Toward a Science of Program Design (1977)

Toward a Science of Program Design

Robert A. Knisely
Far Muse
Mathias, WV 26812

(Originally prepared for delivery at the ORSA-TIMS meeting
in Atlanta, Georgia, on November 9th, 1977)


President Carter is correct in saying that the Federal government needs an overhaul, but too much of his current reorganization effort is focused on how the pieces fit together, and not enough on how the pieces work. A major part of his reorganization effort must be directed toward the redesign of individual programs, and the development of a more scientific way of designing programs in the first place.

Program design and redesign depend upon two things: first, an understanding (which we do not have) of how programs work, and second, a set of guidelines (which does not exist) for use in designing new programs. Given the explosion of governmental responsibilities and new programs in the past two decades, it is not surprising that there is no science of program design. It is time to start one.

Three Principles in Program Design

Three fundamental principles must guide any program design or redesign efforts for the Federal government. First, the people of the United States are not affected by departments such as Health, Education, and Welfare (HEW), nor by their constituent agencies such as the National Institutes of Mental Health (NIMH). People are only affected by programs, such as the Community Mental Health Centers program in NIMH within HEW. Departments and agencies are only an administrative convenience; they may provide centralized direction, planning, budgeting, evaluation, and so forth, but they have no direct contact with the problem. Therefore, the efficiency and effectiveness of each program is of primary concern. How programs fit into departments is not very important. For example, the departments of Interior and Commerce have been feuding for years about which should have jurisdiction over deep seabed mining, Interior’s Bureau of Mines or Commerce’s National Oceanic and Atmospheric Administration (NOAA). The two agencies have claims that are equally logical, but the jurisdictional squabble has certainly delayed the implementation of any program.

The second fundamental principle is that real improvements in efficiency (and probably in effectiveness) can only be achieved through substantial changes in the machinery of individual programs. Industrial productivity rises almost exclusively because of improved machinery and not through exhortation or exploitation of the workforce. Service industries, including government, are no different. The design and recurring redesign of the machinery of programs are the keys to more productive government. For example, Commerce’s Economic Development Administration (EDA) in all of fiscal year 1976 obligated $146 million for public works under its Title I program. In the last half of 1976 EDA funded $2 billion worth of public works under its new Local Public Works (LPW) Program, with no more than twice as much effort. The types of projects funded are quite similar. The difference is that the LPW program solicited applications nationwide and processed 25,000 of them in two months via computer, while the Title I program relies on applications that are “hand-crafted” over a period of years. The Congress and President Carter liked the LPW machinery well enough to add another $4 billion to the program in May of 1977.

The third fundamental principle is that the costs of government to the people, in terms of time, money, and aggravation are nothing but overhead. They do not add to the real productivity of this or any other nation. Like hospitals and law courts, government is a necessary evil, however necessary it may be.

Governmental intervention into the lives of people is therefore better the simpler, more precise, and cheaper it can be. Individual programs constitute the interventions, and it is the programs that must be designed as sparely as possible yet accomplish their goals.

The Federal Deposit Insurance Corporation’s best known program to prevent bank failures is a model of simplicity and precision: since everyone knows that the FDIC insures their bank deposits, “runs” on banks and the ensuing bank failures are prevented; the low cost of the insurance reflects the low number of bank failures. It is a beautiful program.

An Art, not a Science

Unfortunately, the design of Federal programs is not a skilled art, much less a science. It is an art practiced almost exclusively by civil servants and lawyers who work for the Congress, the executive branch, and the various lobby groups. Their attention is always focused on the problem at hand, and the government has grown like Topsy. No one in Washington compares programs across agency and department lines in terms of what they do and how they do it. We do not even have a language with which to compare them. For example, the Office of Revenue Sharing (ORS) in Treasury had 100 employees and disbursed $6.4 billion in fiscal 1976; in Commerce’s EDA, the Office of Technical Assistance (OTA) had 62 employees and gave out $11 million in the same period. Is ORS 355 times better than OTA because it disbursed 355 times as many dollars per employee? Of course not; the very question seems unfair. The only rejoinder, however, is that ORS and OTA are designed to different things differently; this is not helpful. If it is not fair to compare OTA to ORS, what programs are so similar to OTA that a comparison of, say, administrative costs and disbursements per employee would be fair? What does “similar” mean in the context of specific Federal programs?

Understanding How Programs Work

One must understand the programs that already exist in order to move program design from a rather primitive art form toward a science. One must begin with a list of today’s Federal programs, then compile a list of the generic types of programs, and finally sort that list into the types of design components that make up similar programs. In March of this year (1977) Senator Muskie’s Subcommittee on Intergovernmental Relations published the very first list of all Federal programs. This Table of Federal Programs lists over 2000 separate programs, was in preparation for over a year, and does not contain changes that occurred in 1976. Still, it is a start.

