Choosing an employer of record: infrastructure models, liability chains, and a provider selection framework for cross-border hiring

Abstract

A previous paper (Schuchmann, 2026) proposed a taxonomy distinguishing four global employment intermediary models: Employer of Record (EOR), Professional Employer Organization (PEO), Global Payroll, and Contractor of Record (COR). That taxonomy answers the between-category question: which model applies to a given hiring scenario. This paper addresses the within-category question that follows: once EOR is the appropriate model, how should a company choose between providers whose legal structures, and therefore risk profiles, differ in ways their marketing does not reveal?

The central problem is that the most consequential structural distinction in the EOR market, whether the provider employs workers through its own local entities or through a network of in-country partners, is discussed almost exclusively in vendor-authored content, where each provider argues for the superiority of whichever model it happens to operate. No neutral framework exists. This paper proposes one. It formalises three infrastructure models (owned-entity, partner-network, and hybrid), introduces the liability chain as the analytical device that reveals what the infrastructure question actually determines, and argues that the owned-versus-partner binary is a proxy debate: the operative question is not which label a provider carries globally, but who the legal employer is in each specific country and how many contractual steps separate that entity from the client. A six-dimension selection framework and a sequenced decision protocol are offered for practitioners.

This is a conceptual paper. It evaluates structures, not vendors, and makes no empirical claims about the performance of either infrastructure model. Independent empirical evidence on comparative compliance outcomes does not yet exist and is identified as a priority for future research.

Keywords: employer of record, provider selection, vendor selection, owned entity, in-country partner, aggregator, global employment, international HRM, HR outsourcing, liability, cross-border hiring, data protection

JEL classification: M16 (International business administration), M54 (Labor management), J41 (Labor contracts), L24 (Contracting out)

Defined Terms

Consistent with the approach of the preceding taxonomy paper, the definitions below establish this paper's terminological basis. They do not claim universal legal status; they establish the consistent usage applied throughout. Where a definition departs from common vendor usage, that is noted.

Employer of Record (EOR): as defined in Schuchmann (2026): a third party that assumes full legal employer status under local law in the worker's jurisdiction, while the client retains operational direction. This paper concerns selection within this category only.

Owned-entity model: an EOR infrastructure model in which the provider (or a wholly-owned subsidiary within its corporate group) is itself the legal employer in the worker's country. Sometimes marketed as "wholly-owned infrastructure" or "direct" EOR.

Partner-network model: an EOR infrastructure model in which the provider contracts with independent local firms, each of which is the legal employer in its own country. Commonly called the "aggregator" model in vendor literature. The local firms are referred to as in-country partners.

In-country partner (ICP): an independent local company that acts as the legal employer of record in its jurisdiction under a commercial agreement with a global EOR provider. The ICP is not owned by, and is a separate legal person from, the provider the client contracts with.

Hybrid model: an infrastructure model combining owned entities in some countries with in-country partners in others. In practice, many providers marketed under a single-model label operate hybrid infrastructure. This paper treats hybrid as the empirically common case rather than an edge case.

Liability chain: introduced in this paper: the sequence of legal persons and contracts standing between the client and the entity that holds employer obligations for a specific worker in a specific country. The chain's length (number of contractual steps) and composition (ownership versus arm's-length contract at each step) determine how employer obligations, indemnities, and recourse actually flow.

Sub-processor: under Regulation (EU) 2016/679 (GDPR), Article 28: a processor engaged by another processor to carry out specific processing on behalf of a controller. In partner-network arrangements, ICPs typically process employee personal data as sub-processors, with consequences for transparency and cross-border transfer compliance.

Offboarding portability: introduced in this paper: the practical and contractual ease with which a client can move workers off a given EOR arrangement, whether to another provider, to the client's own newly established entity, or out of employment, including notice obligations, transfer mechanics, data return, and any exit fees.

Total cost of engagement (TCE): the full cost of an EOR arrangement over its expected life, including the headline fee (flat or percentage of salary), statutory employer costs, benefits administration margins, foreign-exchange spreads, security deposits, onboarding and offboarding charges, and charges contingent on salary changes.

