Introduction

In the context of Federal statistical programs, interagency funding means the transfer of funds from one Federal agency to another for the performance of data collection, analysis, or a related statistical activity. In each case, there is involvement of one or more sponsor agencies which provide the funds and a performing agency which does the work.

To put the discussion of interagency funding in the proper perspective, the close relationship between the issues of interagency funding and organizational questions should be recognized. In recent years, a principal emphasis of efforts to reorganize Federal statistical activities has been to limit the number of Federal offices directly engaged in statistical operations and to establish an appropriate number of statistical centers to perform statistical operations for all of the other Federal offices. The by-product of such a reorganization effort is an increase in reimbursable funding, that is, an increase in the transfer of Federal funds to the recognized statistical centers. Inversely, the cases in which recognized statistical centers are not heavily engaged in reimbursable programs indicate areas in which the concept of statistical centers has been only partially carried out. This of course is not clear cut. For example, the total reimbursable work done by the National Center for Health Statistics (NCHS) in fiscal year 1976 was only $221 thousand but a part of the short-fall in its role as a reimbursable center is its lack of flexible staffing to respond to outside calls for assistance.

Another issue to which reimbursable funding has frequently been linked is the role of market demand in the funding of statistical activities. Stated simply, the proposition is that an agency which needs data should pay for them. If an agency is unwilling to pay, there is the presumption that failure to pay is the equivalent of a market response which states that either there is no need or the expressed need is of a lower order of priority than other needs. There are certain situations where willingness to pay is a legitimate test of priority, but any broad application must confront the nature of the budget formulation process which definitely limits applicability. In short,

the budget process tends to specify, in advance, the objects of expenditure and hence to limit the possibilities for setting priorities within broad areas.

Another factor which complicates the situation is that many statistical programs are addressed to both agency-specific and general-purpose needs. The market is broader than an identifiable agency and, therefore, willingness to pay does not provide a good measure of the overall need and hence the real priority of the project from the point of view of overall governmental needs.

In spite of all of the above limitations, market demand does have a definite role in shaping the outputs of the Federal Statistical System. Within broader budget constraints the reimbursable mechanism provides flexibility for funding some major statistical projects and not funding others as agency priorities shift.

Current Activities

Only a small number of agencies perform significant amounts of statistical work for other Federal agencies. Since data collection is more expensive and requires more specialists than data analysis, most of the interagency funding reimburses data collection rather than analytic projects and thus the principal performers are the core collection agencies. The degree of concentration in only three agencies, the Bureau of the Census, the Bureau of Labor Statistics, and the Statistical Division of the Economics, Statistics, and Cooperatives Service of the Department of Agriculture, however, is quite striking and may be surprising to those who are used to thinking of the U.S. Federal Statistical System as highly decentralized. When reimbursable situations are included, these agencies account for approximately 40% of the Federal statistical budget. With regard to deciding what data are needed, the system is correctly described as decentralized, but statistical collection is, in fact, quite centralized.

'See Section II, “The Organizaion of U.S. Federal Statistics" for discussion of the roles and missions of this group of agencies.

The second possibly surprising fact is the extent to which the functional statistical centers use the core statistical centers as their collecting agents. This practice, in fact, is what leads to highly concentrated statistical operations. The Bureau of the Census, for example, is involved in major data collection projects for the National Center for Health Statistics (NCHS), the National Center for Education Statistics (NCES), the Law Enforcement Assistance Administration (LEAA), and the Department of Housing and Urban Development (HUD). In fiscal year 1977 these four agencies together planned obligations of $28.8 million to the Bureau of the Census for reimbursable projects. Of this total, NCHS estimated obligations of $4.3 million, NCES $0.9 million, LEAA $11.9 million, and HUD $11.7 million. None of these agencies, on the other hand, is engaged to any significant degree as a collection agent for the Bureau of the Census or for any other Federal agencies. These agencies are functional collection agencies only in the sense that their own data collection or collection sponsored by them is multipurpose in nature.

The Bureau of Mines does function as a collection agent for energy and minerals data although the volume of work performed on a reimbursable basis is a great deal less than that performed by the Bureau of the Census, the Bureau of Labor Statistics, and the Economics, Statistics, and Cooperatives Service. The total estimated 1977 obligations to the Bureau of Mines for reimbursable statistical work amounted to $0.5 million, whereas the Bureau of Labor Statistics estimated $14.4 million of obligations and the Bureau of the Census estimated $62.7 million in accruals. In the data analysis area, both the Bureau of Economic Analysis (BEA) and the Economic Research Division of the Economics, Statistics, and Cooperatives Service (ESCS) are engaged in a substantial number of projects on a reimbursable basis. BEA estimated $2.2 million of obligations primarily related to seven separate projects, and ESCS/Economics estimated $4.0 million related to 16 separate projects in 1976, of which by far the largest were river basin studies and investigations of watershed and flood prevention programs, which are sponsored by the Soil Conservation Service. The Bureau of Labor Statistics also includes data analysis projects in its reimbursable account.

