The public health community has expended much time and effort developing models for classifying public health services in an effort to link performance measures to different public health activities. Several closely related frameworks have emerged, including the “Ten Essential Services” proposed by the Core Public Health Functions Steering Committee at the instigation of the US Department of Health and Human Services (HHS)1 and the National Association of State Budget Officers (NASBO) framework based on the mission categories developed by the Core Public Health Functions Steering Committee.2 Members of the public health community have expressed concern to us about the lack of engagement of elected officials in using these frameworks. This article discusses the criteria that were important to legislators and legislative staff in Georgia during their effort to try to develop a program structure as a part of Georgia's transition to program-based, performance-oriented budgeting. This model led to a program framework that was similar to the one promoted by the NASBO but varied significantly from the “Ten Essential Services” framework. Key reasons why differences emerged were that the legislators were concerned about maintaining control, which, in turn, depends on clear lines of accountability. They also have to take into account practical, strategic, and political considerations.
The Transition From an Object Class to Program Budget
In 2005, the Georgia General Assembly switched its entire appropriations process from appropriating money by “object class” to appropriating money by program. In prior years, Georgia budgeted by selecting functional categories, such as human resources, facilities, and administration, and by “object class” or input categories, such as personnel, travel, motor vehicles, equipment, and contracts. The legal weight of the budget fell on the object classes, because the appropriations bills stipulated that an administrator could only increase funds in an object class category by two percent and any increases had to come from a source internal to the agency—the total budgeted amount could not change. This requirement restricted administrators' ability to move funds between the object classes in the budget, and agencies were audited to ensure compliance. In contrast, agencies could move funds freely between functional categories although they might have to report such changes to the governor's budget office.
With fiscal control centered around object classes, legislators focused on the money associated with inputs and, thus, were likely to ask questions about whether allocations for inputs such as personnel, travel, or motor vehicle purchases were too much or too little.3,4 Strategically, this traditional budget allowed legislators to control aspects of agency activities such as the number of people employed, or whether an agency contracted out a service or provided it in-house. The problem with this orientation was that legislators were never going to know as much as administrators about how best to deploy resources. The Georgia legislature, like the majority of states, is part time,5 and the exercise of object class control leads, often rightly, to charges of micromanagement.* It can also lead to very low levels of agency accountability since the information provided sheds no light on the services the agency is providing or whether the agency is achieving meaningful results with the money it receives. While legislators may want to continue to keep an eye on object classes to make sure that agencies are not “hiring too many people” or “going to too many conferences,” using an input orientation as the centerpiece of budgetary information and control is likely not an effective use of legislative or administrative resources.
What legislators and elected officials should know better than many managers are the expectations and goals of the public about a public service. In addition, legislators are positioned outside the executive-administrative relationship and so can provide an external source of accountability and oversight.6 When Georgia switched to a program budget, the transition was intended to reorient legislative decision making toward the goals and objectives that they wanted an agency to achieve, the application of resources to this end, and the evaluation of adequate agency progress. Figure 1 shows the differences in the presentation of the object class budget for Division of Public Health (DPH) in state fiscal year (SFY) 2005 and the program budget in SFY 2006.
Developing a Program Structure
What is a program?
The switch to programs requires agencies, the executive, and the legislature to answer a key question: “What is a program?” The “textbook” criterion for defining a program in a budget is that it should be a cluster of activities oriented toward a single objective or end product.7 Most state definitions reflect this general orientation. In Georgia, the executive branch's Office of Planning and Budget defines programs as “systematic sets of activities undertaken to carry out an agency's core business.”† South Carolina defines a program as “something an organization does to accomplish its goals and objectives. Moreover, an activity is defined as something which consumes resources and produces a product, service, or outcome.”8 Texas defines a program as a “method to achieve goals and objectives. Formulated from goals and objectives a [program] is a means of transforming inputs into outputs and ultimately outcomes with the best use of resources.”9 Virginia describes programs as areas of expenditure that support one or more products or services.10,*
However, one of the fundamental dilemmas of program budgeting is that the definitions described can encompass a number of differing programmatic arrangements. Figure 2 shows some of the program and subprogram changes proposed by the DPH (note the differences from Figure 1), and the program structure that the legislature adopted for the SFY 2007 budget. Each box represents a distinct “program” for budget purposes.
As Figure 2 shows, a program can be defined at different levels of aggregation and still meet the state definition of a “program” described above. Furthermore, some might argue that the “subprograms” should properly be “programs” based on the definitions above.