1. Introduction

Once a company has established that the Employer of Record model is the right construct for a given hire, an employee, in a country where the company has no entity and is not planning one (Schuchmann, 2026), a second decision immediately follows: which provider. That decision is harder than it looks, because the EOR market conceals a structural distinction that materially changes what the client is actually buying.

Some providers employ workers through legal entities they own. Others contract with independent local firms, in-country partners, each of which is the legal employer in its own jurisdiction. Many do both, owning entities in some markets and using partners in others, while marketing themselves under a single label. From the client's side of the platform, these arrangements can look identical: same dashboard, same onboarding flow, same invoice.

Legally, they are not identical. The identity of the employing entity, its relationship to the provider the client signed with, and the number of contracts standing between the client and that entity all change where obligations sit, how disputes resolve, and what happens when something in the chain fails.

The existing discussion of this distinction has a specific defect: it is written almost entirely by the vendors themselves. Providers operating owned-entity infrastructure publish content arguing that partner networks produce inconsistent service and diluted accountability. Providers operating partner networks publish content arguing that owned-entity providers lack genuine local expertise and cannot match partner-network coverage. Both sets of claims are made confidently, both are self-serving, and neither is supported by independent evidence, because no independent evidence exists. There is no peer-reviewed study comparing compliance outcomes, service quality, or worker experience across EOR infrastructure models. A company researching the question will find a literature in which every source's conclusion can be predicted from its author's business model.

This paper does not resolve that empirical question. It cannot; the data has not been gathered. What it does instead is reframe the question so that practitioners can make structurally informed decisions without relying on vendor claims. The reframing has three parts. First, the owned-versus-partner binary is replaced with a three-model classification (owned-entity, partner-network, hybrid) that reflects how the market actually operates. Second, the infrastructure label is demoted from answer to proxy: what a client actually needs to establish is the liability chain for each specific country it intends to hire in, and this paper introduces that construct and shows how to trace it. Third, infrastructure is placed as one of six selection dimensions alongside coverage match, pricing structure, data-flow architecture, offboarding portability, and provider viability, dimensions that jointly determine the quality of an EOR engagement but that no existing framework assembles in one place. The paper deliberately evaluates structures rather than vendors. No provider is named, ranked, or recommended. The market research context is the same as in the preceding paper: one research firm estimates the EOR market at USD 4.42 billion in 2023, projected to reach USD 8.59 billion by 2030, though forecasts vary considerably across research providers (Verified Market Research, 2024). A market of that size, growing at that rate, built on structures that clients demonstrably do not understand, needs a neutral selection framework. This paper attempts to provide one.

Note: nothing in this paper constitutes legal advice. How EOR arrangements and the liability of the parties are characterised legally varies by jurisdiction. Anyone making decisions in this space should work with local legal counsel.

2. Literature Review

2.1 From taxonomy to selection

The preceding paper in this line of work (Schuchmann, 2026) established EOR, PEO, Global Payroll, and COR as distinct constructs separated by four dimensions: legal employer status, entity requirement, jurisdictional portability, and co-employment structure. Dahan (2025) provides the most substantial legal treatment of the EOR model itself, mapping the inconsistent patchwork of leasing laws, PEO regulation, and service intermediation rules across more than a dozen jurisdictions. Both works stop at the boundary of the present question. The taxonomy tells a company that EOR is the right category; Dahan explains how the category is regulated; neither addresses how providers within the category differ structurally or how a company should choose among them.

2.2 Vendor selection in the outsourcing literature

The outsourcing literature offers relevant general machinery. Transaction cost economics explains why intermediaries exist at all: firms contract out when the cost of performing a function internally, here, establishing and running a foreign employing entity, exceeds the cost of procuring it in the market, and it predicts that contracting hazards rise with asset specificity and the difficulty of monitoring performance (Williamson, 1985). Employment is a high-hazard case on both counts: the "asset" is a legal relationship with a human being, and the client cannot easily observe whether the provider is actually compliant in a jurisdiction the client, by definition, does not know well.

