age group, which matters because Portland metro wages are substantially higher than rural areas and the cost model accounts for that.
I spent three days last fall reconciling a set of cost estimates for a Clackamas County prosecutor's office that needed updated figures for a grant application. The standard OCCIF output was from fiscal year 2022-2023, but they needed 2024 estimates. Running a quick projection based on percentage change from the prior two years gave numbers that were clearly too high — burglary costs had spiked disproportionately because there had been a regional surge in vehicle break-ins that year, which inflated the property crime category enough to make the per-crime average look inflated for general purposes. What I ended up doing was pulling the raw incident counts from the Oregon State Police dashboard, calculating the change specific to Clackamas County rather than applying a statewide adjustment factor, and re-anchoring the per-incident cost using the most recent available Department of Revenue tax return data to adjust for local income shifts. The final adjusted estimate came in about eight percent lower than the straight projection would have suggested, which turned out to be closer to what the actual incident reports showed when I cross-referenced with the county sheriff's quarterly statistics. One thing that most people who work with this data get wrong the first time is assuming the cost-per-crime figures are static. They are not. The model updates its multiplier values every cycle, but the updates are conservative and lag behind actual changes in costs like emergency room visits or legal processing expenses. During the 2020-2021 period, for example, the per-assault medical cost in the model remained anchored to pre-pandemic hospital rates for nearly two full cycles because the data collection pipeline moves slowly. If you are citing these numbers in a report or presentation and the figures look slightly low compared to what practitioners on the ground are actually seeing, that is usually why. Another counter-intuitive detail is how victim costs are treated. The model does not simply calculate the value of stolen property at retail price. It uses depreciated replacement cost for most property crimes, which means a five-year-old laptop that gets stolen and is valued at full current retail in your head will show up in the OCCIF data at roughly forty percent of that value. That is not a bug, it is by design — the model is trying to capture actual economic loss, not the price tag someone might slap on something during an emotional reaction. It does matter when you are trying to explain to a city council why the property crime cost figure seems lower than the sum of individual victim claims. They are measuring different things.
The indirect cost component is where the model does the most speculative work. Lost productivity from incarceration, long-term healthcare for violent crime victims who suffer permanent injuries, and the quality-of-life adjustments for communities with high crime rates all require assumptions that vary considerably depending on who is making them. The Oregon model uses a willingness-to-pay approach for the quality-of-life adjustment, which means it is essentially asking survey respondents what they would pay to reduce crime risk in their area. That produces a number, but it is fundamentally different from the hard accounting of direct costs and carries a much wider confidence interval. I have seen people cite the total cost figure including indirect costs as if it were a settled calculation, when the indirect portion alone can vary by fifteen to twenty percent depending on which survey data and discount rate assumptions are applied. If you need the actual data, the Oregon Criminal Justice Council publishes the full methodology report and the underlying dataset on their website. The raw numbers are available in CSV format, which is useful if you want to do your own adjustments rather than working from the published summary tables. The download is free, no registration required, and the file is usually under two megabytes. I tend to pull the CSV and run a quick check on the category totals against the published report before using anything — it takes maybe five minutes and has saved me from citing an outdated figure at least twice in the last two years. The main limitation to keep in mind is that OCCIF covers only reported crimes. Unreported crimes, particularly sexual assault and domestic violence, are estimated using supplementation factors from the National Crime Victimization Survey, but those factors carry substantial uncertainty. The model also does not include organized retail theft as a separate line item in a way that captures the full scale of the problem as it has been described in legislative hearings over the last three years. Property theft that falls under organized retail crime gets folded into standard retail theft categories, which means the cost figures for that segment are likely understated relative to what law enforcement agencies are reporting in their own internal assessments.
Get the Full Details
For anyone doing serious analysis, I would recommend pairing the OCCIF data with the Oregon State Police's supplementary incident-level data and the Department of Corrections' sentencing and cost-per-inmate figures. That combination gives you enough granularity to adjust the statewide averages for your specific county or demographic group. The OCCIF numbers are a solid starting point, but they are not fine-grained enough to stand alone if you are making decisions about resource allocation or policy recommendations at the local level.
