IESET.
Hypotheses·labour·immigration_crime_rate_vs_native_controlled

Per-capita crime rates (measured by police-recorded offences per 100k population, by offence type) among foreign-born residents in developed destination countries are NOT systematically higher than among native-born residents once age, gender, and socioeconomic status are controlled.

This aggregate pattern is the robust finding of the criminology meta-literature (Light-Miller 2018; Bell-Fasani-Machin 2013; Fasani-Mastrobuoni-Owens-Pinotti 2019). Within the aggregate, there IS measurable variation by specific origin group, duration of residence, and legal status — but these differences are best explained by (a) selection effects in the migration channel (e.g., asylum cohorts have different age/gender distributions than native population, producing age-composition-driven offending differentials before any ethnicity effect), (b) labour-market integration policy, (c) neighbourhood concentration and policing intensity, NOT by ethnicity as a causal variable.

INCONCLUSIVEengine/runs/immigration_crime_rate_vs_native_controlled

INCONCLUSIVE_DATA_PENDING — no decomposition channel loaded; missing: ['constructed: % of group aged 15-34 (primary offending age band); WDI + destination-country admin data', 'constructed: % male in group', 'constructed: median income / unemployment rate / education attainment for the group; destination-country admin data', 'constructed: average duration of residence', 'constructed: % of group on work-visa / family / asylum / undocumented']

confidence cueResult card produced; verdict unclassified.

policy briefCoverage too thin

In ordinary language

In plain terms, this asks whether age composition is a real pathway to better or worse crime rate per capita by nativity and offence from 2000 to 2023.

plain answer

This test cannot make a firm call yet. no decomposition channel loaded; missing: ['constructed: % of group aged 15-34 (primary offending age band); WDI + destination-country admin data', 'constructed: % male in group', 'constructed: median income / unemployment rate / education attainment for the group; destination-country admin data', 'constructed: average duration of residence', 'constructed: % of group on work-visa / family / asylum / undocumented']

why it matters

Labor-market rules often help some workers while risking job loss or slower hiring for others. This test looks for that tradeoff in observable employment or unemployment data.

how the test works

It compares 8 country or place units from 2000 to 2023, using a panel fe decomposition design, with fixed effects for destination country and year.

what was measured
Possible pathway
  • Age composition
  • Gender composition
What we checked
  • Crime rate per capita by nativity and offence
what this does not prove

A single test is not the whole truth. It narrows the claim under a specific sample, time period, and method. Strong policy conclusions need the pattern to survive nearby tests, alternative data, and serious objections.

verification

No evidence packet has been generated yet.

Results

engine/runs/immigration_crime_rate_vs_native_controlled
1007550250200020122023USAGBRDEUNLDSWENORDNK
illustrative sketch · run pending
No coefficients yet. When the model fires, this chart will show crime_rate_per_capita_by_nativity_and_offence across 8 sampled countries over 2000–2023.
The shapes above are stylised — none of the lines are real data.
Placeholder for immigration_crime_rate_vs_native_controlled. Published chart will be generated from engine/runs/immigration_crime_rate_vs_native_controlled/chart_data.json.

Pre-registration

pre-registered
first-spec commit bae09ab · 2026-04-29T22:09:42Z
run generated · 2026-06-29T17:53:36Z

Per-capita crime rates (measured by police-recorded offences per 100k population, by offence type) among foreign-born residents in developed destination countries are NOT systematically higher than among native-born residents once age, gender, and socioeconomic status are controlled. This aggregate pattern is the robust finding of the criminology meta-literature (Light-Miller 2018; Bell-Fasani-Machin 2013; Fasani-Mastrobuoni-Owens-Pinotti 2019). Within the aggregate, there IS measurable variation by specific origin group, duration of residence, and legal status — but these differences are best explained by (a) selection effects in the migration channel (e.g., asylum cohorts have different age/gender distributions than native population, producing age-composition-driven offending differentials before any ethnicity effect), (b) labour-market integration policy, (c) neighbourhood concentration and policing intensity, NOT by ethnicity as a causal variable.

