Multi-Institutional Research Collaborations

A multi-institutional research collaboration is not just a larger version of a single-institution project. It requires a distinct set of administrative decisions — which legal instrument moves the money, which institution the funder holds accountable, how authority is split between principal investigators at different sites, and how contribution gets credited when the people involved don’t share an employer — that a single-PI, single-site award never has to resolve. This guide covers those mechanisms specifically: the agreement types that structure a collaboration, the difference between a subaward and a consortium agreement, lead-institution versus distributed multiple-PI (MPI) governance, and how authorship and credit allocation work when contributors sit at different institutions. It does not cover collaboration strategy, team science soft skills, or how to find collaborators — for the agreement that governs how collaborating institutions actually exchange research data once a project is underway, see CASRAI’s guide to data sharing agreements between collaborators and institutions.

What “multi-institutional” changes about research administration

Adding a second institution to a project introduces at least four separate administrative questions that a single-site award doesn’t have to answer: which organization is legally and financially accountable to the funder; what instrument governs the money (and the compliance obligations that travel with it) moving to the other institution; how scientific and administrative authority is divided between principal investigators who don’t report to the same institution; and how authorship, credit, and intellectual property get allocated across people who may be subject to different institutional policies on all three. Sponsored programs offices, not investigators, are usually the ones negotiating the first three; the fourth is jointly the investigators’ and, increasingly, the institutions’ concern, since funders and journals both now expect a documented answer rather than an assumed one.

How the money and accountability actually flow: three real models

Funders differ meaningfully in how they structure a multi-institutional award, and the terminology isn’t consistent across them — the same phrase, “consortium agreement,” describes two structurally different instruments depending on which funder is involved. Research administrators need to know which model a given funder uses before they can negotiate the right document.

Model A — lead institution, subaward or “consortium agreement” (NIH’s default)

Under NIH’s model, a single institution — the direct recipient — is accountable to NIH for the entire project: performance, expenditure of funds by every participant, and all reporting obligations. Collaborating institutions receive a subaward, which NIH’s own Grants Policy Statement calls a consortium agreement — a formalized arrangement in which the recipient performs a substantive role, not merely a conduit of funds. Per NIH GPS §15.2.1, the agreement must identify the consortium’s lead investigator, set out procedures for directing and monitoring that institution’s research effort, and specify how it will be reimbursed. This is the same instrument CASRAI’s Consortium Agreement and Subaward dictionary entries both describe from different angles — NIH’s usage functionally is a subaward, executed under the “consortium agreement” name.

Model B — separately submitted, separately funded awards (NSF’s alternative)

NSF allows collaborating institutions two real options, not one. A collaborative proposal can be submitted as a single package with the lead institution administering subawards to the others — functionally identical to Model A. Or it can be submitted as separately submitted collaborative proposals: each organization files its own linked proposal, reviewed together as one package, and — if funded — each organization receives its own direct award number from NSF rather than a subaward from a lead institution. Per NSF’s Proposal & Award Policies and Procedures Guide (PAPPG), Chapter II, this is a genuinely distributed funding structure, not just a distributed governance structure — no institution is a subrecipient of another, and each is independently accountable to NSF for its own portion.

Model C — direct multi-beneficiary award, governed internally by a consortium agreement (Horizon Europe)

Under Horizon Europe, the funder (the European Commission or an executing agency) signs a single Grant Agreement directly with every participating organization — each is a named beneficiary, paid directly, with no institution acting as a pass-through for another. The internal relationships between beneficiaries — governance bodies, intellectual property categories (background, foreground, sideground), access rights, internal payment schedules, default and dispute-resolution procedures — are set out separately in a Consortium Agreement, commonly built on the DESCA model template, which CASRAI’s dictionary entry for consortium agreement covers in detail. This is the opposite structure from NIH’s use of the same term: the Horizon Europe consortium agreement governs relationships between co-equal direct grantees, not a funding flow-down from one institution to another.

The practical takeaway for a research administrator negotiating a multi-institutional award: don’t assume “consortium agreement” means the same thing across two funders’ documents. Confirm, for each funder, whether money flows through a lead institution (Models A and B’s subaward path) or directly to every participant (Models B’s separately-submitted path and Model C) before drafting or reviewing the agreement.

Inter-institutional agreements (IIAs): the layer that isn’t about funding

Not every multi-institutional collaboration needs a subaward or a consortium agreement — a project can involve two institutions’ researchers without any money crossing between the institutions at all, and still need a formal document. This is where the term inter-institutional agreement (IIA) is used in practice, covering at least three distinct situations:

Non-funded collaboration. When researchers at different institutions collaborate on work related to a funded project but with no direct funding exchanged, sponsored programs offices commonly use a memorandum of understanding or IIA to establish roles, responsibilities, and expectations that an award agreement doesn’t cover.
Joint intellectual property ownership. When an invention, dataset, or other IP-protectable output is co-created by researchers at two or more institutions, an IIA is the instrument technology transfer offices typically use to govern joint ownership, licensing rights, and revenue sharing — distinct from, and usually negotiated by, the technology transfer office rather than the sponsored programs office. See CASRAI’s guide to material transfer agreements for the related instrument used when physical research materials, rather than intangible IP rights, move between institutions.
Single-IRB and multi-site human-subjects research. Where NIH’s single-IRB (sIRB) policy applies, participating sites still need a documented way to allocate local review responsibilities, reporting relationships, and the costs of sIRB review among themselves — NIH’s own guidance on the policy (NOT-OD-16-109) discusses inter-institutional agreements specifically in this context, distinct from the funding agreement itself.

