The Alphabet Soup — Part IV:
- Angie D
- Feb 25
- 9 min read
Updated: Mar 18
Parallel Filters Across Ontario’s Oversight Bodies
ESA / MOL
Reverse-Mathing the Wage Theft Funnel
Ontario’s Ministry of Labour reports 11,940 Employment Standards Act (ESA) claim investigations in 2024–25. At face value, that number suggests an active enforcement environment. Thousands of cases are being examined, files are moving, and complaints are being processed.
But 11,940 is not a measure of wage theft. It is a measure of administrative throughput.
An “investigation” reflects a file that has progressed far enough through the intake and screening process to warrant formal review. It does not capture the number of violations experienced. It does not capture the number of workers who never filed. It does not capture the volume of unpaid wages that were never assessed, ordered, or collected.
To understand what 11,940 represents, the number must be placed within its enforcement context.
In the same reporting period, the Ministry reports:
11,940 ESA claim investigations
813 proactive inspections
111 prosecutions (99 Part I tickets and 12 Part III prosecutions)
Now reverse the ratios.
Prosecutions vs. investigations: 111 ÷ 11,940 = 0.9% Less than 1% of investigated ESA claims escalate to prosecution-level enforcement.
Prosecution is not expected in every case. It is typically reserved for serious or repeated violations. However, the ratio reveals the structural ceiling of escalation. Even within the subset of complaints that advance to formal investigation, the proportion that proceeds to enforcement with penal consequences is extremely small.
The system, in other words, is designed primarily for administrative resolution rather than punitive deterrence.
This pattern becomes more significant when layered with wage collection data. Independent reviews of ESA enforcement over various reporting periods have found that only a fraction of wages assessed as owing are ultimately collected. In some periods, collection rates have been cited in the range of roughly 25–35% of assessed wages.
Even allowing for variation across years, the structural implication remains: the amount of unpaid wages assessed is not equivalent to the amount actually returned to workers.
So the 11,940 investigations represent a very specific group of workers — those who:
recognized a violation,
knew their rights,
were willing to file,
navigated the intake process,
did not withdraw,
did not privately settle under pressure,
did not abandon the claim due to delay or fatigue.
The Ministry does not publish comprehensive data on:
the number of workers who experienced violations but never filed,
the number of claims withdrawn or abandoned,
the number of files closed without violation findings,
the number of orders to pay issued,
the percentage of orders fully satisfied.
Without those layers, the public figure of 11,940 appears larger and more complete than it actually is. It reflects cases that entered and remained within the administrative pipeline — not the totality of wage violations occurring in the labour market.
The structural math problem is straightforward:
If 11,940 investigations are visible, fewer than 1% escalate to prosecution, and only a fraction of assessed wages are collected, then 11,940 is not a wage theft number. It is an administrative processing number.
The funnel is narrower than the headline suggests.
Health & Safety
The Harm We Count vs. The Harm That Exists
If ESA reveals the wage theft funnel, Ontario’s occupational health and safety data reveals a different structural feature: the distinction between experienced harm and reportable harm.
Under the Occupational Health and Safety Act (OHSA), employers are legally required to report certain events to the Ministry of Labour, including critical injuries and fatalities. These reports form the basis of publicly released harm statistics.
In 2021, Ontario recorded 2,387 critical injury events reported under OHSA.
That number reflects severe incidents that crossed a statutory reporting threshold. It is not a claim count. It is not a compensation count. It is a count of events that met the legal definition of “critical injury” and were reported accordingly.
But not all workplace injuries meet that definition.
The OHSA defines “critical injury” narrowly. Many serious injuries — fractures not meeting the statutory test, psychological injuries, chronic conditions, gradual-onset harm — may not trigger mandatory reporting in the same way. Reporting obligations are event-based and threshold-driven. This distinction matters.
The publicly reported number reflects harm that:
was recognized,
met statutory criteria,
and was formally reported.
It does not capture:
injuries that did not meet the reporting definition,
injuries that went unreported,
gradual or cumulative harm,
psychological harm that did not trigger acute reporting requirements,
workers who continued working despite injury.
Recognition is not synonymous with experience. The health and safety dashboard shows reportable harm events. It does not show all harm occurring in workplaces. It shows what crossed a legal line.
Visibility, in this system, is threshold-based. Harm becomes visible when it meets definitional criteria and triggers a reporting obligation. Everything below that line remains statistically silent.
When placed alongside ESA investigations and OLRB applications, the pattern repeats:
The visible number reflects events that entered a statutory stream. It does not represent the full volume of underlying harm. Across systems, recognition requires threshold crossing, and threshold crossing is not the same as harm occurrence.
