Police Funding in Canadian Municipalities: Longitudinal Study – Comprehensive Study Notes

Police Funding in Canadian Municipalities: A Longitudinal Study — Study Notes

  • Topic and scope

    • Examines municipal police service expenditures and crime rates over time in Canada.
    • First Canada-wide longitudinal analysis of police funding across multiple cities.
    • Time frame: 2010–2021 for budgets; crime data 2010–2019 (CSI); population data up to 2020; inflation-adjusted to 2020 dollars.
    • Data sources: municipal operating budgets (re budget documents), Statistics Canada population estimates, Statistics Canada Crime Severity Index (CSI).
    • Data are made public in the Online Appendix; limitations exist due to missing budgets for several municipalities.

  • Research questions and aims

    • (1) How much do municipalities spend on policing?
    • (2) How does spending on policing vary across Canada?
    • (3) To what extent are variations in police spending associated with changes in crime rates over time?
    • Hypotheses (pre-registered in paper): expect no strong positive or negative correlations when data are disaggregated; no causal claims; anticipate variation by municipality and context.
    • Emphasis on associational evidence rather than causality; calls for more robust multivariate evaluations in future work.

  • Key findings at a glance

    • Police services were the top operating expenditure in a majority of municipalities in 2019 (12 of 20; 60%).
    • Mean share of police spending in total operating budgets in 2019: $%
    • Per-capita police spending in 2019 (real 2020 dollars): mean $342.28 (SD $75.67); Vancouver highest ($495.84 per person); Quebec City lowest ($217.05 per person).
    • Across 2010–2020, per-capita spending increased in 16 of 20 municipalities; several municipalities showed notable increases or decreases (e.g., Quebec City +$63.14 per capita; Toronto −$4.59; Montreal −$41.94).
    • There is substantial heterogeneity in the relationship between police funding and crime across municipalities; no uniform inverse association across all cities.
    • Correlations between police funding and CSI are generally weak and inconsistent across municipalities; some notable exceptions (e.g., Gatineau negative, Saskatoon positive).
    • Lagged analyses (change in spending in year t with CSI in year t+1) show weak overall associations, with one municipality (Winnipeg) showing a significant negative lag effect (r = −0.716, p = 0.046).
    • Net change in CSI and net change in spending per capita across all municipalities show a modest negative trend but not strong enough to support causal inferences (average r ≈ −0.206, p ≈ 0.035 for the lagged relation; overall net-change correlation r ≈ 0.43, p ≈ 0.061).

Data and sample

  • Municipalities studied
    • 20 largest urban municipalities or regions by population (as of 2020):
    • Toronto, Montreal, Peel Region, Calgary, York Region, Edmonton, Ottawa, Winnipeg, Vancouver, Waterloo Region, Surrey, Quebec City, Hamilton, Halifax, Laval, London, Gatineau, Saskatoon, Burnaby, Longueuil.
    • Geographic spread across seven of Canada’s ten provinces; 2020 population estimates ranged from 2,988,408 (Toronto) to 252,828 (Longueuil).
  • Budget data and collection
    • Budgets and operating expenditures (2010–2021) collected from municipal websites; where not publicly available, contact by email/phone.
    • When actual expenses were unavailable, used approved budget amounts or projected spending.
    • Population estimates from Statistics Canada; CSI data from Statistics Canada.
    • All budget data adjusted to 2020 Canadian dollars using Bank of Canada inflation calculator.
    • Some municipalities lacked complete budget data for all years (only 11 of 20 had complete 2010–2021 budgets); missing years noted (Peel, Waterloo, Surrey, Hamilton, London, Saskatoon, Longueuil).
  • Crime data
    • Crime Severity Index (CSI) data collected for 2010–2019; not available for 2020–2021 at collection time.
    • CSI accounts for both the volume and severity of police-reported crimes; adjusted for population; standardized to 100 with base year 2006.
  • Inflation and currency adjustments
    • All budget figures inflated to 2020 Canadian dollars for comparability.
    • When comparing 2019 budgets across municipalities, 2019 was used as the comparison year because it had the most complete data across all municipalities at the time of collection.

Measures and variables

  • Police funding measures
    • Gross policing budget: total operating expenditures for police services (inflation-adjusted to 2020 dollars).
    • Per-capita spending on police services: real spending per resident, i.e., spending per capita adjusted to 2020 dollars.
    • Proportion of municipal gross operating budget allocated to police services: share (%) of the total operating budget.
  • Crime measure
    • Crime Severity Index (CSI): accounts for both volume and severity of police-reported crimes; population-adjusted; standardized to 100 using base year 2006.
    • CSI used as a continuous outcome; values available 2010–2019; used in correlation and lag analyses.
  • Covariates and comparability notes
    • Population estimates (2010–2020) from Statistics Canada; 2021 not available at time of data collection.
    • Budget data include both tax and non-tax revenues (gross expenditures); BLM studies suggest tax-dollar shares may differ from gross budget shares.
    • Comparability limited by municipal budget organization (some include utilities, others do not) and by differences in property tax bases and urban form.

