Victorian Evidence Base · September 2026
This analysis links publicly available disaster declaration, household impact, and family violence data at the Local Government Area level. It identifies a structural gap in Australia's disaster recovery funding framework and outlines the data required to quantify it.
Natural disasters increase the incidence of domestic and family violence. This relationship has been documented in peer-reviewed research across multiple countries, disaster types, and methodologies. Parkinson (2019) observed increased family violence following the 2009 Black Saturday bushfires. Gearhart et al. (2018) found significant post-disaster increases in domestic violence across US events. True (2013) established the link in Pacific disaster contexts. Victoria's Royal Commission into Family Violence (2016) recognised natural disasters as a risk factor for family violence.
The established causal pathways include displacement from housing, financial stress, loss of community support infrastructure, disrupted service access, and trauma. These conditions are present in the aftermath of every major natural disaster.
Australia's Disaster Recovery Funding Arrangements (DRFA) allocate Commonwealth and state resources on the basis of immediate physical damage: dwellings destroyed, infrastructure lost, and roads damaged. Spending thresholds trigger escalating levels of federal cost-sharing (Categories A through D). This model is structurally blind to the social consequences that the peer-reviewed literature identifies as predictable outcomes of household-level disaster impact, including family violence, displacement, and mental health deterioration.
No DRFA trigger exists for proactive family violence service funding. There is no "social consequence multiplier" in the formula. Services in disaster-affected areas absorb increased demand from existing budgets or rely on ad hoc grants allocated after the need has already materialised.
This evidence base assembles publicly available data at the LGA level to support a quantitative assessment of the relationship between disaster impact severity and family violence incidence in Victoria. It identifies two critical data gaps that, if addressed through data requests to the Crime Statistics Agency and Emergency Management Victoria, would enable a dose-response model suitable for informing DRFA reform.
Family violence incidents recorded by Victoria Police rose from 93,439 in 2020–21 to 106,430 in 2024–25. The largest single-year increase of 7,619 incidents (+7.7%) occurred in 2024–25, following sustained major disaster activity across multiple LGAs in the preceding 24 months (CSA 2025).
Victoria experienced 53 DRFA-declared events since October 2020. The October 2022 floods (AGRN 1037) affected 63 LGAs and damaged 5,017 buildings (AIDR 2023). Baw Baw Shire was included in 16 separate declarations — the most of any Victorian LGA.
Published data on per-LGA household impacts was identified for five major events. These figures are compiled from council annual reports, parliamentary inquiry submissions, government recovery publications, and independent reviews — not from a centralised source.
| Event | AGRN | LGAs with dwelling data | Key figures |
|---|---|---|---|
| 2019–20 Bushfires | 882 | East Gippsland, Towong, Alpine | 346 dwellings destroyed in East Gippsland; 38 primary residences in Towong (council reports) |
| June 2021 Storms | 969 | Yarra Ranges, Hepburn, + 9 councils | 71 destroyed, 54 damaged, 60 severely damaged in Yarra Ranges (Parliamentary Hansard) |
| October 2022 Floods | 1037 | Campaspe, Gr. Shepparton, Maribyrnong, Mitchell, Gannawarra | 988 homes inundated in Rochester; 4,000 properties isolated in Shepparton (VAGO 2025; council recovery plans) |
| Dec 2023 – Jan 2024 Storms | 1096 | Greater Bendigo, Mitchell, Loddon | 113 above-floor dwellings in Bendigo; 17 in Loddon (parliamentary submissions; local media) |
| Feb 2024 Storms | 1108 | Yarra Ranges (power outage data) | 37 homes left uninhabitable statewide; 500,000 lost power (AER 2024) |
| AGRN | Date | Hazard | LGAs affected | Notes |
|---|---|---|---|---|
| 1037 | Oct 2022 | Flood | 65 | 5,017 buildings affected across 63 LGAs (AIDR 2023) |
| 969 | Jun 2021 | Flood/Storm | 41 | Category C/D with AGDRP payments; 330,000 lost power |
| 983 | Oct 2021 | Storm | 39 | Category C/D |
| 1108 | Feb 2024 | Storm | 41 | Category C/D; 620,000 homes lost power (AER 2024) |
| 1242 | Jan 2026 | Bushfire | 38 | Category C/D — most recent major event |
| 1096 | Dec 2023 | Storm | 29 | Category C/D; $1.61B insured losses nationally (ICA) |
| 1147 | Aug 2024 | Storm | 28 | Category C/D |
| 882 | Nov 2019 | Bushfire | 18 | 458 houses impacted (IGEM Phase 1); $2.32B national insured losses |
The table below compares family violence trends across 20 Victorian LGAs, grouped by disaster exposure level. The 13 LGAs with six or more DRFA-declared events since 2020 include the state's most disaster-affected communities. Seven metropolitan LGAs with three or fewer events are included for comparison. Family violence data is sourced from CSA annual publications (2020–21 through 2024–25).
