Every Decision in This System Rests on Data Someone Had to Build
Every actor in the American retirement system — Congress, the Social Security Administration, academic researchers, actuaries, insurers, asset managers and recordkeepers, employers designing benefit plans, and individuals deciding when to claim Social Security or how much to save — depends on a body of data and research that had to be built, largely from scratch, starting in 1935. I have spent nearly fifty years inside that system, first as a researcher, then for thirty-seven years running EBRI, and now at the Consumer Policy Center. What follows is the history I have watched, and lived a large part of, followed by an account of how each sector actually uses what that history produced.
Origins: The Office I Would Later Rely On
The federal government’s retirement-data function began the moment it had to administer a retirement program. The Social Security Board was created when President Roosevelt signed the Social Security Act on August 14, 1935, and one of the Board’s five original service bureaus was the Bureau of Research and Statistics. That office — renamed the Division of Research and Statistics in 1963 and the Office of Research and Statistics in 1965 — is the single most important fact in this entire history: the United States has had a continuous federal retirement-research office for ninety years. No private institution existed to do this work in the 1930s, 1940s, or 1950s. What data existed on how Americans actually supported themselves in old age came from this one federal office, from the decennial Census, and from ad hoc academic surveys with no continuity from one to the next.
Data Enters the Policy Arena
The 1960s turned this quiet administrative function into a live input to national policy. In 1962, SSA’s research office and the Census Bureau jointly fielded the first nationwide survey of the aged population’s income — the 1963 Survey of the Aged. That same year, President Kennedy established the Committee on Corporate Pension Funds to examine the financial stability of the rapidly growing private pension system. Its most consequential empirical work came from career SSA researchers detailed onto it, chief among them Walter Kolodrubetz, whose studies of private pension coverage and vesting underpinned the Committee’s 1965 report and fed directly into the legislative process that produced ERISA in 1974. In 1969, SSA’s Office of Research and Statistics designed the Retirement History Study, a ten-year longitudinal survey following the same workers through the retirement transition. Outside government, research capacity remained thin — BLS did not run its own annual Employee Benefits Survey until 1979, and actuarial firms kept their client studies private.
The Gap I Was Determined to Close
The years after ERISA made the research gap acute. Employers, unions, insurers, and regulators were operating under a complex new federal statute with no ongoing, nonpartisan institution positioned to track how it was actually working. SSA’s research office, however capable, was an internal government function — not a forum where a labor union, an insurer, and a Fortune 500 employer could sit on the same board and agree to trust the same numbers. I had a personal stake in that problem before I had a professional one. My grandfather, Howard Ries, spent thirty-two years as a CLU with the Equitable and died in 1954 with no pension — the family home was sold, and my grandmother moved in with us. Whether a benefit promise gets kept has never been an abstraction to me. That is the gap EBRI was built to close in 1978, on three principles: benefit plans serve an essential economic function, there is an ongoing need for objective information about that system, and a broad, competing membership base is better served by disinterested research than by advocacy.
Parallel Tracks, 1978–1998
President Carter established the Commission on Pension Policy in July 1978, drawing again on SSA’s research bench — Virginia Reno was detailed from the Office of Research and Statistics, and Tom Woodruff, out of the Labor Department, ran it as executive director. Its 1981 report proposed a mandatory universal pension add-on to Social Security that was never enacted, but it shaped debate for years, exactly as the Kennedy Committee’s report had sixteen years earlier. Like the Kennedy Committee, it disbanded once its report was delivered. EBRI, by contrast, was built to be permanent, and spent this period building the research programs — including what became our long-running Retirement Confidence Survey — that gave industry, employers, and policymakers a continuously updated, non-governmental source of retirement data for the first time in the field’s history.
The Expansion Era, and What It Built
This is the period in which retirement data reached its broadest institutional footprint across every sector that uses it. In 1998, SSA launched the Retirement Research Center program — the first centers at Boston College and Michigan, NBER added in 2003, consolidated in 2019 into the six-institution Retirement and Disability Research Consortium. At the same time, a very different kind of retirement data was accumulating outside any government or academic channel: administrative recordkeeping data held by Vanguard, Fidelity, BlackRock, and Empower, each sitting on tens of millions of participant accounts — a scale no survey could match, but privately held and rarely released for outside replication. By the late 2010s we had reached a genuine high-water mark: a federal statistical base, a federally funded academic layer, EBRI’s independent nonpartisan research, and a large body of proprietary administrative data all operating simultaneously.