OLRB
Reverse-Mathing the Labour Relations Funnel
The Ontario Labour Relations Board occupies a central position in Ontario’s workplace governance architecture. It adjudicates labour practice disputes, duty of fair representation complaints, Employment Standards appeals, and Occupational Health and Safety reprisal claims. In structural terms, it is the forum where workplace conflict crosses from internal dispute into formal statutory adjudication.
At face value, the Board’s annual application volumes suggest active oversight and meaningful access to remedy. Thousands of matters enter its jurisdiction each year. Files are scheduled, mediated, and disposed of. As with ESA investigations and OHSA reporting thresholds, the critical question is not how many applications are filed. It is how many survive the funnel to become visible adjudicative outcomes.
The Intake Layer (2024–25)
In 2024–25, the Board received 3,135 new applications. It also carried 1,642 open cases from previous years and reopened 100 files, for a total of 4,877 files processed during the year.
The Board reports that 3,024 cases were disposed of (with/without a hearing) and 1,853 were carried into 2025–26, leaving 1,577 active files pending.
Those numbers are already a funnel:
total processed files (stock + flow)
files disposed of (throughput)
files carried forward (backlog/continuity)
The deeper funnel appears when we reverse the Board’s own “resolved without final hearing” metric.
The Adjudication Visibility Layer (2024–25)
The OLRB reports that 90.2% of all cases were resolved without a final hearing (with breakdowns: LRA 90.6%; ESA 87.9%; OHSA appeals 86.3%; OHSA reprisal 91.8%).
Now reverse the implication:
Resolved without a final hearing does not mean “no work happened.” It means the file concluded through mechanisms like mediation, withdrawal, settlement, abandonment, screening, or other non-final-hearing routes — outcomes that do not generate the same level of visible jurisprudence.
Reverse-math (based on the Board’s own 2024–25 dispositions):
Total cases disposed of: 3,024
Share resolved without final hearing: 90.2%
Estimated share reaching a final hearing: 9.8% (100% − 90.2%)
So the visible adjudication layer is structurally narrow:
Final-hearing share of closed files: ~9.8%
Estimated final-hearing count: 3,024 × 9.8% ≈ 296 cases (approximate, using the Board’s published ratio)
The point is not the exact integer. The point is the structure:
Most files do not become final-hearing outcomes. Which means the publicly “legible” part of the system (formal decisions with reasons) represents only a fraction of intake and closure volume.
The Scheduling Reality Layer (2024–25)
The Board also reports the difference between planned adjudication activity and adjudication that actually occurs.
In 2024–25:
4,592 hearings were scheduled (involving 6,346 files)
1,525 hearings were actually held (involving 2,230 files)
Reverse the ratio:
Hearings held vs scheduled: 1,525 ÷ 4,592 ≈ 33.2%
Files heard vs files scheduled: 2,230 ÷ 6,346 ≈ 35.1%
Scheduling is not adjudication. The difference between “scheduled” and “held” is another point where attrition, settlement, withdrawal, adjournment, and procedural re-routing shape the visible outcome.
The “Second-Order” Funnel: DFR
(Here Is Where It Gets Brutal)
Duty of Fair Representation (DFR) applications are structurally important because they often arise after a primary dispute has already been filtered through multiple layers.
Workplace harm → grievance → union decision → worker challenges union → OLRB DFR application. So DFR is not a first-order harm pathway. It is a representation-quality pathway.
In 2023–24, the Board reports 248 DFR files received (plus re-openings), with 240 DFR files closed. Of those closed DFR files, 93.3% were resolved without a final hearing, and 16 proceeded to a final consultation.
Reverse the ratio:
Final consultation share: 16 ÷ 240 = 6.7%
Non-final-hearing share: 93.3%
Even in the stream specifically designed to evaluate whether representation conduct crossed a legal line, the portion that reaches formal adjudicative visibility is small.
Cross-Stream Confirmation (ESA Appeals + OHSA Reprisals at OLRB)
This is where the “Alphabet Soup” logic becomes undeniable: the same filtration pattern shows up inside the OLRB across different statutes.
ESA Appeals at the OLRB (2023–24)
The OLRB received 609 ESA cases in 2023–24. Of 595 closed, 88.9% were resolved without a hearing, and 66 went to a final hearing.
Reverse-math:
Final hearing share: 66 ÷ 595 = 11.1%
Resolved without hearing: 88.9%
OHSA Reprisals at the OLRB (2023–24)
The Board received 304 OHSA s.50 reprisal complaints in 2023–24. Of 278 unlawful reprisal cases closed, 94.6% were resolved without a final hearing.