| LGA | DRFA Events | FV 2020–21 | FV 2024–25 | Change (%) | Major events |
|---|---|---|---|---|---|
| Baw Baw | 21 | 1,028 | 1,280 | +24.5% | Oct 2022 Floods, Jun 2021 Storms |
| Yarra Ranges | 18 | 1,727 | 2,041 | +18.2% | Oct 2022 Floods, Jun 2021 Storms, Jan 2026 Bushfires |
| East Gippsland | 16 | 1,567 | 2,499 | +59.5% | Oct 2022 Floods, Jun 2021 Storms, Jan 2026 Bushfires |
| Wellington | 16 | 1,343 | 1,508 | +12.3% | Oct 2022 Floods, Jun 2021 Storms, Jan 2026 Bushfires |
| South Gippsland | 15 | 532 | 604 | +13.5% | Oct 2022 Floods, Jun 2021 Storms |
| Macedon Ranges | 13 | 444 | 544 | +22.5% | Oct 2022 Floods, Jun 2021 Storms, Jan 2026 Bushfires |
| Latrobe | 9 | 2,522 | 3,022 | +19.8% | Oct 2022 Floods, Jun 2021 Storms |
| Greater Shepparton | 8 | 1,623 | 2,442 | +50.5% | Oct 2022 Floods, Jan 2026 Bushfires |
| Greater Geelong | 8 | 4,318 | 5,586 | +29.4% | Oct 2022 Floods, Feb 2024 Storms |
| Cardinia | 8 | 1,550 | 2,079 | +34.1% | Oct 2022 Floods, Jun 2021 Storms |
| Wodonga | 7 | 769 | 1,116 | +45.1% | Oct 2022 Floods, Jan 2026 Bushfires |
| Mitchell | 7 | 1,223 | 1,378 | +12.7% | Oct 2022 Floods, Jun 2021 Storms, Jan 2026 Bushfires |
| Ballarat | 6 | 2,367 | 3,020 | +27.6% | Oct 2022 Floods, Jun 2021 Storms |
| Comparison — metropolitan LGAs with low disaster exposure | |||||
| Monash | 3 | 1,809 | 1,923 | +6.3% | Jun 2021 Storms, Feb 2024 Storms |
| Hume | 2 | 4,183 | 3,885 | −7.1% | Oct 2022 Floods, Jun 2021 Storms |
| Brimbank | 2 | 2,989 | 2,898 | −3.0% | Oct 2022 Floods |
| Whitehorse | 2 | 1,461 | 1,495 | +2.3% | Jun 2021 Storms, Feb 2024 Storms |
| Wyndham | 1 | 3,975 | 4,151 | +4.4% | Oct 2022 Floods |
| Melbourne | 1 | 2,138 | 2,178 | +1.9% | Oct 2022 Floods |
| Hobsons Bay | 1 | 1,086 | 1,078 | −0.7% | Oct 2021 Storms |
Among the 13 high-exposure LGAs examined, the mean increase in recorded family violence incidents between 2020–21 and 2024–25 was 28.4%. Among the seven metropolitan comparison LGAs with three or fewer DRFA events, the mean change was 0.6% over the same period. The statewide average increase was 13.9%. East Gippsland — affected by 16 separate DRFA-declared events including the 2019–20 Black Summer bushfires and the October 2022 floods — recorded the largest proportional increase, with incidents rising from 1,567 to 2,499 (+59.5%). Greater Shepparton, where over 4,000 properties were isolated or inundated during the October 2022 floods, recorded the second-largest increase (+50.5%).
