STATEMENT FOR THE RECORD
STEPHEN A. MESKIN, PH.D., FSA, MAAA,
DEPARTMENT OF VETERANS AFFAIRS
"THE AMERICAN COMMUNITY SURVEY:
THE CHALLENGES OF ELIMINATING THE LONG FORM FROM
THE 2010 CENSUS"
THE UNITED STATES HOUSE OF REPRESENTATIVES
COMMITTEE ON GOVERNMENT REFORM
SUBCOMMITTEE ON TECHNOLOGY, INFORMATION POLICY, INTERGOVERNMENTAL RELATIONS,
AND THE CENSUS
May 13, 2003
I am very pleased to have the opportunity to record full support for the Census Bureau’s American Community Survey (ACS) program on behalf of the Department of Veterans Affairs (VA). This vitally important data collection effort would be a great benefit to the 25.4 million veterans VA is here to serve, their dependents, and the nation as a whole.
A key function of VA’s Office of the Actuary is to collaborate with other federal agencies on the collection of data on veterans from non-VA sources. The ACS will become one of the most critical sources of these data. Indeed, the Office of the Actuary serves officially as VA’s representative to the Interagency Committee advising on the ACS. As such, we work closely with others throughout VA to assess VA’s data needs and to make sure those needs are addressed during the planning of the proposed ACS.
Our major interest in the ACS stems from its unique design to collect census sample, or “long form” data, regularly at intervals much shorter than the ten–year period of the traditional decennial census, reliable for relatively small geographic areas, and ready to use relatively soon after collection. Because veteran status is an integral part of the questionnaire, the promise of ACS is ultimately to provide timely and reliable annual local area data on the number and characteristics of veterans in the general population across America. VA cannot plan services delivered to veterans where they live if we do not have up-to-date, small area, veteran-specific information, particularly since we know that veterans differ from the general adult population in many ways, including their age, sex, and racial composition, socioeconomic and disability status, and migration patterns. The data we currently use, from the decennial census, administrative records, and surveys, collected under VA and non-VA auspices, are either quickly out-dated or limited in scope or not reliable below the national level. As a practical matter, the ACS would be the only data source not having any of these deficiencies.
A major need for data at VA is to help comply with select provisions of federal legislation or VA-based initiatives. Much of the legislation and initiatives require VA to plan for the delivery of services to meet critical needs of veterans where they live, or requires VA to gather information on relatively small, but important groups of veterans, such as women, the disabled, and ethnic and racial minorities. To meet all of these provisions, summary data are required which can help us estimate not just the number of veterans, but also the joint distribution of where they live, and what they look like, including their sex, race, age, education, income, marital status, functional disabilities, and household composition. Data from our own (and non-VA) sample surveys are usually rich in content. But cost constraints keep them infrequent and too small to provide estimates below the national level or to provide reliable estimates of relatively small groups, such as female or minority veterans. The data become quickly out-dated and are limited in geographic scope. While administrative databases alone generally provide recurring local area data, the content is often limited and the universe of veterans to which the data apply is confined to those whom VA already serves. For a reliable estimate of the demand for services, data on veterans in the general population are needed. The ACS will be the only source to provide such data—that is, data rich in content for veterans in the general population, collected annually and large enough, when pooled over time, to provide small area estimates.
The census long-form, of course, has in the past provided data rich in content on veterans in the general population, and reliable for small areas. But the data were collected only once a decade, making inter-censal estimates necessary. The reliability of those estimates is inversely related to the time between censuses. In contrast, the promise of the ACS will make inter-censal estimates unnecessary. We will be assured of timely data, which becomes more important as the veteran population changes rapidly and new issues emerge. Over the decade of the nineties, the proportion of older veterans, minority veterans and female veterans increased and the population experienced significant shifts from the Rustbelt to the Sunbelt, particularly among older veterans. For example, although the overall veteran population in the U.S. and Puerto Rico declined by 5 percent between 1990 and 2000, veterans 65 or older increased by 35 percent, and women veterans by 41 percent. Rapid changes in the characteristics of the veteran population and shifts in where veterans live require timely data to make adjustments to the mix of services VA must be prepared to provide and where they are delivered.