The information technology outsourcing literature, the most developed empirical body of work on outsourcing decisions, consistently finds that sourcing outcomes depend less on the decision to outsource than on how the arrangement is configured and governed: contract design, provider capabilities, and relationship governance recur as the determinants of success (Lacity, Khan and Willcocks, 2009). Barthélemy (2003) catalogues recurring failure modes, notably selecting a vendor on price and headline capability while neglecting the vendor's underlying delivery structure, that map closely onto the EOR selection errors this paper addresses. From the procurement side, Kraljic (1983) argues that purchasing approaches should vary with supply risk and business impact; cross-border employment is high on both axes, which argues against treating EOR selection as a commodity purchase decided on subscription price.

The HR outsourcing literature adds the domain specifics. Cooke, Shen and McBride (2005) distinguish transactional from strategic HR outsourcing and note that the risks of outsourcing rise as the outsourced function moves closer to the employment relationship itself; the EOR model is the limiting case, in which the employment relationship is not supported by the provider but held by it. Brewster, Sparrow and Vernon (2007) document the tension between global consistency and local compliance that multinational HR must manage, the same tension that the owned-entity and partner-network models resolve in opposite directions: the owned-entity model prioritises consistency of process, the partner-network model prioritises depth of local practice, and neither escapes the trade-off.

2.3 What the literature does not cover

None of this literature addresses EOR infrastructure models. The owned-versus-partner question appears only in vendor marketing content and trade commentary, which supplies vocabulary ("aggregator," "wholly-owned," "in-country partner") but no neutral analysis. The ILO's work on triangular employment (ILO, 2006; ILO, 2016) is relevant in one under-appreciated way: a partner-network arrangement is not a triangle but a quadrilateral. The worker is employed by an ICP, directed by the client, and administered through a platform provider that is party to contracts with both but employer to neither. Each additional party in the structure is a place where responsibilities can be ambiguously divided, which is precisely the hazard the ILO identifies in multi-party employment structures, and Davidov (2004) shows that when employer functions fragment across parties, workers can end up with less protection than any party intended. The length of the chain is therefore not a commercial detail. It is the structural fact a selection framework should be built around.

3. Infrastructure Models

3.1 Methodology

As in the preceding paper, the approach is a conceptual classification in the tradition of Doty and Glick (1994). The three infrastructure models were identified from provider disclosures, contractual documentation, and over a decade of direct observation of the global employment solutions market, including the evaluation of dozens of EOR providers. The classification was refined against the same criteria used previously: each model must occupy a distinct position on the dimensions that matter (identity of the legal employer, chain length, coverage economics), and no model should be redundant. Whether the three-model classification captures practitioner reality across jurisdictions is an empirical question deferred to future research.

Terminological note. The terms "aggregator" and "wholly-owned" circulate widely in vendor content; "in-country partner" is standard industry usage. What this paper contributes is not the vocabulary but the formalisation: explicit definitions, the hybrid model treated as a first-class category rather than a footnote, the liability chain as the analytical device connecting infrastructure to risk, and the demotion of the global infrastructure label in favour of per-country analysis. These contributions are the author's own formulation.

3.2 The owned-entity model

In the owned-entity model, the legal employer of the worker is an entity within the provider's own corporate group: either the contracting company itself or a wholly-owned local subsidiary. The client signs one commercial agreement with the provider, and the employment relationship sits inside the provider's group.

The structural properties follow from ownership. Because the employing entity and the platform provider are under common control, process standards, service levels, and remediation can be imposed by management decision rather than negotiated by contract. The liability chain is short: client to provider group, provider group to worker. When something goes wrong, there is no third party to whom responsibility can be deflected; the group that took the fee also holds the employment.

The structural costs are equally direct. Entities are expensive to establish and maintain, which constrains coverage: an owned-entity provider's country list is limited by its willingness to carry fixed local infrastructure, and expanding to a new country is a capital decision, not a contracting decision. Ownership of an entity also does not by itself guarantee local operational depth. A subsidiary can be thinly staffed, administered regionally rather than locally, or recently established with limited accumulated practice in the jurisdiction. Ownership guarantees control; it does not guarantee expertise. Clients should assess the two separately.