Which office negotiates an IIA depends on which of these three it is — typically sponsored programs for the first and third, technology transfer for the second — so confirm ownership internally before treating “IIA” as a single, single-office instrument. For how these different research-administration functions typically divide inside an institution, see CASRAI’s guides to pre-award vs. post-award office roles and departmental vs. central sponsored programs offices.

Subrecipient monitoring once the award is active

Once a lead institution issues a subaward under Model A (or the subaward path of Model B), federal Uniform Guidance imposes two separate obligations that research administrators need to track as distinct steps. First, before an award is even structured as a subaward, the pass-through entity has to make — and document — a genuine determination that the relationship is a subrecipient relationship rather than a contractor (vendor) relationship, per 2 CFR §200.331: does the other party have programmatic decision-making responsibility, is its performance measured against the objectives of the federal program, does it determine eligibility under the program — these are the substantive factors, not the label on the document. Getting this determination wrong at the outset misapplies the entire compliance framework that follows.

Second, once the subaward is issued, 2 CFR §200.332 requires the pass-through entity to evaluate each subrecipient’s risk of noncompliance, include a defined set of information in the subaward document itself (subrecipient identity, Federal Award Identification Number, performance period, obligated amount, indirect cost rate, and more), and then actively monitor the relationship for the life of the award — reviewing financial and performance reports, following up on audit findings, and issuing management decisions where deficiencies are found. This is a genuinely different obligation from the upfront determination in §200.331, and it’s the piece CASRAI’s subaward dictionary entry cross-references as subrecipient monitoring. For the broader Uniform Guidance framework these sections sit inside, see CASRAI’s guide to Uniform Guidance (2 CFR 200).

Lead institution vs. distributed PI: two different axes, not one

It’s easy to conflate “which institution is financially accountable” with “who scientifically leads the project,” but they’re separate questions that a well-structured collaboration answers independently.

Financial/legal structure is set by the funding model above: under NIH’s Model A, exactly one institution is the direct recipient and financially accountable to the funder, regardless of how scientific leadership is organized. Scientific/administrative leadership is a separate axis, and NIH’s Multiple Principal Investigator (MPI) policy — established in 2007 (NOT-OD-07-017) — is the clearest example of why the two don’t have to match. The MPI model lets an application name more than one PI as co-equally responsible and accountable for the project, including PIs at different institutions from the lead recipient. Applications using this model must include a Multiple PD/PI Leadership Plan describing the governance structure, communication plan, decision-making process for scientific direction, conflict-resolution procedure, and — if relevant — how budget and resources are distributed among the PIs.

The practical result: a project can have a single lead institution handling all financial accountability to NIH and issuing subawards to every collaborating site, while genuinely co-equal Multiple PIs at different institutions jointly direct the science — that’s the normal NIH pattern, not an edge case. The funding instrument (subaward vs. distributed award) and the leadership structure (single PI vs. Multiple PI) are negotiated separately and don’t have to mirror each other. Distributed-PI governance without a distributed funding structure is common; conflating the two in a proposal’s own description of “who’s in charge” is a common source of confusion for reviewers and, later, for authorship-order disputes (see below).

Authorship and credit allocation across institutions

This is where the funding and governance decisions above intersect with CASRAI’s core subject matter — and it’s the piece a generic “how to collaborate” resource typically skips entirely, because it isn’t a funding-administration question at all. Being named a co-PI on a grant, or having your institution named on a subaward, doesn’t automatically determine authorship on any resulting publication; the two are evaluated separately, against separate criteria, and institutions should not assume one settles the other.