These patterns are consistent with the established literature on post-disaster family violence. However, the annual resolution of currently available CSA data limits the capacity to attribute changes to specific events. LGAs with small populations (such as Towong, Mansfield, and Murrindindi) show volatile year-to-year movements that may reflect statistical noise rather than systematic trends. The monthly data identified in Recommendation 1 would enable event-level analysis.
The table below presents year-on-year percentage change in recorded family violence incidents for all 79 Victorian LGAs across 15 calendar-year transitions from 2010 to 2025. The full 16-year series provides ten years of pre-disaster baseline data (2010–2019) alongside the period since DRFA-declared disasters commenced in Victoria in late 2020. The vertical divider and teal header mark the beginning of the DRFA era: columns to its left show trends during a decade of normal variation, while columns to its right show trends during sustained natural disaster activity. Cells are coloured on a red-to-green gradient: dark red indicates an increase exceeding 20 per cent, orange 10–20 per cent, light yellow 0–10 per cent, light green a decrease of up to 10 per cent, and dark green a decrease exceeding 10 per cent. LGAs are sorted by total DRFA event declarations in descending order, with baseline (zero-event) LGAs separated at the bottom. Where DRFA-declared disasters commenced in the corresponding calendar year, a small badge shows the event count. The critical comparison is whether disaster-affected LGAs — which showed normal variation before 2020 — exhibit distinctly different trends in the DRFA era.
Loading heatmap data…
The heatmap above shows volatile year-to-year movements across all exposure groups. To determine whether the observed pattern is statistically meaningful — or simply noise in annual data — we apply panel econometric methods, hypothesis tests, and correlation analysis to the full 79-LGA × 16-year dataset. All tests use publicly available annual data. Standard errors are heteroskedasticity-robust (HC1).
Mean annual YoY percentage change in FV incidents, grouped by total DRFA event exposure. The pre-disaster period covers 2011–2019 transitions; the post-disaster period covers 2020–2025 transitions.
| Exposure group | LGAs | Pre-disaster (2011–2019) | Post-disaster (2020–2025) | ||||
|---|---|---|---|---|---|---|---|
| Mean % | SD | n | Mean % | SD | n | ||
Panel regressions with LGA and year fixed effects on log FV incidents. The dependent variable is ln(1 + FV incidents). Fixed effects absorb baseline differences between LGAs and common annual trends, isolating the within-LGA effect of disaster exposure.
| Model | Coefficient | % effect | SE | p-value | 95% CI |
|---|
Tests of parallel trends (pre-disaster), post-disaster divergence, event-study comparisons for AGRN 1037 (October 2022 floods), and cross-sectional correlation between total DRFA events and cumulative FV change (2019→2025).
| Test | Statistic | p-value | Key finding |
|---|
The data required to build a dose-response funding model largely exists within Victorian government agencies. Two critical datasets are collected but not published in the form required for LGA-level analysis.
| Dataset | Source | Status | Resolution | Gap |
|---|---|---|---|---|
| FV incidents by LGA | Crime Statistics Agency Victoria | Published | Annual by financial year | Monthly resolution required for event-study design |
| DRFA activations by LGA | NEMA via data.gov.au | Published | Per-event | Binary activation only — no household impact severity |
| AGRN Disaster Register | DisasterAssist.gov.au | Published | Per-event | — |
| Population by LGA | Australian Bureau of Statistics | Published | Annual | — |
| SEIFA disadvantage index | Australian Bureau of Statistics | Published | Census 2021 | — |
| AIDR Disaster Events (Victoria) | Australian Institute for Disaster Resilience | Published | Per-event | Historical context (163 events to 2010); limited post-2010 |
| LGA household impacts (5 events) | Council reports, parliamentary inquiries | Compiled | Per-event × per-LGA | Covers 17 LGAs across 5 events; not comprehensive |
| Monthly FV by LGA | Crime Statistics Agency (requires data request) | Not published | Monthly | Underlying LEAP data records date and location; only the LGA disaggregation is unpublished |
| Damage assessments by LGA | Emergency Management Victoria (requires data request) | Not published | Per-event × per-LGA | Rapid Damage Assessment data exists in DRFA Claims Management System |
The following datasets are available for download. All source data is provided in the form in which it was obtained, with provenance documented in the accompanying README.