Falsification criterion — what would disprove this

set before the run · honoured after

This hypothesis is considered falsified if:

Not supported if, after controls for age-sex-SES + policing, the log- rate ratio between foreign-born and native-born is positive and statistically significant at p<0.05 for violent offences specifically (the highest-salience offence type) in a majority of sample countries AND the gap does NOT close materially between first- and second- generation. Either finding would indicate that channel decomposition is incomplete and ethnicity-or-unmeasured-selection effects remain material. A clean pass: age-sex-SES-controlled violent-crime log- ratio within ±0.10 of zero in most countries, and 2nd-generation convergence measurable.

formal test & threshold
test:      crime_rate_channels_decomposition_by_offence_and_generation
threshold: After age-sex-SES controls: |log_rate_ratio(violent)| < 0.10 in ≥60% of sample countries  AND 2nd-generation log-ratio materially closer to 0 than 1st-generation log-ratio

Method

Template
panel_fe_decomposition
Fixed effects
destination_country, year
Clustering
destination_country
Sample
8 countries · 2000 – 2023
Evidence type
causal

Log-odds decomposition: for each country-year, compute crime rate ratio between foreign-born and native-born. Regress log-ratio on the 5 channels + 2 controls + FE. The KEY test: after controlling for age-composition + SES + gender, does the group-level crime-rate differential converge toward zero or native-born baseline? Robustness specs: - Disaggregate by offence type (violent vs property vs drug) — findings often vary; violent-crime ratios are typically NOT higher for foreign-born once controls applied, but property-crime ratios can be elevated for specific low-SES subgroups. - Second-generation separate analysis — convergence to native rates is the expected finding; where it doesn't happen (specific cases in some Nordic countries for some origin groups), the framework needs to examine destination-country integration + policing mechanisms rather than ethnicity. - Arrest rates vs conviction rates vs self-report victimisation — police-recorded rates are contaminated by differential reporting + arrest rates; the NCVS / Dutch Veiligheidsmonitor self-report data often show different patterns than admin.

Data

VariableSourceTransform
crime_rate_per_capita_by_nativity_and_offence
outcome
constructed:police-recorded offence rates per 100k by nativity (foreign-born vs native-born) and offence type (violent / property / tier 5
level_per_100k
age_composition
channel
constructed:% of group aged 15-34 (primary offending age band); WDI + destination-country admin datatier 5
level
gender_composition
channel
constructed:% male in grouptier 5
level
socioeconomic_status_proxy
channel
constructed:median income / unemployment rate / education attainment for the group; destination-country admin datatier 5
level
years_since_arrival
channel
constructed:average duration of residencetier 5
level
legal_status_mix
channel
constructed:% of group on work-visa / family / asylum / undocumentedtier 5
categorical
neighbourhood_concentration
control
constructed:residential segregation index for group (Dissimilarity Index)tier 5
level
policing_intensity
control
constructed:police officers per 100k × local enforcement priority indicatorstier 5
level

● ready  · ● pending  · ● reconstruct-needed

Detailed result card

Result card — immigration_crime_rate_vs_native_controlled

Verdict: INCONCLUSIVE_DATA_PENDING — no decomposition channel loaded; missing: ['constructed: % of group aged 15-34 (primary offending age band); WDI + destination-country admin data', 'constructed: % male in group', 'constructed: median income / unemployment rate / education attainment for the group; destination-country admin data', 'constructed: average duration of residence', 'constructed: % of group on work-visa / family / asylum / undocumented']

