What do government agencies do? It seems like an obvious question. A research agency writes grants that will lead to new scientific knowledge, a regulatory agency passes rules to keep us safe, and so forth.
But in fact it isn’t an obvious question. Agencies must have a mission, so what kinds of missions could an agency be given? Or conversely, we could begin with the many goals that Congress tells the federal government to pursue. How do these different goals get assigned to different agencies?
So you want to organize the government
An agency can be organized in several ways. Two of the most important1 are by functional category and by subject matter.
An individual agency might have jurisdiction over a functional concept. For example, the National Science Foundations covers the abstract concept of “research” by administering research grants. OSHA covers the abstract concept of “safety” by promulgating workplace safety regulations.
Accordingly, a cabinet level department might be organized into several functional bureaus. In this model, a department has one bureau that administers grants, another bureau that does research, a third that promulgates and enforces regulation, and so forth. USDA’s current org chart illustrates this approach: one bureau covers agricultural research and economics, another bureau covers regulatory programs (such as animal and plant inspection), another covers rural development, and yet another administers welfare programs (food stamps).
But alternatively, an individual agency might instead have jurisdiction over a concrete subject. One example would be USDA’s (now defunct) Bureau of Entomology: this bureau conducted research related to insects, it administered grants to farmers for protecting farms from insects, and it regulated diseased plants to prevent the spread of insects. It combined every function related to the subject of insects. It was therefore a subject matter agency.
In this model, a cabinet department might group its work into several subject matter bureaus. USDA’s past org chart could illustrate this approach. There was (among else) the forest service, a bureau of soils, a bureau of chemistry, and a bureau of animal industry – to say nothing of the bureau of entomology. Each of these subject matter bureaus would mix research, regulation, and grant administration.
The government has always had both types of agencies, both functional and subject matter. But the proportion has varied over time. In the early 1900s, the government was mainly organized by subject matter. Then from the 1930s through 1950s – the FDR through Eisenhower administrations – reformers were very much in favor of functional reorganization, and imposed this on the government. Today the government has relatively few subject matter bureaus, although some agencies such as the Forest Service have survived.
Subject matter agencies were once commonplace but are now rare. Did the US lose anything in this shift?
On the subject of subjects
There were several arguments against subject matter organization. It looked illogical. Cabinet secretaries sometimes were tasked beyond what a sensible2 span of control would call for. There were genuine issues with duplication of work and siloing generally. But nevertheless, there is a strong case for subject matter organization.
First, it led to greater technical expertise by offering an exciting career. The government has never been able to compete with industry on salaries. But if you were a skilled entomologist, say, the Bureau of Entomology offered a breadth of work that no private company could. It offered a chance to work on every aspect of entomology, from research to regulation.
Now consider agencies that focus purely on research (of many kinds) or purely on regulation (of many kinds). Fewer technical experts would be excited to work at such agencies – these missions inherently attract paper pushers.
Second, combining different functions moderated the worst impulses of any one category of employees. What are the failure modes of different types of civil servants? Regulators become aggressive and stop caring about the cost and practicality of their regulations. Researchers become academic and stop caring if their discoveries have any practical use. Grantmakers become captured by interest groups (i.e. their grant recipients), and stop caring about the taxpayer’s interest. We see all these failings in functionally-organized agencies.
So consider the benefit of combining functions – research, regulation, and administration – under a single roof. Scientists were guided toward practical research by their contacts in industry and ordinary civil society. Regulators could craft high quality regulations by relying upon the in-house expertise of researchers, and could craft realistic regulations through their contacts in private industry. Interest groups certainly demanded handouts, but the presence of technical experts gave agencies an esprit de corps3 that enabled them to push back.
Although academic reformers favored functional reorganization, the benefits of subject matter organization were appreciated by administrators at the time. The benefits were powerfully and clearly stated by an experienced bureaucrat with impeccable reformist credentials: the founder of the modern Forest Service, Gifford Pinchot.
Pinchot defends the Ancien Régime
One of the earliest proposals for executive branch reform was FDR’s Brownlow Committee. Like most midcentury reform proposals, it proposed reorganizing the government along functional lines. One specific proposal dealt with land use: it proposed separating applied conservation work from scientific research.
Pinchot delivered a speech4 attacking this proposal, which is worth quoting at length.
A certain committee, known as the Brownlow Committee, and containing in its membership a distinguished citizen of Chicago, has prepared a report on the reorganization of the government, which contains many excellent recommendations, and some even worse than the others are good.