3.3 The partner-network model

In the partner-network model, the provider contracts with an independent local firm, the in-country partner, in each country it serves. The ICP is the legal employer. The client's contract is with the global provider; the worker's employment contract is with the ICP; the provider and the ICP are connected by a commercial agreement to which the client is typically not a party.

The structural properties again follow from the arrangement. Coverage scales by contracting rather than by capital: adding a country means signing a partner, so partner-network providers can typically serve more countries, and enter unusual markets faster, than owned-entity providers of comparable size. The ICPs themselves are often long-established local employment firms with deep jurisdiction-specific practice, frequently deeper than a newly established foreign-owned subsidiary. A network provider can also, in principle, replace an underperforming partner, an option unavailable to a provider that owns its infrastructure. The costs are structural too. The liability chain is longer: client to provider, provider to ICP, ICP to worker, with the critical link, the provider-to-ICP agreement, invisible to the client. Employer obligations rest with a company the client did not choose, cannot directly instruct, and usually cannot audit. Service consistency across countries depends on the provider's partner management rather than on common ownership. Data flows lengthen correspondingly: employee personal data processed by the ICP is typically handled under sub-processing arrangements (Regulation (EU) 2016/679, Article 28), which are lawful but add parties to the processing chain, and cross-border transfers within the chain must satisfy the GDPR's transfer rules (Chapter V) where they apply. None of this makes the model defective; it makes the model different, in ways a client should price in rather than discover during a dispute.

3.4 The hybrid model

The hybrid model combines both: owned entities in some countries, partners in others. Providers typically own entities in their highest-volume markets, where fixed infrastructure is economic, and use partners in the long tail. Trade coverage of the market routinely describes named providers as operating hybrid infrastructure, and provider disclosures confirm that single-label marketing frequently overlays mixed reality.

The hybrid model's significance for this paper is less as a third option than as the demonstration of the central argument: the global label attached to a provider does not describe the structure of any particular engagement. A client hiring in two countries through one "owned-entity" provider may in fact be getting an owned entity in one country and an ICP in the other. The only reliable unit of analysis is the country. This is the basis for the per-country disclosure requirement in the selection framework of Section 4.

3.5 Comparison

Two observations follow. First, neither pure model dominates: each resolves the consistency-versus-coverage trade-off in one direction and pays for it in the other, which is why the vendor debate is undecidable on the vendors' own terms. Second, every structural difference in the table is mediated by one variable, who the legal employer is in the specific country at issue and what stands between that entity and the client. That variable is the liability chain, developed next.

4. The Liability Chain and Six Selection Dimensions

4.1 The liability chain

The liability chain is the sequence of legal persons and contracts standing between the client and the entity that holds employer obligations for a specific worker in a specific country. Tracing it requires answers to four questions, per country:

1. Which legal entity issues this worker's employment contract, and in which jurisdiction is it registered?
2. What is that entity's relationship to the provider the client contracts with: same legal person, wholly-owned subsidiary, or independent partner?
3. Which contracts connect the client to that entity, and is the client a party to, or a disclosed beneficiary of, each link?
4. Where in the chain do indemnities, insurance, and remediation obligations sit, and do they reach the entity that actually holds the employment?

The chain matters because employer obligations are enforced locally, against the local employer, while the client's recourse runs through its own contract, which may terminate one or two steps short of that employer. In a one-step chain, the party the client can sue and the party that holds the employment belong to the same group. In a two-step chain, the client's contractual counterparty and the legal employer are different companies, and the client's protection depends on back-to-back terms in an agreement it has usually never seen. The preceding taxonomy noted that even under EOR the client may retain specific statutory obligations, such as workplace safety and anti-discrimination duties, depending on jurisdiction (Schuchmann, 2026); the chain determines how everything that is transferable actually flows.

The chain also reframes the vendor debate. "Owned entity" is valuable insofar as it produces a short chain with common ownership at every link; "partner network" is acceptable insofar as the longer chain is disclosed, contractually reinforced, and priced. The label predicts the chain imperfectly, which is why the framework below asks for the chain, not the label.