Authorship criteria don’t vary by institution. Under the ICMJE framework, each author — regardless of home institution — independently has to meet all four authorship criteria (substantial contribution; drafting or critical revision; final approval; accountability). A collaborating institution’s own internal authorship norms can’t substitute for this, and can’t be used to justify adding someone who doesn’t meet the criteria (CASRAI’s co-authorship entry covers how order conventions and equal-contribution notes work once authorship itself is established across institutions).
The corresponding author isn’t automatically at the lead institution. Nothing in ICMJE’s criteria ties the corresponding author role to whichever institution holds the prime award or subaward — see CASRAI’s guide to corresponding author responsibilities. Multi-institutional teams should decide this explicitly rather than default to “whoever’s at the lead site,” particularly since the corresponding author takes on real post-publication obligations (data requests, corrections, disclosures) that should sit with whoever is actually positioned to handle them.
CRediT statements make cross-institutional contribution visible in a way a flat byline can’t. A byline order alone doesn’t communicate who at which institution actually ran the analysis versus who supervised it versus who provided the samples. A CRediT contribution statement (ANSI/NISO Z39.104-2022) records role-by-role what each person did, independent of institutional affiliation or funding relationship — useful precisely in multi-institutional teams where contribution type correlates more with role (e.g., the subrecipient site ran Formal Analysis; the lead site handled Funding Acquisition and Project Administration) than with seniority.
Author-order disputes across institutions have an institutional-politics dimension that single-site disputes don’t. A common friction point is an institution expecting its PI to be listed first or last for internal promotion or reporting purposes, independent of actual contribution. COPE’s authorship-dispute guidance applies the same way regardless of how many institutions are involved — order should reflect contribution and discipline convention, not funder or institutional accounting — see CASRAI’s guide to resolving authorship order disputes.
Affiliation listing should reflect where the work was actually done, not the awardee institution by default. Each author’s affiliation is listed independently at the time of contribution; a subrecipient institution’s staff list that institution as their affiliation even though the lead institution holds the prime award (see CASRAI’s affiliation entry for how this is modeled in a CRIS).
Non-author contributors still need to be credited — just not as authors. Core-facility staff, project coordinators, or statisticians at a partner institution who don’t meet full authorship criteria should be named in an acknowledgments section or funding statement, per ICMJE’s separate recommendation on this point, rather than either being silently omitted or added to the byline as a courtesy (a pattern COPE identifies as gift authorship).

The practical fix is timing: agree on the authorship and credit-allocation approach across all participating institutions early in the project — ideally at the same point the subaward or consortium agreement is negotiated — rather than leaving it to be worked out during manuscript drafting, when institutional and career-stage pressure on order and inclusion is highest.

Setting up a new multi-institutional collaboration: a practical checklist

Confirm the subrecipient-vs-contractor determination for each collaborating institution before choosing an agreement type (2 CFR §200.331, if federal funding is involved).
Identify which funding model the sponsor actually uses — lead-institution subaward, separately-submitted distributed award, or direct multi-beneficiary award — rather than assuming; the correct agreement template follows from this.
Draft and execute the funding instrument (subaward, NIH-style consortium agreement, or funder-issued award to each institution) with the terms that flow down from the prime award, not a generic vendor contract.
Separately determine whether a non-funding inter-institutional agreement is needed — for joint IP, for non-funded related work, or for single-IRB reliance allocation — and route it to the correct office (sponsored programs vs. technology transfer).
Decide and document the scientific leadership model (single PI vs. Multiple PI) explicitly, including a leadership/governance plan if using an MPI structure, independent of how the funding instrument is structured.
Set a subrecipient monitoring plan (risk-based, per 2 CFR §200.332) once any subaward is active — don’t treat this as a one-time step at award setup.
Agree on the cross-institutional authorship and CRediT approach at project start, not at manuscript submission — including who will serve as corresponding author and how non-author contributors will be acknowledged.

Frequently asked questions

What’s the difference between a subaward and a consortium agreement?

It depends on the funder. Under NIH, they’re effectively the same instrument — NIH’s Grants Policy Statement uses “consortium agreement” for what is functionally a subaward. Under Horizon Europe, a consortium agreement is a completely different, non-funding-flow instrument that governs relationships between co-equal direct grantees who are each paid directly by the funder. Confirm which model applies before assuming the terms are interchangeable.

Does NIH ever fund a collaborating institution directly, without a subaward?

No — under NIH’s standard model, funds go to the single direct recipient institution, which then issues consortium agreements/subawards to collaborating sites, even under a Multiple PI structure where a co-PI is based elsewhere. This is a real difference from NSF’s separately-submitted collaborative proposal option, where each institution can receive its own direct award.

Who negotiates an inter-institutional agreement — sponsored programs or technology transfer?

It depends on what the IIA is for. Agreements governing non-funded collaboration or single-IRB reliance allocation typically sit with sponsored programs; agreements governing joint intellectual property ownership typically sit with the technology transfer office. Confirm ownership internally rather than assuming.

Does a subaward automatically make someone a co-author on resulting papers?

No. Authorship is determined independently, against the same criteria (ICMJE’s four-part test) regardless of funding relationship. An institution receiving a subaward, or a PI named on a Multiple PI leadership plan, still has to independently meet authorship criteria on any given output — the two determinations are related in practice but not automatically linked.

Does a CRediT statement replace the need to agree on authorship across institutions?

No. CRediT records what each person did; it doesn’t determine who qualifies as an author in the first place, and it doesn’t set author order. A CRediT statement is a useful complement to an authorship agreement across institutions, not a substitute for deciding authorship and order.