The list of program types is probably both finite and manageable, although it can only be developed via a thorough review and categorization of the programs in Senator Muskie’s list. The Federal government builds things, either directly or indirectly; gives assistance with more or fewer strings attached to governments, organizations, and individuals; promotes the use or conservation of resources; regulates both individual and corporate behavior; and so forth. Each category can be further broken down: different groups in America are given different forms of assistance, including advice (farmers), loans (small businesses), loan guarantees (Lockheed, among many others), grants (hospitals and universities), contracts (the construction industry) and tax preferences (almost everyone).

The program components involved vary by generic program type. For a grant program the components would include the following: regulations, guidelines, application forms, internal procedures, the management structure, review processes, field staff, external contracts, planning and evaluation requirements, audit procedures, and so one. The lists of components for all program types is finite and can be organized, although it is extensive to say the least.

Once similar programs have been disaggregated to their components, the efficiency and effectiveness of the components can be compared with each other. This is a very manageable task, in sharp contrast to a comparison, say, of revenue sharing with technical assistance.

From an understanding of how the same components function in programs of the same type (regardless of agency or departmental location), one can make specific recommendations for change, requiring desirable aspects of one program be in another, such as “every program granting money to public bodies upon application must require timely local publication of the fact and content of each application”. Further, one can seek to improve the program components, and hence programs, in terms of a set of specific indicators, such as applicant expense, processing time, staff time involved in processing, potential for arbitrary decision-making, accountability, or whatever.

In time, model components would be developed for each program type, and the task of creating new programs, and of changing old programs to reflect new conditions, would be greatly simplified.

Developing Design Criteria

Significant but not dramatic increases in efficiency and effectiveness of government programs will occur through “tidying up” existing programs after the analysis and comparison of similar programs and their components described above. Dramatic change will only occur when more efficient program machinery is chosen to accomplish the same objectives. For example, if a direct loan program can be changed to a loan guarantee program, the private sector’s interest in an accurate determination of each applicant’s credit-worthiness may tend to keep the program in touch with current economic realities. Less government time and money will go into loan processing, and perhaps the paperwork burden on the applicant will be reduced.

Where more significant savings are sought, the process of inquiry must begin with the problem to be addressed, not with the existing program or programs designed in the past around similar problems. There is an apocryphal story in Washington of a young attorney working on Capitol Hill some years ago who was asked to draft a bill to regulate some aspect of the airlines industry. He started by looking up and copying the law that addressed similar concerns with the railroads. Seeing some provisions that did not seem applicable to railroads, he did more research and discovered that the railroad bill had been lifted 75 years earlier by another attorney from a law passed early in the 19th century to regulate the canals. Justice Oliver Wendell Holmes said “in the law, an ounce of history is worth a pound of logic”. In program design, we need the pound of logic.

I have collected a few of the basic questions that program designers should ask themselves; there are undoubtedly many more. To start with:

(1) How good and how recent is our understanding of the problem or problems to be addressed? If we are not sure what we are trying to do, or if our program is aimed at the symptoms and not the causes of the problem, we won’t get very far. For example, the Office of Minority Business Enterprise (OMBE), started in 1972 in Commerce, has for years funded Business Development Organizations (BDOs) whose mission is to help people in minority communities start small businesses. The failure rate for all small businesses is 90% in the first year, so even if the BDOs’ success rate were three times the national average (which it has not been), they would still be wasting 70% of their resources.

Seeing the odds against this intervention strategy, OMBE recently started de-emphasizing the BDOs in favor of Minority Purchasing Councils that attempt to get contracts for minority-owned businesses that have survived the first several years. Although this new program strategy seems a big improvement, OMBE still does not know why there is so little minority entrepreneurship. Do the minority communities lack a history of entrepreneurship? Role models? Access to capital? Access to emerging technologies? People who can afford to take a gamble on a new business? People with managerial experience? Or has entrepreneurship changed so greatly since America’s major corporations were formed that there is no room left for anyone new — whether black, white, grizzly, or gray? Programs should not be designed around hunches or one person’s experience; they should be designed around the best problem knowledge we can get. Programs are expensive.

2. How is the problem addressed by other organizations, whether Federal departments, state governments, other countries or international organizations, or even corporations? Here the understanding of the program types outlined above would be very useful. For example, not all countries that grant patents do so after an examination of the novelty and merit of each invention. Several western countries merely record and file the inventor’s claim and the date and time it reached the patent office. In America about a thousand patent examiners, all of whom are scientists and one third of whom are also attorneys, decide whether ideas deserve patents. Our patent process leaves the inventor uncertain about the protection his idea will get for at least 19 months, keeps the thousand government patent examiners reviewing the work of others rather than inventing, supports the private patent bar handsomely, and costs a lot of money. It also does not provide very good protection for the inventor: Each patent that is challenged goes through a de novo review by law clerks and judges, and in about half the cases that reach the U. S. Courts of Appeals the patent is overturned. Choosing the winning side 50% of the time can be done blindfolded.

It has been said that the uncertainty of the 19 month “patent pending” period is a contributing factor to America’s relative decline in innovation during the past few years. Business does not like to invest in uncertainty. We are told that in some fields corporations are keeping new ideas as trade secrets and not patenting them at all. An increase in trade secrets is the best measure of a patent system’s failure.