One of the reasons that Figure 2 shows different program structures is that program definitions have implications related to accountability and control. Where previously an “object class” designation locked an agency into a restricted level of spending on inputs, now agencies are locked into a restricted level of spending on programs. In SFY 2006, agencies could add only a two percent increase of funds or $250,000 whichever was less, to any program category, and the audit process assessed agency spending by program.† Thus, for strategic and management reasons, agencies and the executive prefer to define programs broadly. Large programs encompassing many different services give the agencies and the governor the discretion to move money freely between many competing public services and thus the power to determine state priorities.
In contrast, the legislators in Georgia preferred program definitions that are discrete, closely associating resources with a particular product or service that leads to a distinct outcome. More narrowly defined programs make state priorities more visible to the public, and because of the legal restrictions attached to a program designation, they also allow legislators greater control over setting funding priorities. As a result, program budgets may lead to tension among the agencies, the executive, and the legislature.11 In Figure 2, the left-hand side shows the agency and gubernatorial preference for large and inclusive programs while the right-hand side shows the legislative tendency toward more differentiated program definitions.
A framework for legislative considerations
This article draws on the personal experiences of the legislative staff in developing the program structure for the DPH in Georgia and uses the criteria that influenced their deliberations to develop a “decision tree” shown in Figure 3 These criteria are not normative and are not meant to be representative of how other legislative bodies might approach program budgeting. However, the decision tree does illustrate how a program structure might be altered in the face of legislative concerns about control and accountability as well as practical, strategic, and political concerns. The decision tree can be broken into roughly four sets of considerations. The first tier describes the initial effort to try to cluster programs around outcomes, products, and services, and reflects a results-based policy orientation—does a program encompass a single product, service (or core business) associated with a distinct set of outcomes? The second set of considerations about the choice to aggregate or disaggregate the proposed program reflects legislative concerns about control of priorities, oversight and evaluation, and the political prominence of a particular issue. The third tier reflects more practical and managerial considerations—here described as “costs” of aggregation or disaggregation. Finally, the fourth tier reflects an initial staff assessment of political will: Are the legislators willing to bear the potential costs of aggregation or disaggregation, and if there are no costs, do legislators want to aggregate/disaggregate a program for political reasons?
Linking resources to outcomes
From a pure results-based orientation, program definitions originate with the identification of outcomes desired and then an assessment of the clusters of activities or strategies being used to achieve these outcomes. These clusters of activities form the basis for “programs,” which are then funded in the budget process. This definition is based on an “outcomes-sequence model” that forms the foundation for program evaluation.12 For purposes of clarity, accountability, and control, legislators tended to look for a very clear relationship between funding, program designations, and an outcome or interrelated cluster of outcomes.
In Figure 2, there are several examples in the public health budget where clusters of activities with fairly distinct outcomes were grouped in the same program. For instance, in the DPH's proposed program structure, epidemiology and vital records were put under the program “Health Information and Assessment.” Although epidemiology may rely to some degree on the data in vital records, and both are involved in data collection, public expectations and the outcomes associated with these public services are distinct. Vital records are the public service responsible for collection, management, and distribution of birth and death certificates, as well as record-keeping on marriage, divorce, terminated pregnancies, and stillbirths. Vital records' outcomes are associated with good customer service in making these records available to the public as well as accuracy of record-keeping. In contrast, epidemiology is concerned with tracking and monitoring diseases, injuries, and at times population health indicators. The public might ask that epidemiologists catch early warning signs of serious outbreaks of communicable diseases and publish reports that guide citizens on health issues of concern to the population at large. Given public health concerns such as severe acute respiratory syndrome, human immunodeficiency virus/acquired immunodeficiency virus, or bird flu, legislators might want to add funds to epidemiology and having this service blended with vital records obscures the relationship between funding and outcomes.
Combining epidemiology and vital records raises an important concern for legislators, which is the issue of control. Control, and the closely associated issue of accountability, requires that programs be defined so that the legislators or their staff can monitor, evaluate, and take action to change expenditures for a particular public purpose.13 If epidemiology is in the same program as vital records, legislators might allocate additional funds to epidemiology but without a specific epidemiology program, they have no way of ensuring that the agency actually spends the funds on epidemiology. Given the DPH's proposed program structure, the agency could allocate the epidemiology funds entirely to vital records. Since audit and legal controls are associated with the program definitions, legislators would have difficulty monitoring this shift in funds and then would have little recourse to ensure that this transfer did not happen again.
