Across 878 permanent employees in six organizations, the perceived share of non-permanent workers on a team predicted almost nothing. Organizational trust: B = โ0.16, p = 0.18. Job insecurity: B = โ0.02, p = 0.85. Team commitment: B = โ0.29, p = 0.07. Three hypotheses, three nulls (Schaltegger et al., 2026). Meanwhile the variable most operations leaders never instrument โ whether employees believe the organization has their back โ moved all three, hard, at p < 0.001.
If you run a contractor ratio cap, you are running a control that this data says does not touch the outcome you bought it for.
The control you are running was never actually tested
The contractor ratio cap is one of the most common governance instruments in mid-market operations. It usually appears as a policy line: no team exceeds 20% or 30% non-permanent headcount. The stated rationale is almost always about the permanent staff, not the contractors โ the belief that a rising external share erodes the core team's trust, raises their fear of being next, and loosens their commitment to the team.
That belief is intuitive. It maps onto psychological contract theory: if the employer signals that roles are fungible, employees revise what they think they are owed. It is also, in the mid-market, largely untested. Most organizations set the threshold by instinct or by copying a peer, then never measure whether the number does anything.
The Schaltegger study is the first mixed-methods look at that specific question at scale โ and it was designed by people who expected the intuitive answer to hold.
Getting this wrong is not free. A ratio cap set to protect morale has a real operational price: it blocks capacity you could have used, it pushes hiring managers into workarounds and reclassification games, and it consumes governance attention that could be spent on a lever that works. Worse, it creates false assurance. A team sitting comfortably under the cap reads as safe, so nobody looks further โ even when the actual risk condition is present and unmeasured.
What the 878-employee study actually found
The research, published 4 August 2026 in Frontiers in Organizational Psychology by a team at FHNW Switzerland and Europa-Universitรคt Flensburg, ran in two parts across six Swiss organizations (Schaltegger et al., 2026).
Study 1 surveyed 878 permanent employees โ roughly 48% IT, 33% management, 19% nursing โ measuring the perceived proportion of non-permanent workers on their team against three outcomes: organizational trust, job insecurity, and affective team commitment. Study 2 took 22 of those participants into qualitative interviews.
Three nulls where the theory predicted three hits
Hypotheses H1, H2 and H3 all failed. The perceived proportion of non-permanent workers was not significantly related to trust, insecurity, or commitment. Commitment came closest (p = 0.07) and still did not clear the bar. In regression models running roughly N = 783โ788 after listwise deletion, the headcount mix was, statistically, close to inert.
The variable that did move everything
Perceived organizational support (POS) โ the employee's read on whether the organization values their contribution and cares about their wellbeing โ was directly related to all three outcomes: trust B = 0.78, job insecurity B = โ0.22, team commitment B = 0.45, all at p < 0.001. The trust model explained 48% of variance. That is a large effect in organizational research, and it dwarfs anything the ratio produced.
This is consistent with the broader evidence base. A meta-analysis of 558 studies covering more than 200,000 employees found POS strongly associated with lower burnout and turnover intention and higher commitment and satisfaction (Kurtessis et al., 2017). POS is not a soft variable. It is one of the best-evidenced predictors in the field.
Where the contractor ratio does bite โ and how small that effect is
One result partially survived. POS moderated the relationship between contractor share and organizational trust (B = 0.36, 95% CI [0.05, 0.67], p = 0.02). The simple slopes tell the operationally useful story: at low perceived support, a rising contractor share was negatively related to trust. At high perceived support, that relationship disappeared.
That is the finding worth carrying into a planning meeting. The ratio is not harmless in all conditions โ it is harmless when support is high, and corrosive when support is low. The mix is a conditional risk, not a standalone one.
I would be doing you a disservice not to price the caveats, because they are real:
- The interaction added ฮRยฒ = 0.004. Statistically significant, practically tiny.
- The authors ran a robustness check without control variables, and the trust interaction lost significance. Some findings were sensitive to model specification, and they say so plainly.