In sum, the Departments of Agriculture, Commerce, and Labor all have established statistical centers which accept a substantial volume of data collection and analysis projects on a reimbursable basis from other agencies both within and outside their own Departments. Outside of this small group of Departments and agencies, however, statistical collection and analysis centers are much less well established.

Are the existing practices satisfactory or should there be further development of statistical centers to perform substantial amounts of reimbursable work for others? In answering this question, one can take at least a brief look at the current constraints to further development. Some of the easily identifiable problems are organizational. Within the Justice Department, where both the Law Enforcement Assistance Administration and the Federal Bureau of Investigation are responsible for major segments of criminal justice statistics, basic issues concerning departmental responsibilities for data collection and analysis need to be resolved before the appropriate role of any Department of Justice agency with regard to reimbursable work can be contemplated. Within the Interior Department, at least with regard to energy and minerals data, the concept of a statistical center performing work for others on a reimbursable basis is well-rooted in the Bureau of Mines and could probably develop further without organizational changes although the Energy Information Administration in the Department of Energy has competing responsibilities. In addition, the Bureau of Mines does a considerable amount of work for other agencies without charge. Examples are world minerals data sent to the Federal Power Commission and the Departments of State and Defense, and petroleum and coal data which were previously sent to the Federal Energy Administration and the Energy Research and Development Administration before the Department of Energy was formed. This does not imply, however, that the existing organization within the Department of the Interior could not be further improved.

Lack of adequate staffing of statistical centers has also led to limitations on the extent to which they have been able to perform reimbursable work. The National Center for Education Statistics (NCES), for example, in spite of a favorable organizational location within the Department of Health, Education, and Welfare (DHEW), has a limited pool of expertise because of the small numbers of statisticians and professional employees in closely related fields. For this reason alone, NCES is not and cannot now be a major statistical resource available either to the other education agencies of DHEW or to other departments sponsoring education data collection activities.

Even agencies with substantial numbers of highly qualified professional statisticians, however, may find that their capability to undertake reimbursable

'The chapter on "Professional Staffing and Professional Staff Training" contains a detailed analysis of employment of statisticians by agency and further discussion of the adequacy of staff resources in fulfilling agency roles and missions.

work for other agencies is very limited. The reason is that the entire staff may be required to operate the agency's base program. Among the statistical agencies, only the Bureau of the Census, the Bureau of Labor Statistics, and the National Center for Health Statistics have employees whose positions are budgeted exclusively for the performance of reimbursable work. Prior to 1977, NCHS was allowed only 9 positions for reimbursable work. The fiscal year 1977 authorization was 34. The Economics, Statistics, and Cooperatives Service, in spite of the large amount of reimbursable work it performs, has never been allocated staff explicitly for such work.

It may seem paradoxical that ESCS can undertake large amounts of reimbursable work without fulltime permanent staff devoted to these projects. It must be recognized, however, that some part of the statistical work for other agencies can be absorbed into the base program of a large statistical center. This limits the need for staff positions devoted to reimbursable work and in many situations may be the appropriate way to satisfy a given demand.3

Recommendations

The preceding sections have focused on the small number of Federal agencies which are currently performing statistical work, either data collection or analysis, for other Federal agencies. The problems in delivery of these services which have been identified and discussed represent the major constraints to the supply of such services. Not surprisingly, therefore, one set of policy recommendations is addressed to removing or reducing these constraints as appropriate.

First, further organizational changes need to be made. Emphasis needs to be placed on revising the role and mission statements of selected agencies with the general objective of assuring that departments have at least one statistical center in each major functional area, that is, an office which includes in its mission the performance of statistical work for other agencies.

Implementation of this recommendation would lessen the high degree of concentration of reimbursable work in three agencies which now prevails. More importantly, however, it should contribute to reducing the total number of agencies engaged in data collection on their own account and would, hence, lead to more centralization for the statistical

'The Appendix to this chapter discusses the rationale of deciding whether a data collection or analysis project should be handled as a reimbursable project or as part of the base program of the statistical center.

system as a whole. This objective has been generally accepted as desirable and should not have any significant negative consequences if properly executed.