Of course, a special evaluation could be conducted to separate out spending on epidemiology and vital records, but this effort would require someone knowing that there might have been a problem in the first place and often requires a more expensive external audit. In sum, a legislator is concerned both about the association of resources with a particular result and about his or her ability to control or adjust those resources based on information about results or to reflect legislative priorities.
Georgia is required constitutionally to balance its budget with the revenue estimate for state funds issued by the governor. Because of the finite availability of dollars, it is also necessary to have programs that are associated with distinct outcomes so that legislators can prioritize the needs of the state and fund accordingly. For example, in a recession, the legislators might still want to maintain or increase the state's epidemiological capabilities, but because of declining revenues, the General Assembly might feel that the vital records program needs to become self-sufficient or should cut costs. This transition is only legally enforceable through separate programs.
Relating programs to short-term or intermediate outcomes
The emphasis on distinct outcomes as well as accountability and control leads to an emphasis on outcomes that can be measured and evaluated within a budgetary cycle. Here, the concepts of “short-term, intermediate, and long-term outcomes” are particularly useful. Short-term and intermediate outcomes are ones over which an agency is likely to have more direct control and are often easier to measure in a budget cycle, whereas long-term outcomes are more difficult to measure and may not be within a particular program's control.12
When Georgia legislators meet each January for their session, they look at two budgets: one that adjusts funds for the remainder of the fiscal year (which ends June 30) and the budget for the next fiscal year. In effect, they prioritize funding for 18 months at a time. Defining programs so that they are associated with short-term to intermediate outcomes allows the General Assembly to monitor progress toward their goals at the beginning of each session.
For instance, Georgia's DPH might claim that their most important outcome is to improve the health of Georgians, and to some degree their program “chronic disease prevention and health promotion” (in Figure 2) had this sort of broad-based outcome, but this outcome might take years to accomplish, is dependent on a variety of external factors, and encompasses almost every aspect of public health funding. Legislators are interested in long-term outcomes, but more useful measures for legislative oversight and annual appropriations are short-term or intermediate outcomes (or even outputs) that indicate progress toward achieving longer term outcomes. Although the actual program structure developed represents a compromise, if the state were to truly show the relationship between state funds and outcomes, these programs would need to be allocated with even greater detail. For instance, it would be useful to see in the budget how much is being spent on teen pregnancy prevention and associated short-term outcomes such as demonstrated changes in attitudes toward teen pregnancy.
This legislative focus on clearly defined short-term or intermediate outcomes as well as the concern about accountability and control was one of the primary reasons that the legislative program designations did not align with the “Ten Essential Services.” This framework, in its current form, defines services so broadly that legislative staff could not figure out how to use it to hold the agency accountable for distinct purposes and outcomes. For example, the program “inform, educate and empower people about health issues” could apply to a range of public health services, including ones as diverse as cancer screening, family planning, or emergency preparedness. This vagueness made it difficult to understand what the program was supposed to accomplish in the near term and how the state could budget to these ends. The difficulty associated with classifying funds by the Ten Essential Services has been observed in other contexts.2,14
In contrast, the programs in the Georgia budget more closely aligned with program categories used by the NASBO which are easier to associate with discrete services, but even here some categories were too general. Looking at Figure 2, the state DPH divided the NASBO category of “prevention of epidemics and the spread of disease” into three distinct programs. The General Assembly divided it into five programs. For instance, legislators wanted to see the association of resources with the agency's progress in areas such as immunization and food establishment inspections—both fairly distinct services and sets of outcomes.
A separate issue related to outcomes occurs when a program is a support service rather than a “core business.” Best practices in cost accounting generally recommend fully allocating all costs to the “core business” programs.15 An example in the public health budget was the original identification of laboratory services as a distinct program. The public health laboratories provide routine testing such as newborn screening as well as public health emergency services such as assessing the spread of disease in epidemics. Although its services are critical to public health, laboratory services do not have public policy outcomes independent of other programs. As a result, the DPH recommended (and the General Assembly agreed) that the funds for laboratory services should be allocated to the various programs that utilize laboratory services. In effect, these programs will “purchase” services from the internal service center. This change is primarily an accounting and budgetary shift; the facilities that provide laboratory services will still function as before.