4.2 The six dimensions

Infrastructure is one of six dimensions that jointly determine the quality of an EOR engagement. The other five are routinely underweighted because the infrastructure debate absorbs the attention.

Dimension 1: Infrastructure and liability chain (per country). As above. The unit of analysis is the country, not the provider. A provider unwilling to disclose, in writing, the identity of the employing entity for each country in scope has failed the dimension regardless of its marketing.

Dimension 2: Coverage match. Aggregate country counts are marketing statistics. What matters is depth in the specific countries the client will hire in: the maturity of the employing entity or partner there, local restrictions that constrain the model itself (for example, Germany's Arbeitnehmerüberlassungsgesetz limits the leasing of an individual worker to the same hirer to 18 consecutive months (AÜG, §1 Abs. 1b), a constraint that binds regardless of provider quality), and whether the provider's structure in that country has been tested at the client's intended scale. Schuchmann (2026) treats these jurisdiction-specific restrictions as EOR failure modes; at the selection stage they become screening questions.

Dimension 3: Pricing structure and total cost of engagement. EOR pricing is typically either a flat fee per worker per month or a percentage of salary. The structures allocate risk differently: percentage pricing couples the provider's fee to the client's compensation decisions, so every raise increases the intermediary's take; flat pricing decouples them. The headline fee is also a minority of variable cost in many engagements. A total-cost-of-engagement analysis should price foreign-exchange spreads on salary funding, security deposits and their release conditions, benefits administration margins, onboarding and offboarding charges, and any fees contingent on termination. Selecting on subscription price while ignoring delivery structure is the classic outsourcing error (Barthélemy, 2003), and the opacity of EOR pricing beyond the headline fee makes it easy to commit.

Dimension 4: Data-flow architecture. An EOR processes the most sensitive category of routine corporate data: identity documents, compensation, bank details, health-related benefits data. The client should be able to map where employee personal data is stored, which entities process it, and under which roles. Where the GDPR applies, processing by ICPs will generally occur under Article 28 sub-processing arrangements, and transfers outside the EEA must rest on a Chapter V mechanism (Regulation (EU) 2016/679). The practical selection question is whether the provider can produce a current sub-processor list and transfer-mechanism documentation for the countries in scope without treating the request as unusual.

Dimension 5: Offboarding portability. Every EOR engagement ends: the client establishes its own entity, switches providers, or exits the market. The preceding paper's practitioner rule of thumb, that EOR begins to look expensive relative to entity establishment at roughly 10 to 15 employees in one country (Schuchmann, 2026), implies that successful engagements graduate. Selection should therefore examine the exit before the entry: notice periods, cooperation obligations on transfer of workers to a successor employer, treatment of accrued entitlements, return and deletion of data, and any charges triggered by leaving. In partner-network structures the exit involves a third party (the ICP) whose cooperation the client cannot contractually compel; the provider's obligations to procure that cooperation should be explicit.

Dimension 6: Provider viability and concentration. The client's workers are employed by, or through, the provider; the provider's failure is therefore an employment event, not merely a service interruption. Financial durability, insurance coverage, and the concentration structure of the network matter: in a partner-network model, the failure of a single large ICP can affect workers across many of the provider's clients simultaneously. This dimension is standard in outsourcing governance (Lacity, Khan and Willcocks, 2009) and routinely absent from EOR marketing comparisons.

4.3 Dimension summary

5. Practitioner Decision Protocol

5.1 Sequence

The dimensions are not equally weighted, and they are not usefully assessed in parallel. The following sequence orders them so that each step can disqualify providers before effort is spent on the next.