Second, once organizationally established, all statistical centers must have in-house both program generalists and specialists sufficient for a major base program in at least one functional area. An agency without such a diversified staff cannot become a significant data collection or analysis agent for other agencies.

The basic argument for insisting that statistical centers have fairly extensive operations is that smaller operations cannot utilize on a full-time basis all the specialists which combine to produce the most professional Federal statistical output. One possible danger, or pitfall to avoid, would be increasing professional staff in anticipation of reimbursable work. One result could be staff underutilization or less than optimal utilization by spending time seeking contracts with other agencies.

Third, established statistical centers with demonstrable demand for reimbursable services should analyze those projects which the center has had to refuse because of staff constraints. Important types of information which should be included are: (1) prospective sponsoring agency or agencies; (2) general type and scope of project, for example, data collection, data analysis; (3) periodicity, that is, single-time, continuing; (4) resources required, especially personnel; and (5) functional area, for example, health statitics or education statistics. For continued justification for reimbursable positions already allowed, a similar analysis could be performed with regard to reimbursable work actually being performed by the agency. Both kinds of analysis could make important contributions to assuring that reimbursable positions are in line with effective and legitimate demand for the performance of reimbursable statistical activities.

In contrast to the first three recommendations, which are basically directed at reducing constraints to the supply of reimbursable statistical services, the final recommendation in this section is directed toward reducing undue constraints to the demand for such services. In brief, what is called for is analysis of all statistical services which agencies purchase from (1) private contractors, (2) State or local governments, and (3) other Federal agencies. For group (3), a further break by individual agency would also be relevant. Such analysis could lay the groundwork for discussions by the purchasing agencies concerning the current and potential capability of the Federal statistical centers as "suppliers." To return for a

moment to the previous discussion of the market demand aspects of the production and analysis of Federal statistics, one may point out that the flexibility in choosing who is to perform the statistical work may be as great as the latitude in deciding what is to be produced. A major challenge would be to identify changes which would make the statistical centers more competitive as suppliers.

Steps To Be Taken

Specific steps on functional organization and staffing recommendations are contained in other chapters. Implementation of the recommendation on analyzing the demand for reimbursable work could proceed by including discussion of the proposed analysis on the agenda for spring planning meetings with the core multipurpose statistical agencies. At that time, details of preparing material specifically for justifying reimbursable positions could be discussed. Reports could be timed so that they would be available during the fall period of budget review.

In implementing the recommendation concerning the purchase of statistical services, preliminary analysis on which "buyer" agencies to contact could be undertaken by the Office of Federal Statistical Policy and Standards. As already mentioned, some of the major buyers of statistical services are the core multipurpose statistical collection agencies themselves. Others are the program collection and analysis agencies. Representative agencies from both groups should be contacted in the implementation of this recommendation. Again, early spring planning meetings would provide an appropriate opportunity to raise the issue and detailed reports could be requested for review in the summer prior to submission of agency budget requests.

Appendix

The material which follows describes selected issues which must be addressed by the agencies in the implementation of specific activities under reimbursable program agreements.

Reimbursable Versus Base-Program Funding

In cases where a statistical agency is planning to produce some output for another agency or agencies, the first operational question is location of the required funds in the budget. If funds are located in the budget of the statistical agency providing service to other agencies, the project essentially becomes a part of its base program and no reimbursement is required. In a technical sense, it is not a reimbursable project at all. Coincidentally, the question of related personnel will disappear as a separate issue. There is, moreover, solid precedent for this type of

arrangement. One good example is the Professional, Administrative, Technical, and Clerical (PATC) survey used to help establish comparability between Federal and private-sector pay. The Bureau of Labor Statistics regularly performs this survey as a part of its base program even though the users are the Civil Service Commission and the Office of Management and Budget. On the other hand, funds for the Annual Housing Survey (AHS) are located in the budgets of the sponsoring agencies, the Department of Housing and Urban Development and the Department of Transportation, rather than in the budget of the service agency, which is the Bureau of the Census.

Although practice varies, a general rationale may be set forth to help in deciding whether a project should be funded in the budget of the sponsor or the service agency. The key elements in making this decision should be: (1) whether the project is single time or repetitive and (2) whether the number of sponsors is limited, i.e., one or two, or more numerous. The situation may be conveniently summarized by the matrix below:

Single-time projects for one or two user agencies should generally be funded by them. Single-time projects do not lend themselves well to becoming part of another agency's program and the transfer of funds should be relatively uncomplicated. In contrast, repetitive projects with numerous users should be funded in the base program of the service agency. Because of the predictability of budget requirements, they can become part of a program, and the excessive budgetary fragmentation associated with numerous small transfers will be avoided.