Despite the legislators' desire to focus on clear associations between resources and outcomes, as well as control and accountability, program definitions often have to be tempered by a realistic understanding of agency operations, intergovernmental grant requirements, and cost accounting problems as well as legal constraints on the separation or aggregation of funds. A further consideration is whether an agency (or the state) has the financial information systems to divide funding along different dimensions.
A particular dilemma in public health was the cost associated with disaggregating existing programs into smaller units, which raised issues of cost accounting, discontinuities between existing organizational arrangements, and program definitions, as well as legislative concerns over having too many small programs to track in the budget. In the end, a number of programs were left in more highly aggregated form because of these practical considerations.
For instance, injury prevention was initially broken out of the large chronic disease prevention and health promotion program with the idea that preventing injuries was a discrete set of activities and had outcomes that differed from the variety of infant, adult, and adolescent health promotion and treatment activities also in the program. However, to truly capture the resource-product/service-outcome relationships, the budget would have had to disaggregate injury prevention into its component parts since the short-term and intermediate outcomes of interest were really associated with programs such as suicide prevention and child and automobile safety. Yet, injury prevention only had a total budget of around $400,000 in state funds, in a budget of more than $18 billion, and disaggregation would both create highly fragmented funding streams for the agency and clutter the legislative budget with very small programs.
Duplication of purposes and outcomes
A consideration in clustering activities into programs is also the organizational “home” of a program. Although program budgeting helps budgeters identify duplication and overlap in services, sometimes there are trade-offs in combining programs. For instance, Emergency Preparedness is located in the DPH, but other emergency response activities are located in the Governor's Office of Homeland Security and the Georgia Emergency Management Agency. These programs collaborate in many ways, and it might be cost beneficial to group them under a single agency as a single program. Aggregating these programs would also make a single agency accountable for results. In the end, legislative staff left these separate since combining government services may also create managerial problems, such as loss of specialized knowledge and organizational confusion.
Political and strategic considerations
While the policy-based and pragmatic considerations form the core of the debate over program structure, a number of practical and political considerations influenced the final program structure of the DPH as well. Part of the political dynamic was whether the legislators were willing to expend political capital on a fight with the agency and the governor over program structure. There were also different agendas in each chamber of the legislature. The political dynamic is evident in the program “adolescent and adult health promotion.” The Senate staff initially proposed that this program included a number of distinct outcomes and thus should be disaggregated. Domestic violence prevention and stroke and heart attack prevention, for example, would have different outcomes during a budget cycle. However, in an agreement with the House, the legislature created a “catch-all” program for adult health promotion activities. Legislative willingness to demand a more refined program structure was influenced both by the political salience of particular policy issue and by their relationship with agency staff. If legislators trust that the agency will use funds in line with legislative intent, then they are more likely to accommodate broader program designations. In contrast, legislators might make a point of trying to designate programs carefully and specifically if they are concerned about a diversion of resources or poor management.
Another important consideration is whether a program is “short term” versus ongoing. A problem in other places in the budget has been that short-term programs intended to address a specific problem “vanished” into the base budget even though their purposes had been achieved. An important way to prevent the loss of a program into the base is to specifically designate it in the budget. Although not currently an issue in the public health budget, an emergency fund for an epidemic might be an example of a short-term program that should be “de-funded” once the program purposes have been achieved.
Finally, program budgeting operates on the philosophy that an agency is responsible for determining the method of accomplishing the goals of the program; however, there will always be a tug and pull between the different branches of government. Legislators may choose to dictate specific policy methods as well as policy direction—including the providers, prices, or activities. This can be done for political reasons, previous mismanagement by the agency, or a disagreement about the best method to achieve the desired outcomes. It can also be done to increase the comfort with monitoring agency activities. Ideally, in so far as the legislators intervene in the details of management, legislators should also assume greater responsibility for results. A good example is the legislature dictating that a particular provider be used to produce a particular output. If the agency asserts that that particular provider is not the best choice, then the legislature should assume the responsibility for any negative results.
This article describes some of the criteria used by legislative staff to develop the DPH's program structure, and the decision tree shown in Figure 3 summarizes the process. Program definitions for budgeting purposes are founded on ideas about linkages between resources and results, but they must also accommodate concerns about clarity of resource-result relationships, accountability and control, practical problems, as well as political and strategic concerns. These legislative considerations lead legislative budgeters away from very broad programmatic frameworks such as the “Ten Essential Services” and toward more concrete definitions of programs based on the specific strategies.