Pre-registration

  • Claim: Per-capita crime rates (measured by police-recorded offences per 100k population, by offence type) among foreign-born residents in developed destination countries are NOT systematically higher than among native-born residents once age, gender, and socioeconomic status are controlled. This aggregate pattern is the robust finding of the criminology meta-literature (Light-Miller 2018; Bell-Fasani-Machin 2013; Fasani-Mastrobuoni-Owens-Pinotti 2019). Within the aggregate, there IS measurable variation by specific origin group, duration of residence, and legal status — but these differences are best explained by (a) selection effects in the migration channel (e.g., asylum cohorts have different age/gender distributions than native population, producing age-composition-driven offending differentials before any ethnicity effect), (b) labour-market integration policy, (c) neighbourhood concentration and policing intensity, NOT by ethnicity as a causal variable.
  • Falsification rule: Not supported if, after controls for age-sex-SES + policing, the log- rate ratio between foreign-born and native-born is positive and statistically significant at p<0.05 for violent offences specifically (the highest-salience offence type) in a majority of sample countries AND the gap does NOT close materially between first- and second- generation. Either finding would indicate that channel decomposition is incomplete and ethnicity-or-unmeasured-selection effects remain material. A clean pass: age-sex-SES-controlled violent-crime log- ratio within ±0.10 of zero in most countries, and 2nd-generation convergence measurable.
  • Falsification test: crime_rate_channels_decomposition_by_offence_and_generation

Estimate

  • Error: no decomposition channel loaded; missing: ['constructed: % of group aged 15-34 (primary offending age band); WDI + destination-country admin data', 'constructed: % male in group', 'constructed: median income / unemployment rate / education attainment for the group; destination-country admin data', 'constructed: average duration of residence', 'constructed: % of group on work-visa / family / asylum / undocumented']

Variables resolved

  • constructed: police-recorded offence rates per 100k by nativity (foreign-born vs native-born) and offence type (violent / property / drug / public order). Destination-country admin data — SCB Sweden, BKA Germany, ONS England-Wales, FBI UCR US (with caveats), CBS Netherlands. → crime_rate_per_capita_by_nativity_and_offence (outcome, publisher=constructed, n=192)

Variables missing data

  • constructed: % of group aged 15-34 (primary offending age band); WDI + destination-country admin data (decomposition_channels, name=age_composition) — vintage not on disk
  • constructed: % male in group (decomposition_channels, name=gender_composition) — vintage not on disk
  • constructed: median income / unemployment rate / education attainment for the group; destination-country admin data (decomposition_channels, name=socioeconomic_status_proxy) — vintage not on disk
  • constructed: average duration of residence (decomposition_channels, name=years_since_arrival) — vintage not on disk
  • constructed: % of group on work-visa / family / asylum / undocumented (decomposition_channels, name=legal_status_mix) — vintage not on disk
  • constructed: residential segregation index for group (Dissimilarity Index) (controls, name=neighbourhood_concentration) — vintage not on disk
  • constructed: police officers per 100k × local enforcement priority indicators (controls, name=policing_intensity) — vintage not on disk

Generated by scripts/run_panel_fe.py at 2026-06-29T17:53:36+00:00

Strongest opposing argument

Every hypothesis ships with its charitable opposing argument. The framework earns credibility by handling objections at their strongest, not weakest.

Notes

This is a politically charged topic handled with framework rigour: strong demographic + SES controls, disaggregation by offence type, explicit comparison of 1st vs 2nd generation to test convergence, and a steelman that engages the strongest counter-arguments rather than the weakest. The pre-registration format prevents selective reporting: if the data shows an unexpected elevation in a specific offence type that does not converge, the framework MUST report it honestly per DISCLOSURE.md, while also reporting the destination-country-policy interaction that best explains the result. Data access: US FBI UCR is coarse; NCVS and state-level data more useful. UK Crime Survey England-Wales provides victimisation data. Sweden BRÅ (crime council) publishes detailed breakdowns. Germany BKA and Bundesrepublik statistics are partly legally restricted on ethnicity reporting — Dutch CBS and Norwegian SSB have more permissive breakdowns.

Authored framework. Read the transparency note.