While granting that the patent registration system may not be as suitable to America as our patent examination system, certainly the various patent systems should be systematically compared every so often. Perhaps we should adopt the system used in the Netherlands, where examination is done even more thoroughly than in America. The point is, without comparison, who knows?

(3) What known program design options, and what conceivable program design options can we think of? This step requires both research and imagination. The bases for the research are described above, but the imagination must come from a number of creative people with varied backgrounds — not only is a lawyer likely to create a different program than a civil engineer, but a civil servant in one department will begin the design process from his or her own experience, which will be quite different from that of someone from another department. For example, the goal of more energy-efficient household appliances might be sought through tax incentives, voluntary industry standards, mandatory industry standards, or consumer labeling, depending on whether the program designer is in Treasury, Commerce, the Federal Energy Administration, or the proposed new Agency for Consumer Protection. Another possibility, which like several of the above interventions is now underway, is to convince America’s largest purchaser of household appliances to make buying decisions based on life-cycle costing (purchase price plus operating and maintenance costs) rather than purchase price alone. If a manufacturer is guaranteed one very large contract, he can afford to design and manufacture a more energy-efficient appliance, and sell part of his production directly to consumers. Since the Federal government’s General Services Administration (GSA) is by far the largest purchaser of most household appliances, an innovative little program at the National Bureau of Standards called ETIP (the Experimental Technological Incentive Program) has been working with GSA to see whether large government purchases can pull better technology out of producers.

(4) What are the probable consequences, both intended and unintended, of the various program design options? People designing programs are seldom aware of the vast range of consequences that even a small program can have. Multi-disciplinary design teams, knowing the designs that have produced positive or at least hopeful results in the past, should spend time hypothesizing unintended consequences. A famous example is the dramatic increase in schistosomiasis in Egypt that followed the construction of the Aswan Dam. The dam provided the ideal environment for the snail that carries this debilitating disease. An example closer to home was the surprising preference of the local powers in some unreconstructed rural Southern counties for the food stamp program over the old commodities distribution program. Most people in the North thought the food stamps far more liberal and humane. In 1968, while travelling in upstate Louisiana for the Office of Economic Opportunity, I found out that the food stamp program (a) enriched the local storekeepers, (b) permitted both the hiring of white clerks and the firing of Black warehousemen, and (c) was driving the poorest of the poor out of the state. They could not afford the minimum cost of the food stamps, and their source of commodities was shut off. I dare say that few of these consequences were uppermost in the minds of the major sponsors of the food stamp legislation.

Benefits of a Science of Program Design

Here are a few of the conclusions that might emerge from an analysis of program design independent of subject matter:

(1) Loan guarantee programs should be preferred over direct loan programs, as mentioned above.

(2) Tax incentives can induce behavior that would be almost impossible to coerce bureaucratically. An example is the recent tax incentive that allows a write off for added construction costs that make a building accessible to the physically handicapped.

(3) Adding relevant requirements as conditions of large existing programs is a very simple and precise mechanism. This was used to eliminate billboard along interstate highways.

(4) Precision targeting and Federal control may have to be sacrificed so that a program will have immediate impact, as in the case of the two public works programs discussed earlier.

5. A program aimed at one set of problems may learn from the history of a similar program or programs aimed at another set of problems. America is presently beginning to exploit systematically the living resources of the sea. Present day exploitation is done by fishermen in small vessels with little government aid or counsel. Today’s mariculture is in striking parallel with the agriculture of the 1880’s and 1890’s when governmental intervention in farming began in earnest. Much has been written about the effects of these programs on today’s agriculture. Unless we understand what happened to create agribusiness and then apply the lessons learned, we are condemned to create maribusiness, to paraphrase Santayana. If we recognize that categorical grants, block grants, special revenue sharing, and general revenue sharing are only points on a continuum of Federal control over the expenditure of Federally collected revenues, we may be able to tailor programs to our perception of the recipients’ readiness for responsibility, and not solely to the fashion of the day. We may even find other points on the continuum.

Next Steps

The development of a science of program design could be undertaken either within government or by a university or other institution. If it is done within the government, access to information will be easier and the experience gained in doing the research will be available for applying the results. The project should be housed in the Office of Management and Budget (OMB) because only OMB has a government-wide perspective, only OMB can set standards for program design components, and because OMB is now a major participant in program design as an art form.

What is needed is a small office that will perform the analysis of program structure, review and comment on the design of new programs, and oversee the review of the design of older programs as an adjunct to President Carter’s reorganization effort.


The ultimate choice among program design options is a political one, not very susceptible to the rational process. It would be encouraging, however, to see the thumb of rationality weigh more heavily on the scales of decision-making. Decisions cannot help but be improved with the knowledge gained from a comparative study of program types, their strengths and weaknesses, and their past successes and failures. When Federal programs are being designed, there should be room at the table for someone familiar with many program designs, for as the psychologist Abraham Maslow once said, “if the only tool you have is a hammer, you tend to see every problem as a nail”.