Crime Statistics Agency data is published under Creative Commons Attribution 4.0. NEMA DisasterAssist data is published under Creative Commons Attribution 2.5 Australia. AIDR data is published for public use. Council reports and parliamentary submissions are public documents. This compilation is provided for research purposes.
The relationship between natural disasters and family violence is established in the literature. The analytical objective of this evidence base is to support quantification of the magnitude and timing of that relationship at the Victorian LGA level, in a form suitable for translation into a disaster recovery funding formula.
Disaster declaration data was obtained from the NEMA DisasterAssist register (data.gov.au) and cross-referenced with the AIDR disaster mapper. LGA-level family violence data was obtained from CSA annual publications. Per-LGA household impact data was compiled from publicly available council reports, parliamentary inquiry submissions, government recovery publications, VAGO audit reports, and local media coverage for five major events (AGRN 882, 969, 1037, 1096, 1108).
The disaster–FV panel dataset merges DRFA activation records with annual FV counts by LGA. This permits descriptive comparison of FV trends in disaster-declared versus non-declared LGAs, but the annual temporal resolution limits causal inference.
With monthly FV data by LGA, a difference-in-differences design would compare FV rates in disaster-declared LGAs against non-declared LGAs before and after each event, with LGA and time fixed effects absorbing baseline differences. This would measure the additional FV incidence attributable to disaster impact. The literature suggests FV effects peak 6–12 months post-disaster (Parkinson 2019), which monthly data would capture.
With per-LGA household damage counts from EMV, this analysis would estimate whether LGAs with greater dwelling damage experience proportionally larger FV increases. Count models (Poisson or negative-binomial with population offset) are appropriate given the overdispersion typical of crime data. Chronically affected LGAs such as Baw Baw (16 separate declarations) would require a stacked Callaway–Sant'Anna estimator to handle treatment heterogeneity across multiple events.
The target output is a coefficient of the form: for each additional N dwellings damaged in an LGA, an estimated Y additional family violence incidents are expected over the following Z months. This coefficient, combined with cost-of-service estimates, would provide the basis for a "social consequence multiplier" in disaster recovery funding allocations.
All data processing scripts are included in the source repository. The pipeline consists of: raw data download and cataloguing; LGA name normalisation (mapping council names, shire/city suffixes, and historical names to ABS LGA identifiers); disaster–FV panel construction; household impact extraction from event-specific CSV files; and summary statistics generation.
Two data requests and one policy reform are identified. The data requests are preconditions for the quantitative analysis required to inform the policy reform.
A request for monthly family violence incident counts by LGA from January 2018 to the most recent available period. The CSA publishes monthly data at the state level and annual data by LGA; the underlying LEAP database records date and location for each incident. The LGA-by-month disaggregation is the minimum temporal resolution required for an event-study design.
info@crimestatistics.vic.gov.auA request for per-LGA counts of residential properties assessed by damage category (destroyed, uninhabitable, damaged but habitable, minor damage) for all 53 Victorian DRFA-declared events since October 2020. This data exists within the DRFA Claims Management System and is collected through the Rapid Damage Assessment process conducted after every declared disaster. It is the essential input for dose-response analysis.
emv.info@emv.vic.gov.auThe peer-reviewed literature establishes that disasters increase family violence. If the dose-response analysis confirms the magnitude at the Victorian LGA level, the evidence would support inclusion of a social consequence component in DRFA funding allocations. This component would automatically trigger funding for family violence services, emergency housing, and legal assistance in proportion to household impact — allocated prospectively at the point of disaster declaration, based on damage assessment data, rather than retrospectively after services have been overwhelmed.
The analysis assembled in this evidence base, combined with the requested EMV and CSA data, would provide the quantitative basis for specifying the multiplier: for every N dwellings damaged, fund Y additional hours of DV casework, Z emergency housing placements, and W legal assistance sessions.