Reverse-math:
Estimated final-hearing share: 5.4% (100 − 94.6)
This is the filtration pattern in its clearest form:
Reprisal is a “high stakes” allegation — and yet the adjudicative visibility layer remains narrow.
What the OLRB Numbers Actually Signal
OLRB statistics are accurate within mandate. Reverse-mathing shows the same structural pattern as ESA and OHSA reporting:
The public count reflects intake. The visible jurisprudence reflects adjudicated survival. The attrition between those points is not fully visible — and is largely non-narrated in headline reporting.
When the OLRB’s ratios are viewed in isolation, they reflect ordinary tribunal function. High settlement rates are not inherently problematic. Procedural screening is part of adjudicative design. Not every dispute requires a final hearing.
When the numbers are reverse-mathed, the structural pattern becomes clear:
For every 100 matters closed at the Board, roughly 90 resolve without a final hearing. For ESA appeals at the OLRB, approximately 89% resolve without a hearing. For OHSA reprisal matters, roughly 95% resolve without a final hearing. For DFR applications, over 93% resolve without proceeding to final consultation.
The visible jurisprudence — the decisions that shape precedent and inform public understanding — represents only a small portion of total intake. This does not mean the other matters lacked substance. It means they resolved within procedural channels that do not generate equivalent visibility.
Across ESA enforcement, OHSA reporting, and OLRB adjudication, the pattern holds:
The system counts what enters jurisdiction. It publishes what survives to formal outcome, it does not measure what dissipates between those stages.
The OLRB numbers are not an outlier. They are a confirmation. Oversight is jurisdiction-bound. Adjudication is filtered. Harm, however, is not segmented in lived experience. When placed alongside ESA and OHSA data, the Board’s ratios reinforce the same structural observation:
The visible output is always a filtered subset of experienced conflict. That pattern is not evidence of bad faith. It is evidence of design.
Structural Convergence Across Systems
When ESA, OHSA, and OLRB data are examined separately, each appears internally coherent. Investigations are processed. Harm events are reported. Applications are adjudicated or resolved. The systems function within their statutory mandates, and the published numbers reflect that activity.
However, when those numbers are reverse-mathed side by side, a structural pattern emerges. Across all three systems, entry is counted. Escalation narrows. Adjudication is filtered. Visibility depends on survival through procedural thresholds. The public sees investigations, reported events, applications filed, and decisions released. Those numbers are accurate within mandate. What is not visible is the attrition curve between harm experience and formal recognition.
Workers who never file are not counted. Claims withdrawn before adjudication are not visible in jurisprudence. Matters resolved through mediation or settlement do not generate the same public record as final hearings. Injuries that do not meet reporting definitions do not appear in official harm statistics. Disputes filtered out at screening stages disappear into administrative closure categories. This is not an allegation of institutional bad faith. It is an architectural observation.
Ontario’s oversight bodies are designed to process matters within defined jurisdictional boundaries. They are not designed to measure total harm prevalence across systems. They are not designed to track cumulative cross-forum impact. They are not designed to narrate attrition.
Across ESA enforcement, OHSA reporting, and OLRB adjudication, the pattern is consistent. Harm enters through multiple statutory gates, but only a fraction survives to formal visibility.
In the final article of this series, we move from analysis to architecture. If the filtration effect is structural, then solutions must also be structural. We will examine how oversight bodies could begin measuring attrition, designing cross-system continuity, and reducing visibility gaps without exceeding jurisdictional mandates.
The problem is measurable. That means it is solvable.
Data Sources and Methodological Note
All numerical data referenced in this article are drawn from publicly available annual reports and statistical publications issued by the Province of Ontario and its oversight bodies.
Employment Standards Act (ESA) investigation, inspection, and prosecution figures for 2024–25 are derived from Ministry of Labour, Immigration, Training and Skills Development public reporting.
Occupational Health and Safety Act (OHSA) critical injury reporting figures are drawn from Ministry of Labour public statistics for the 2021 reporting year.
Ontario Labour Relations Board (OLRB) intake, disposition, hearing, and stream-specific data (including LRA, ESA appeals, OHSA appeals and reprisal applications, and Duty of Fair Representation matters) are drawn from the OLRB Annual Reports for 2023–24 and 2024–25.
All ratio calculations, reverse-math percentages, and structural interpretations are the author’s own calculations based on the reported figures contained within those public documents.
This analysis does not allege institutional misconduct or bad faith. It examines publicly reported data to assess structural filtration effects across statutory systems.
Conclusions reflect interpretive analysis of published intake and disposition statistics and are presented for policy discussion purposes.
Readers are encouraged to consult the original annual reports and statistical publications for full contextual detail.
Comments