The Brownlow Report constitutes, in my opinion, the most dangerous attack that has been made upon the national conservation policy since it was first laid before the people of the United States by Theodore Roosevelt, at the great conference of governors in 1908.
Now, for background. The Forest Service was located in the Department of Agriculture, as it is now. However, the Interior Department had long wanted to gain jurisdiction over forests, and many reformers were sympathetic.
The Forest Service administered the national forests and also conducted research on forestry, which they made available to foresters in the private-sector. The Brownlow report proposed a functional reorganization of government: it proposed that administration of federally owned lands should be in a proposed Department of Conservation, while research related to agriculture and forestry should be in the Department of Agriculture.
Pinchot goes on to explain this proposal, and strongly objects. In his view, there were significant benefits to having a subject-matter agency like the Forest Service, which concentrated on every aspect of forestry.
It recommends that Secretary Fall’s plan for the transfer of the national forests back to the Interior department shall be carried out. In addition, it recommends that the conservation work shall be dismembered and scattered between two departments.
[…]
The Brownlow Committee suggests renaming and camouflaging the Department of Interior under the name of Department of Conservation, whose duty it shall be […] “to administer the public lands, parks, territories, and reservations, and enforce the conservation laws with regard to public lands.’
The duty of what is to remain of the Department of Agriculture would be: […] “To conduct research, and the conservation and development of private lands.”
All this means, if the English language has any meaning, that, in the opinion of the Brownlow Committee, the forest on private land is one thing, and ought to be handled in one department, while the precisely similar forest on public land is another thing, and ought to be handled in another department.
[…]
In other words, the Brownlow Report proposes to divide up the natural resources with which the federal government deals, not on the basis of use and protection, which is the only sensible basis, but on the artificial and constantly changing basis of ownership.
[…]
To be specific, the Brownlow Report means that the Biological Survey, the Soil Conservation Service, the Forest Service, the management of pasture and range, and all the rest of the innumerable questions which concern the soil and growth from the soil, shall be split up in accordance with an artificial distinction which has no existence in nature—and all that under the sacred name of avoidance of duplication and putting similar subjects together.
In his view, functional reorganization was a triumph of logical niceties over practicality. He ultimately reaches a key argument, well worth considering today:
The proposals to transfer the national forests to the Conservation Department and leave forest research in the Agricultural Department, is bad administration of the very worst kind. If anything is proved in government work, it is that to separate administration and research means bad administration every time. As good an illustration as I know is the General Land Office in the Department of the Interior. Research it had none, and its mishandling of the public lands became a scandal, the stench of which is with us yet.
“To separate administration and research means bad administration every time”!
In Pinchot’s day, the federal government largely did not conduct pure research. Today, could we add that to separate administration and research means bad research every time?
Postscript
This discussion raises further questions about these two types of agencies. I aim to explore the following questions in the future:
Why and when, exactly, did the US transition from subject-matter to functional agencies?
How was scientific research conducted by subject-matter agencies? Was it less bureaucratic than the grant based system today?
How did subject-matter agencies manage and budget for projects? Did they have a greater product focus than the government has today?
Gulick’s standard classification has four ways to organize an agency. One of the other two is by recipient: the VA, for instance, provides different services to veterans. The other is by geography: the Appalachian Regional Commission, for instance, provides different services to Appalachia. (All of the types of organizations are my explanations and not the precise terms that he uses.)
Gulick, Luther Halsey. Papers on the Science of Administration. New York: The Rumford Press, 1937.
The Secretary of Agriculture had nearly 20 bureau chiefs reporting directly to him in the early 1900s. This is an extreme example.
The Forging of Bureaucratic Autonomy explains, for instance, how the highly trained Railway Rail Service was able to resist corruption and expand Post Office services. It covers also the early USDA’s resistance to interest group capture – also directly attributable to the highly trained and disciplined bureaus such as the Bureau of Chemistry. The subtitle of this article is a nod to the book.
Carpenter, Daniel. The forging of bureaucratic autonomy: Reputations, networks, and policy innovation in executive agencies, 1862-1928. Princeton University Press, 2002.
Stoeckeler, J.H. “Pinchot Says Reorganization Plan Wrecks Conservation Policy: Former Forest Chief Flays Brownlow Committee Report as to US Natural Resources.” Forestry News Digest, June 1937.
Do you have a taxonomy of currently existing functional vs. subject matter agencies? It seems to me that the federal bureaucracy currently exists as a mix of both (as you note with the Forrest Service) and it would be interesting to see a breakdown of the balance between the two ways of organization.