Consolidated Out of Existence: What University Mergers Are Doing to America's Research Data

When American universities merge, restructure, or shutter departments, the administrative consequences are visible and well-documented. Logos change. Letterheads are reprinted. Faculty offices are reassigned. The scientific consequences — entire research data collections quietly deleted, migrated into incompatibility, or simply abandoned — rarely make the press release.

Over the past fifteen years, the United States has witnessed a sustained wave of higher education consolidation. The University System of Georgia merged eight institutions between 2012 and 2018. Wisconsin, Louisiana, and North Dakota have each restructured significant portions of their public university systems under fiscal pressure. Private institutions have not been immune: small liberal arts colleges have closed at a rate that would have seemed extraordinary a generation ago. In nearly every instance, the fate of institutional research data repositories has been addressed last — if at all.

What Gets Lost in the Transition

The losses are not always dramatic. There is rarely a single moment when a dataset ceases to exist. More commonly, a server is decommissioned without a migration plan. A data librarian's position is eliminated as a cost-saving measure, and the institutional knowledge required to maintain a specialized repository evaporates with her departure. A storage contract lapses because the department responsible for renewing it no longer exists.

Agricultural research institutions are among the most acutely affected. Land-grant universities and their associated experiment stations have spent generations accumulating longitudinal datasets on soil composition, crop yield variation, pest population cycles, and regional climate interactions. These collections are not merely historical curiosities. They represent the observational baseline against which contemporary agricultural science measures change. When a rural extension program is folded into a larger university system and its servers are decommissioned to cut overhead, that baseline is severed.

Data librarians — a profession that has grown in institutional importance even as it has shrunk in institutional funding — describe a consistent pattern. Merger planning committees prioritize personnel, real estate, and accreditation continuity. Data infrastructure is treated as an IT problem to be solved later, which in practice often means never. "Later" arrives when the new institution's systems team discovers that migrating a legacy repository requires custom scripting, format conversion, and metadata reconciliation that no one in the current org chart is equipped to perform.

The Economics of Deletion

There is a straightforward financial logic that drives deletion over preservation. Maintaining a research data repository is not free. Server costs, software licensing, staff time, and security compliance represent recurring line items. A merged institution facing budget shortfalls has every structural incentive to treat legacy data storage as a liability rather than an asset.

Preservation, by contrast, requires upfront investment with diffuse and delayed returns. The researchers who would benefit from access to a thirty-year soil moisture dataset from a shuttered agricultural experiment station are not a constituency with lobbying power inside a provost's office. The researchers who generated that data have often retired or moved on. The institutional memory of why the collection matters is gone.

Legal frameworks offer limited protection. Federal data retention requirements apply to specific grant-funded projects and carry defined timelines — often five to ten years post-publication — after which deletion is not merely permitted but, from a compliance standpoint, arguably encouraged to limit liability exposure. State public records laws vary considerably and rarely contemplate scientific datasets as a distinct category deserving special treatment.

Documented Cases, Incomplete Counts

Precise national figures on research data lost to institutional consolidation do not exist, which is itself a significant part of the problem. There is no federal registry of decommissioned university data repositories. There is no systematic audit process. What survives in the literature are case reports: an agricultural economics dataset from a merged Midwestern extension program, cited in fifteen subsequent publications, that now returns a 404 error. A longitudinal study of irrigation efficiency conducted across three decades at a California state institution that was folded into the UC system and whose raw data was never migrated.

Data librarians who have attempted informal surveys report that the problem is widespread but nearly impossible to quantify because the losses are, by definition, undocumented. You cannot count what has been erased.

What Preservation Would Actually Require

Solutions exist, but they require deliberate policy choices that current incentive structures do not support. Pre-merger data audits — mandatory assessments of existing repositories conducted before consolidation is finalized — would at minimum create an inventory of what is at risk. Several research library associations have proposed model policies along these lines, with limited adoption.

Distributed preservation networks offer a more resilient architecture. Rather than relying on a single institutional server, datasets ingested into federated repositories such as those built on open-source platforms retain redundant copies across multiple nodes. If one institution decommissions its infrastructure, the data persists elsewhere. This approach requires inter-institutional agreements, standardized metadata schemas, and ongoing coordination — none of which are impossible, but all of which require sustained commitment.

Funding agencies could require, as a condition of grant renewal and institutional eligibility, that recipient universities demonstrate active data preservation plans covering legacy collections — not merely the outputs of currently funded projects. This would shift the cost-benefit calculation meaningfully, making preservation a condition of future revenue rather than a discretionary expenditure.

The Accumulated Cost

Agriculture is a domain where the consequences of lost longitudinal data are not abstract. Decisions about crop insurance programs, drought resilience investments, and soil conservation policy depend on long-term observational records. When those records are destroyed because an experiment station's server room was repurposed for administrative offices, the scientific foundation for those decisions is quietly degraded.

The wave of institutional consolidation in American higher education is not over. Demographic pressures, state budget constraints, and endowment volatility will continue to drive mergers and closures for the foreseeable future. Each consolidation carries the same risk: that the research infrastructure built over decades will be treated as overhead rather than inheritance, and deleted accordingly.

The data that disappears in these transitions does not announce its own absence. It simply becomes unavailable — first to the researchers who might cite it, then to the policymakers who might rely on it, and eventually to everyone.