- The design is cross-sectional. This is association, not causation.
- Mean perceived contractor proportion was 0.15 (SD 0.20). This sample tells you little about teams running at 50% external.
- Null results are not proof of absence. They are failure to detect.
So the honest reading is not "the ratio never matters." It is: at the contractor shares most mid-market teams actually run, the ratio is a far weaker lever than the support climate you are not measuring.
The obvious counter-argument, and why it does not rescue the cap
There is strong evidence that heavy use of agency workers damages business performance. Eldor and Cappelli found that otherwise-identical workplaces making greater use of agency temps showed lower service quality, in a study published in the Academy of Management Journal (Eldor & Cappelli, 2021). If agency use hurts performance, is a cap not justified?
It is โ but for a different reason than the one on your policy page. Eldor and Cappelli's mechanism runs through operational integration: knowledge loss, coordination cost, weaker firm-specific capability. That is a capability argument. The cap on your policy page is justified by a morale argument โ protecting the core team's trust and commitment. The Schaltegger data says the morale argument is the weak one.
Two conclusions follow. If your cap exists to protect service quality and institutional knowledge, keep it and defend it on those grounds. If it exists to protect how your permanent staff feel, you are using a headcount instrument to manage a perception problem, and the instrument does not reach.
The distinction matters because the two rationales imply different thresholds, different owners, and different failure signals. A capability cap should be set per function, owned by the function lead, and reviewed when service metrics move. A morale cap โ if you kept one โ would need to be set per team, owned by HR, and reviewed when sentiment moves. Most mid-market policies I see conflate the two into a single company-wide number that answers neither question well.
What the interviews found: support or replaceability
Study 2's 22 interviews produced seven themes and 15 sub-themes (Schaltegger et al., 2026), and the pattern is the part I would put in front of a Head of Operations. Almost nothing was inherently positive or negative.
Workload split both ways โ 28 codings for workload reduction, 26 for additional workload. Competence split both ways. The same structural arrangement was read as relief by some and as burden by others.
What separated the readings was organizational conditions. The two highest-frequency codes in the entire dataset were proactive socialization behavior (42 codings across 19 of 22 interviews) and conducive work practices (41 codings across 20 interviews). On the negative side: restricted system access and slow information flow (18 codings), unfavorable organizational conditions (20), ostracism (12), unrest and uncertainty (14).
Read that list again. Every one of those is a management-controlled variable, and not one of them is a ratio. Whether a contingent hire reads as backup or as evidence that I am replaceable is decided by onboarding, system access, information flow, and how the arrangement is explained โ not by the percentage.
What to instrument instead
Concretely, for a 200-FTE operation planning contingent capacity:
- Measure POS before you scale external headcount, not after. A short validated POS scale in your existing engagement survey, segmented by team. You need the baseline before the blend changes, or you cannot attribute anything.
- Treat low-POS teams as the constrained ones. This is the study's one actionable interaction. If a team's support scores are weak, that is where a rising contractor share plausibly costs you trust. Sequence your contingent growth into high-support teams first.
- Fix access before headcount. Restricted system access and slow information flow appeared in 10 of 22 interviews. That is an IT provisioning problem masquerading as a culture problem, and it is cheap to fix.
- Explain the why. The signal your permanent staff decode is intent. A contingent hire framed as capacity relief for a specific overload reads differently than one that appears without explanation.
- Stop defending the cap on morale grounds. Keep it if it protects capability. Retire the morale rationale, or you will keep optimizing a number that does not move the outcome.
The decision for this quarter
The contractor ratio cap is doing something in your organization. It is probably not the thing you think.
Before you set next year's contingent headcount plan, run one measurement you almost certainly are not running: perceived organizational support, segmented by team. It takes a handful of survey items. It predicts trust, insecurity, and commitment at magnitudes the ratio cannot approach โ and it tells you which teams can absorb external capacity and which will read it as a threat.
You have been managing the denominator. The evidence says the answer is in a variable you have not been counting.