Analytical approach and methods

  • Descriptive statistics
    • Assess share of police spending in total budgets; identify top/bottom per-capita spenders in 2019.
    • Examine year-by-year trends in per-capita and gross spending; identify increases/decreases and years with exceptions.
  • Longitudinal analysis
    • Evaluate changes in police spending per capita and gross spending over time (2010–2019/2020 depending on data availability).
    • Compute year-to-year changes and net changes over the study period.
  • Correlation analyses
    • Pearson correlation coefficients to assess associations between:
    • Real per-capita police funding and CSI (2010–2019). Report r and p-values.
    • Change in real per-capita police funding (2011–2018) and change in CSI in the next year (2012–2019) to account for potential lag effects. Report r and p-values.
    • Net change in CSI and net change in spending per capita across municipalities (2010–2019). Report r and p-values.
    • Significance threshold: p < 0.05.
  • Sensitivity and robustness considerations
    • Analyses using real-dollar spending (inflation-adjusted to 2020 dollars).
    • Acknowledgement of missing data and potential biases due to incomplete budgets for several municipalities.
    • Recognize that correlations do not imply causation; results are associative and context-specific.

Key results and details

  • Budget prominence and regional variation (descriptive findings)
    • In 2019, police services were the largest operating expenditure in 12 of 20 municipalities (60%).
    • Police was the second-largest line item in 8 municipalities (Ottawa, Laval, London, Gatineau, Saskatoon, Longueuil, etc.).
    • Mean proportion of the municipal operating budget allocated to police services in 2019: 15% (SD = 5%).
    • The municipality with the largest share of its budget for police in 2019: Winnipeg (26%). The smallest shares were in Quebec City and Ottawa (8% each).
    • Mean per-capita spending on police services (2019, 2020 dollars): xˉpc=342.28,extSD=75.67.\bar{x}_{pc} = 342.28, ext{ SD} = 75.67.
    • Highest 2019 per-capita spending: Vancouver ($495.84). Lowest: Quebec City ($217.05).
    • Range across municipalities in 2019 per capita: 278.79.278.79. Vancouver spent more than double per capita compared with Quebec City in 2019.
    • The minimum per-capita spending across 2010–2020 for Quebec City was $155.26 in 2010; the maximum across the studied period was $560.68 in Longueuil (2017).
  • Time trends in police funding (descriptive longitudinal findings)
    • Across the 11-year window with complete data, most municipalities showed net increases in police spending (16 of 20; 80%).
    • Notable decreases in per-capita spending included: Toronto (−$4.59; −1.10%), Montreal (−$41.94; −9.98%), Calgary (−$4.43; −1.15%), Halifax (−$12.28; −4.23%).
    • Quebec City had the largest net per-capita increase (+$63.14; +41.67%).
    • When considering the period ending in 2019, similar patterns held for changes in gross and per-capita spending (Appendix A).
  • CSI trends
    • CSI values showed a net decrease in 15 of 20 municipalities over 2010–2019; mean net change was −9.09 (SD = 14.35).
  • Correlations: police funding and crime (cross-sectional across municipalities)
    • Real per-capita policing and CSI (2010–2019): average r = −0.192 (p = 0.115).
    • Notable municipality-level correlations (selected, p < 0.05):
    • Gatineau: r = −0.860, p = 0.001 (strong negative relationship).
    • Saskatoon: r = 0.804, p = 0.016 (strong positive relationship).
    • Other municipalities showed weaker or non-significant associations (e.g., Toronto r = −0.519, p = 0.125; Edmonton r = 0.777, p = 0.008).
  • Lagged correlations: change in spending and CSI in the next year
    • Average correlation between year-on-year spending changes and CSI change in the following year: r = −0.206, p = 0.035.
    • The strongest negative lagged relation: Winnipeg, r = −0.716, p = 0.046.
    • Other municipalities showed weaker lagged associations; overall pattern did not show consistent evidence that spending reductions lead to crime increases or that spending increases reliably precede CSI reductions.
  • Net-change correlation across all municipalities
    • Correlation between net change in CSI and net change in spending per capita (2010–2019): r = 0.43, p = 0.061 (not statistically significant at 5% level, but suggests a weak positive association in this sample).
  • Overall interpretation
    • No consistent inverse (crime-reducing) effect of increased police funding across municipalities.
    • Substantial local variation indicates that context, local policy, and other socio-economic factors likely shape the police–crime relationship.
    • Findings align with broader literature that changes in police budgets alone do not reliably predict crime outcomes; cautions about causality and the importance of multivariate analyses.

Tables and figures referenced

  • Table 1: Changes in per-capita and gross spending on police services by municipality (2010–2021)
    • Reports net differences, average annual changes, and inflation-adjusted values (2020 dollars).
    • Example entries: Quebec City +63.14 per capita; Toronto −4.59 per capita; Winnipeg +52.19 per capita; etc.
  • Figure 1: Police spending per capita and proportion of budget allocated to police (2019)
    • Visualizes cross-municipality variation in per-capita spending and budget shares.
  • Table 2: Correlation between real per-capita police funding and CSI (2010–2019)
    • Includes r and p-values by municipality; excludes Waterloo and Surrey due to data gaps; average r and 95% CI reported.
  • Table 3: Correlation between change in real per-capita funding (2011–2018) and change in CSI the next year (2012–2019)
    • Includes r and p-values; average and 95% CI reported.
  • Appendix A: Changes in per-capita and gross spending on police services, time period ending in 2019
    • Provides a complementary look at the 2010–2019 window; inflation-adjusted values; percent and absolute changes.