The Census Bureau’s report, Federal Agency Requirements for American Community Survey Data, (revised) January 2003, cites VA’s myriad legislative requirements (along with those of other agencies) which the ACS would address. A few detailed examples of those requirements and one major VA initiative are given below:
VA Health Care Enrollment: In the Veterans’ Health Care Eligibility Reform Act of 1996 Congress required VA to enroll most veterans seeking VA health care. An enrollment priority scheme was specified in the new law giving highest priority to veterans receiving service-connected disability compensation and lowest priority to high income, non-service-connected veterans. Given the discretionary nature of VA health care financing, this law required VA to forecast the expected demand and associated cost of treating enrollees. One outcome of the law was the advent of market studies for VA care where VA is viewed as a major supplier of health care services to veterans. A key aspect of these market studies has been the estimation and forecasting of the number veterans classified by enrollment priority and living in small geographic areas, such as counties. Long form data (including veteran status, age, income, household size and functional disability) provide the only data source for both the measures needed and the statistical power to permit inference at such small levels of geographic aggregation. The frequent recurrence of ACS data promises to make estimates and forecasts more timely and accurate.
u>Long Term Care: Sections 1710, 1710A, and 1710B of title 38, United States Code, establish VA health care priorities for VA nursing home care in particular, and long term care (nursing home, and home and community based care and services, etc.) more generally. That is, VA is required to plan long term care services for eligible veterans, to estimate and project veteran sub-populations at risk of use or need for long term care services, and to estimate and project potential use of VA long term care services. As such, the ACS questions on functional disability, in conjunction with other questions, would provide continuous baseline data for estimating and projecting the demand for long term care.
u>Burial Benefits: A provision of the 1999 Veterans Millennium Health Care and Benefits Act, P.L. 106-117, Section 613(b)(2), required VA to assess the need for additional national cemeteries to the year 2020. The standard to be met was to project the need for cemeteries such that 90 percent of the veteran population would be within 75 miles of a national cemetery by 2005, projecting out to 2020. Such an assessment required data on the veteran population by detailed level of geography. The model used to make the projections incorporated small area data from the decennial census as a baseline for the estimates and projections of the size and location of the veteran population. The need for such planning continues and the ACS would be used to provide a continuous baseline of small area data for updated and accurate estimates and projections.
Minority and Women Veterans: Sections 317 and 318 of title 38, United States Code, established the Centers for Minority and Women Veterans. One of their legislated functions is to conduct and sponsor appropriate social and demographic research on the needs of veterans who are minorities and women and the extent to which programs authorized under title 38 meet the needs of those veterans. Since existing administrative data often do not contain minority status indicators and since existing surveys are usually too small to provide reliable estimates of relatively small populations, the ACS would be a major source of recurring data to comply with this on going legislated requirement.
VA Capital Asset Realignment for Enhanced Services (CARES): In 1999, a GAO study recommended a locally based market assessment of how veterans’ medical needs are aligned with VA’s own assets, that is buildings and facilities. As a result, VA undertook the CARES initiative to help align capital assets with aggregate demand, driven, in part, by the needs of an aging veteran population, shifts in the geographic distribution of the population, and the enrollment system. Such an assessment requires reliable local data on veterans and their characteristics. The model used was based in part on decennial data to provide estimates of local veteran populations. Updates of CARES would greatly benefit from continuous small area data from the ACS.
The Veterans Actuarial Model
Perhaps the primary responsibility of VA’s Office of the Actuary is to develop historical estimates and project future trends in the veteran population, VA program utilization, and associated costs. As part of this function, the Office of the Actuary produces VA’s official veteran population estimates and projections, using actuarial and demographic assumptions and techniques. As official data, the estimates and projections are used extensively within VA for budgeting and planning and by our major stakeholders throughout government and the private sector. The Veterans Actuarial Model is a multi-year effort currently underway, which builds on earlier Office of the Actuary models, to provide population estimates and projections as well as program needs for most VA business lines. Integral components of the earlier models are county-level decennial census data on veterans. Continuous small area estimates from ACS would be critical to the Veterans Actuarial Model and its annual updates. The promised frequency of such data would make our veteran estimates and projections timely and far more reliable. Of particular importance would be the promise of reliable and timely small area data on population movement and data for inferences about veteran-specific deaths to address the two weakest components of demographic projection models, viz., migration and mortality. Furthermore, annual ACS estimates of the veteran population used for the Veterans Actuarial Model would serve as a benchmark to evaluate and adjust independent estimates from other surveys and other data sources.