Solutions for the two intermediate cases are not as clear. A single-time project with numerous users might be funded by the service agency to avoid budgetary fragmentation or by a group of sponsors because of availability of funds in line with their roles and missions. This situation, however, is one which occurs only rarely. The other intermediate situation is the one of a repetitive project with one or two users. Both of the examples previously mentioned (PATC and the AHS) fit into this category. It is practical to handle such a project either as a reimbursable item or as part of the base program of the service agency.

Of course, other factors may enter into the final decision on whether funding should be base program

or reimbursable. One of these is the degree of flexibility which the collection or analysis agency has in accepting additional reimbursable work. Unless an agency can get reimbursable positions when needed, the choice between base program and reimbursable funding may be quite limited.

If an interagency transfer of funds is involved, another operational problem which arises is that of year-specific funding. The transfer of funds from sponsor to service agency generally is made on the basis of cost estimates worked out in advance of performance of the work. Furthermore, with reimbursable work, there is probably some upward bias in the cost estimates. The service agency can return unused funds more easily if costs are overestimated than it can get additional funds if costs are underestimated.

When year-specific funds are returned to the sponsor agency, however, it is sometimes too late in the fiscal year to use them fully, and they ultimately revert to the U.S. Treasury. Agencies utilizing their own funds are less likely to have unused balances. The one readily apparent solution to this problem, however, "no-year" funds for reimbursable projects - would create problems greater than the one solved. For example, the accumulation of carryover funds erodes the normal discipline of direct budgeting. In practical terms, an agency can plan ahead for certain projects for which returned funds could be immediately used.

If the reimbursable project involves data collection, another operational question which emerges is whether the sponsor or the service agency should submit a request for clearance of the statistical project to the Office of Management and Budget (OMB) and its delegated agent the Office of Federal Statistical Policy and Standards. In general, the sponsoring agency should submit the clearance, and be able to defend the contents of the forms even though the service agency may have played a major role in project development. The only exception to this policy is the case of an interagency agreement between an agency whose forms are subject to Federal Reports Act clearance and an agency whose forms are not subject to such clearance. In such a case the subject agency submits the request for clearance whether it is the sponsor or the collecting agent. (See Transmittal Memorandum No. 2, O.M.B. Circular No. A-40, November 5, 1976.)

Rights and Obligations in Interagency Funding

In addition to the obvious need for explicit agreement between a sponsor and service agency on

project content, cost, and timing, there needs to be a clear understanding concerning several other questions. One such question is the extent of responsibility which the collecting agent should assume for data analysis. To a certain extent, this depends on the preferences of both the sponsor and the collecting agent. The collecting agent should, however, perform at least sufficient analysis to be assured that the product it transfers to the sponsor agency meets its own standards of quality. Furthermore, the sponsor should not request raw, unedited data which do not meet such standards. Of course, with mutual agreement, the service agency can assume further analytic tasks, including data tabulations and calculation of indexes. For later stages of analysis, such as index calculation, the sponsor agency will normally be responsible for the methodology used.

Another question about which there has been some lack of clarity is the extent of proprietary interest of the sponsoring agency. Does the sponsoring agency own the individual records? The answer to this question usually depends on the statutory authority for collecting and maintaining the records. The sponsoring agency should be able to receive identifiable records if it has a need for them and if all guarantees of confidentiality are fully protected by both collecting and sponsoring agencies and the respondent is informed of the planned transfer of data. (See Confidentiality Chapter.) In the case of data collected from individuals the Privacy Act provisions for disclosure under the routine use procedure must also be adhered to. On the other hand, the proprietary interest of the sponsor agency does not extend so far as to prohibit the collecting agent from using the records to perform a follow-up quality measurement survey on its own account.

A final question needing clarification is the circumstances under which a collecting or analytic agency should refuse reimbursable work. One such situation occurs when authorized personnel are fully committed to the base program and existing reimbursable work. Existence of reimbursable positions may help to introduce flexibility in this respect and so may the possibility of subcontracting. This latter should only be undertaken, however, with knowledge and consent of the sponsoring agency. Another reason to refuse reimbursable work occurs when the project falls outside the agency's mission. The Economics, Statistics, and Cooperatives Service in the Department of Agriculture, for example, generally limits its work to projects which have a clear