Methodological notes and limitations

  • Data completeness and comparability
    • Only 11 of 20 municipalities had complete budget data for 2010–2021; some years were missing for others (1–3 years missing; Surrey missing six years; Waterloo Region missing eight years).
    • Differences in municipal budget organization (e.g., inclusion of utilities) limit cross-city comparability of police-budget shares.
    • Property tax structures and urban form vary by municipality and province, influencing baseline resources and budget allocations.
  • Crime measurement limitations
    • CSI was used as the sole crime outcome; other measures (e.g., disaggregated crime types, arrests) could yield different patterns.
  • Causality and interpretation
    • Analyses are correlational and do not establish causality between police funding and crime outcomes.
    • Temporal ordering is addressed to some extent via lag analyses, but alternative time lags could produce different results.
  • Data scope limitations
    • 2020–2021 CSI data were not available at collection; some post-2019 trends are not captured.
    • Provincial and federal funding for law enforcement were outside the scope of this study but could influence municipal budgets.
  • Recommendations for future research
    • Employ more robust multivariate models with demographic, economic, and policy controls.
    • Explore alternative crime metrics (e.g., property vs. violent crime; arrests) and longer time lags.
    • Seek standardized, accessible municipal policing-finance data to improve comparability and replication.

Implications and conclusions

  • Variation and policy relevance
    • Municipal police funding varies widely across Canada and has generally increased since 2010, with police budgets constituting a meaningful share of operating budgets in many cities.
    • The lack of a uniform pattern linking higher police spending to lower crime suggests that policing is only one piece of a broader public-safety ecosystem; factors such as demographics, economic conditions, employment rates, and social services likely play critical roles.
  • Research and policy takeaways
    • The study provides a foundation for future Canadian policy analysis of police funding and its effectiveness.
    • Policymakers should interpret correlational findings with caution and consider comprehensive evaluations that incorporate broader contextual factors and causal modeling.
  • Data accessibility and ethics
    • The study highlights persistent data accessibility issues in municipal policing budgets; improving transparency could enable more timely and actionable policy analysis.
    • Given concerns around policing and community safety, rigorous, transparent research is essential to inform debates about resource allocation and governance.

Concluding remark

  • The paper presents a pioneering Canadian longitudinal view of police funding across major cities, showing wide cross-city variation and complex associations with crime; it emphasizes correlational patterns, data limitations, and the need for richer, multivariate, and causal analyses to inform evidence-based budgeting for safe and healthy communities.

References and context (selected)

  • Core literature cited includes Ruddell and Thomas (2015) on determinants of police strength in Canadian cities; Fraser Institute analyses (Di Matteo 2014); U.S. policing-budget-crime literature (Marvell & Moody 1996; Levitt 2002; Kovandzic & Sloan 2002; Becke/Gollstein/Bennett references in the article).
  • CSI methodology and Statistics Canada data sources are foundational to the analysis and interpretation presented in this study.

Data availability

  • Datasets used in the study are publicly available in the Online Appendix accompanying the article.

Notation and equations used in the study (LaTeX)

  • Crime Severity Index description
    • CSI accounts for both the volume and severity of police-reported crimes, adjusted for population, and standardized to 100 with base year 2006:
    • CSIextisadjustedforpopulationandstandardizedto100extwithbaseyear2006.CSI ext{ is adjusted for population and standardized to } 100 ext{ with base year } 2006.
  • Inflation adjustment for budget comparisons
    • Budget<em>2020=Budget</em>nominalimesCPI<em>2020CPI</em>nominalBudget<em>{2020} = Budget</em>{nominal} imes \frac{CPI<em>{2020}}{CPI</em>{nominal}}
    • Budgets were converted to 2020 Canadian dollars using the Bank of Canada inflation calculator.
  • Real per-capita spending and correlations
    • Real per-capita spending and CSI correlations reported as Pearson r values with p-values, e.g.:
    • rext(realper−capitapolicing,CSI)=−0.192ext(p=0.115)r_{ ext{(real per-capita policing}, CSI)} = -0.192 ext{ (p = 0.115)}
  • Lagged relationships
    • Lagged correlation between changes in spending and CSI in the next year reported as:
    • r<em>extΔSpend,extCSI</em>t+1=−0.206ext(p=0.035)r<em>{ ext{ΔSpend}, ext{CSI}</em>{t+1}} = -0.206 ext{ (p = 0.035)}
  • Net-change relationship
    • Correlation between net changes in CSI and net changes in spending per capita across municipalities:
    • r<em>extCSI</em>extnet,extSpendextnet=0.43ext(p=0.061)r<em>{ ext{CSI}</em>{ ext{net}}, ext{Spend}_{ ext{net}}} = 0.43 ext{ (p = 0.061)}