What We Lost in 2025
On February 21, 2025, SSA terminated the Retirement and Disability Research Consortium’s funding entirely, citing cost savings and objections to DEI-focused research within the portfolio. NBER’s center formally closed that July, after twenty-two years of continuous operation. Of the six RDRC institutions, only Boston College’s Center for Retirement Research survived, and only on outside funding. SSA’s internal research office — the direct descendant of the 1935 Bureau of Research and Statistics — continues to operate, but the extramural academic layer built up over more than two decades has, for practical purposes, been reduced to a single surviving center.
I want to be clear about what I am doing in the rest of this section. Everything above is what the record shows. What follows is what I believe it means — and those are not the same statement.
How Each Sector Actually Uses This Data
The history above describes who built and funded the infrastructure. This is who uses it, and for what — drawn from what I have watched policymakers, researchers, industry, employers, and individuals actually do with it.
Government Relies On It To Price Legislation Before It Is Enacted
Congress, Treasury, the Joint Committee on Taxation, and SSA’s own actuaries use retirement microdata to estimate how a proposed change — to contribution limits, to the retirement age, to catch-up provisions — would affect participants across income and age groups, before the vote rather than after. This is the same function SSA’s research office performed for the Kennedy Committee and Carter Commission sixty years ago.
Academia Turns It Into The Evidence Base For Everything Downstream
University researchers, historically concentrated at Boston College, Michigan, NBER, and Wharton, use SSA administrative records, the Fed’s Survey of Consumer Finances, and Census’s CPS to study how retirement timing responds to Social Security rules and how wealth is accumulated and depleted across a lifetime. This is the layer most directly damaged by the 2025 RDRC termination, because it depended on standing grants rather than one-time data access.
Industry Uses Two Different Kinds Of Data For Two Different Purposes, And Readers Should Not Confuse Them
Public federal and survey data — the SCF, BLS compensation data, EBRI’s Retirement Confidence Survey — gets used for competitive benchmarking and regulatory comment letters. Proprietary recordkeeping data — each firm’s own book of participant accounts — gets used for product development and, increasingly, for content marketing published through in-house “institutes.” One is drawn from broad, replicable public data. The other is drawn from a single firm’s own customers and is rarely opened to outside verification. I think that distinction matters more today than at any point in my career, because the volume of the second kind has never been higher.
Employers Make Real Plan-Design Decisions On What This Data Tells Them
Employers without in-house actuarial staff rely on EBRI, BLS, and recordkeeper benchmarking to decide whether to add automatic enrollment, how to set default contribution rates, and how their benefits compare for recruiting. This is the most consequential of the near-term uses of everything described above, because it shapes real retirement outcomes for the employees covered by those plans.
Individuals Encounter The Downstream Product, Rarely The Data Itself
My own parents lived forty-six years in retirement on an annuity and Social Security. They never saw a Survey of Consumer Finances table or an SSA microdata file — they saw a benefit check and an annuity payment that arrived, reliably, for forty-six years. That is what this entire infrastructure is actually for. Retirement calculators, robo-advisors, and planning tools are calibrated, often several layers removed, against the same federal and survey data described above. Individuals rarely encounter the raw numbers; they encounter what research institutions do with them. That is exactly why the independence of those institutions matters as much to an individual saver as it does to a member of Congress.
What I Believe This Requires
Ninety years after the Social Security Board created its first research bureau, the American retirement-data system rests on four layers: a federal statistical base that has existed continuously since 1935, an academic research layer that expanded from 1998 to 2019 and was sharply cut in 2025, a nonpartisan private institution in EBRI that I helped build in 1978 for exactly this kind of gap, and a rapidly growing body of proprietary data held by recordkeepers and asset managers. Every sector described above depends on some combination of these four layers. As of 2026, the oldest and newest of those layers are the ones still standing at full strength. The middle two are carrying a larger share of the field’s independent, non-commercial output than they have carried since the expansion era began — at precisely the moment the commercial, product-linked share of published retirement research is at its highest point in the field’s history. I do not see that imbalance correcting itself. The data on what happens when independent research capacity disappears have been clear since long before I entered this field. The missing ingredient has always been the will to fund the work before the crisis it would have prevented arrives.