1. Step 0: Confirm the construct. Verify that EOR, rather than PEO, global payroll, COR, or entity establishment, is the right model for the hire, using the taxonomy and decision framework in Schuchmann (2026). Provider selection cannot repair a category error.
2. Step 1: Per-country disclosure. For each country in scope, obtain in writing the identity of the employing entity and its relationship to the provider. This converts the marketing label into the actual structure and typically reclassifies several "owned-entity" and "partner-network" providers as hybrids.
3. Step 2: Trace the liability chain. For each country, apply the four questions in Section 4.1. Where the chain has two steps, request the substance of the back-to-back protections: indemnity flow-down, insurance at the ICP level, and the provider's obligations if the ICP fails or is replaced.
4. Step 3: Test coverage depth. Assess maturity and scale of the structure in each country, and screen for jurisdiction-specific restrictions on the EOR model itself. Local legal input belongs at this step, not after signature.
5. Step 4: Model total cost of engagement. Price the full engagement under realistic salary, FX, and duration assumptions, including exit. Compare flat versus percentage structures under the client's expected compensation trajectory.
6. Step 5: Map the data flow. Obtain the sub-processor list and transfer mechanisms for the countries in scope; verify they match the entities disclosed in Step 1.
7. Step 6: Negotiate the exit before the entry. Fix notice, transfer cooperation, data return, and exit charges in the initial agreement, when negotiating leverage is at its maximum.

5.2 Weighting by client profile

The sequence is constant; the weighting is not. Three stylised profiles illustrate.

• Single-country, small headcount (testing a market): coverage depth in the one country dominates. The infrastructure model matters less than the quality of the specific employing entity or partner in that country; a strong ICP can outperform a thin owned subsidiary. Exit terms matter because market tests end.
• Multi-country portfolio (distributed team): consistency and data architecture rise in weight. The client is exposed to the provider's weakest country, and in partner-network structures to its weakest partner. Per-country disclosure (Step 1) is the highest-value step and should be completed for the full expected footprint, not only the launch countries.
• Regulated or data-sensitive industry: the data-flow dimension can become disqualifying rather than merely comparative. Chain length translates directly into processing-chain length, and the client's own regulatory obligations may constrain which structures are acceptable irrespective of employment-law considerations.

In all profiles, worker classification precedes everything: if the worker is not an employee, EOR is the wrong construct and COR boundary conditions apply (Schuchmann, 2026). No selection framework rescues a misclassified relationship.

6. Limitations and Directions for Future Research

This is a conceptual paper, and the limitations are substantial.

First, and most importantly, the paper makes no claim that either infrastructure model produces better outcomes. That question is empirical, and the evidence does not exist. The vendor-authored literature asserts outcome differences in both directions; none of it meets basic standards of independence. Until comparative data on compliance incidents, dispute resolution, and worker experience across infrastructure models is gathered, all outcome claims in this market, including those implied by provider marketing, should be treated as untested.

Second, the liability chain construct is presented analytically, not doctrinally. How a two-step chain actually performs in a dispute depends on the law of the employing jurisdiction, the drafting of two contracts (only one of which the client sees), and facts. The construct tells a practitioner where to look; it does not predict what they will find. Dahan (2025) demonstrates how much the regulatory ground varies across jurisdictions, and that variation applies with full force here.

Third, the paper's observational base skews toward the same markets as its predecessor: the US, Europe, and globally marketed platforms. Infrastructure economics differ in markets where foreign ownership of employing entities is restricted or where local partnering is mandatory, and the three-model classification may need adaptation there.

Four research questions follow.

1. Do compliance and dispute outcomes differ systematically across infrastructure models? Administrative data, insurance claims data, or structured case collection would supply the first independent evidence in a debate currently conducted entirely by interested parties.
2. How accurate are providers' public infrastructure claims? A disclosure audit comparing marketed labels against per-country employing entities, feasible from public registries in many jurisdictions, would quantify the gap between label and structure that this paper asserts qualitatively.
3. How do workers experience the difference? Whether being employed by an ICP rather than a provider subsidiary changes workers' understanding of their employer, their access to remedies, or their treatment in practice connects this framework to the ILO's concerns about multi-party employment (ILO, 2006; ILO, 2016) and to the worker-experience gap flagged in the preceding paper.
4. Where does the EOR-to-entity graduation threshold actually sit? The 10-to-15-employee rule of thumb (Schuchmann, 2026) deserves empirical treatment across jurisdictions, since offboarding portability, Dimension 5, matters in proportion to how often graduation occurs.

7. Conclusion

The EOR market's most consequential structural distinction, who actually employs the worker, and what stands between that entity and the client, is currently explained to buyers almost exclusively by sellers, each of whom explains it in the shape of its own business model. This paper has proposed a neutral alternative: a three-model classification of EOR infrastructure (owned-entity, partner-network, hybrid), the liability chain as the device that converts infrastructure labels into analysable structure, and a six-dimension selection framework, sequenced into a decision protocol, that places infrastructure alongside the coverage, pricing, data, exit, and viability questions it usually crowds out.

The central practical claim is deflationary: the owned-versus-partner debate, as conducted, is not answerable, and does not need to be answered. A client does not need to know which model is better in general. It needs to know, for each country it will hire in, who the legal employer is, how many contracts stand between that employer and itself, and what flows across each link. Providers of either model can pass that examination; providers of either model can fail it.

Terminological contributions. Consistent with the practice of the preceding paper, the elements original to this work are noted explicitly: the formalisation of the three infrastructure models with the hybrid model as a first-class category; the liability chain construct and its four tracing questions; the six-dimension selection framework and its sequenced decision protocol; and the offboarding portability and total cost of engagement constructs as defined terms. The underlying vocabulary of "aggregator," "wholly-owned," and "in-country partner" is established industry usage and is not claimed.

References

Arbeitnehmerüberlassungsgesetz (AÜG) [German Temporary Employment Act], §1 Abs. 1b, as amended 1 April 2017. Bundesministerium der Justiz.

Barthélemy, J. (2003). The seven deadly sins of outsourcing. Academy of Management Executive, 17(2), 87–98.

Brewster, C., Sparrow, P., & Vernon, G. (2007). International human resource management (2nd ed.). Chartered Institute of Personnel and Development.

Cooke, F. L., Shen, J., & McBride, A. (2005). Outsourcing HR as a competitive strategy? A literature review and an assessment of implications. Human Resource Management, 44(4), 413–432.

Dahan, S. (2025). Global workforce compliance and the employer of record model. SSRN Working Paper.

Davidov, G. (2004). Joint employer status in triangular employment relationships. British Journal of Industrial Relations, 42(4), 727–746.

Doty, D. H., & Glick, W. H. (1994). Typologies as a unique form of theory building: Toward improved understanding and modelling. Academy of Management Review, 19(2), 230–251.

International Labour Organization (ILO). (2006). Employment relationship recommendation, No. 198. International Labour Conference, 95th Session, Geneva.

International Labour Organization (ILO). (2016). Non-standard employment around the world: Understanding challenges, shaping prospects. International Labour Office.

Kraljic, P. (1983). Purchasing must become supply management. Harvard Business Review, 61(5), 109–117.

Lacity, M. C., Khan, S. A., & Willcocks, L. P. (2009). A review of the IT outsourcing literature: Insights for practice. Journal of Strategic Information Systems, 18(3), 130–146.

Regulation (EU) 2016/679 of the European Parliament and of the Council of 27 April 2016 (General Data Protection Regulation), Articles 28 and 44–49. Official Journal of the European Union, L 119.

Schuchmann, R. (2026). Global employment solutions: a taxonomy of EOR, PEO, global payroll, and contractor of record (Preprint, v1). Zenodo. https://doi.org/10.5281/zenodo.18861073

Verified Market Research. (2024). Employer of record market size, scope, trends and forecast.

Williamson, O. E. (1985). The economic institutions of capitalism: Firms, markets, relational contracting. Free Press.

Author

Robbin Schuchmann is co-founder of Employ Borderless, an independent advisory and research platform in the global employment solutions sector, headquartered in Singapore. He has over a decade of direct experience advising companies on international employment and has personally evaluated dozens of EOR, PEO, global payroll, and COR providers.

Contact: [email protected] - Entity home: https://robbinschuchmann.com

AI-use disclosure: Drafts prepared with assistance from AI writing tools. All subject matter expertise, analytical reasoning, taxonomic claims, and practitioner judgements are the author's own.