April 12, 2020

On Administration Markets

Governments, in the course of their normal operations, build up skilled bureaucracies to solve the problems they face. If they rented their bureaucratic capacity out to other states, I believe a wide variety of positive outcomes would result.

For example, one can imagine a state- or provincial- government with a bureau for issuing identity documents signing a contract to take over those responsibilities for other states, or a country deciding to rent out certain aspects of its legal infrastructure to its neighbors.1 Using this "G2G (Government-to-Government) Sales" model provides an efficient and tested service to one state, with little administrative overhead, and a new revenue stream to the other, offsetting tax revenue and boosting growth.

There's a kind of dogma that says that states are inherently less efficient than businesses, and there's something to the idea that states aren't often suited to competing head-to-head with the private sector—but there's not much evidence that it's the source of capital that affects efficiency. Rather—businesses are more often provided the incentives and competition that allow them to become near-perfect players in their spaces. By replicating a system of competitive dynamics within the state sector, it should be possible to approximate the dynamics that lead to efficiency in the private sector, without reproducing some of the more pernicious principle agent problems that contracting entails.

The trivial case for such a G2G model is that you get economies of scale, cost savings, and so forth, but there are some less-obvious positive outcomes that are worth expanding on.

Competition Allows Efficiency

The first is that you get a real sort of competitive pressure on the bureaucracy. Today, when someone in charge of the State Department of Road Maintenance does an excellent job, their department won't get any bigger. Instead, it'll likely get smaller as they succeed in saving costs, perversely costing them political leverage. On the flip side, if they do a terrible job, and costs balloon, there's no good option for the state other than trying to improve the organization. As a result, bad organizations don't die, and good organizations don't win. If an exceptional Department of Road Maintenance could take the contract for Delaware and New Jersey, and a bad one had to worry about getting replaced by Pennsylvania's, it's likely that road maintenance would get substantially more efficient.

A second is that you get an experimental basis for iterating on bureaucracies. In the US, we've long understood policy experimentation as part of the "laboratory of democracy," where sufficiently successful bills are bubbled up to a national level. Much of the current structure of the United States government, however, is located not in laws, but in an "Administrative State," where—for reasons of efficiency and expertise—laws establish regulation-making and -enforcing bodies responsible for specific issues and tasks, rather than establishing rules themselves. Despite this, public administration has become routinely ignored in the US in favor of public policy. (The problems of trying to solve such problems using an approach designed from the top down action rather than iteratively while on the ground should be clear to anyone who has read certain classics in the cognitive sciences).

While you can replicate a policy easily, replicating a bureaucracy isn't as easy. Administration markets provide a venue through which bureaucracies can be allowed to scale up and down, letting successful approaches spread more widely. As a result, we should be able to once again treat bureaucracies as a place in which wisdom can be refined through comparison, much as we do with laws today. 2

A Market Based Approach

Private sector operators may be nervous about administration markets, as they fear increasing the size of the state sector will crowd out and shrink the private sector. However, in a G2G model, profits in the state sector go to the state, allowing taxation to be decreased. In 1964 the Nobel Laureate James Meade demonstrated that growing the state sector in this manner can in fact allow the private sector to also increase, because the associated drop in taxation allows for an increase in private sector activity. We should not, therefore, consider administration markets to be “anti-market.”

In fact, I believe administration markets are in most cases the most suitable way to adapt market mechanics to solving government problems.

A previous attempt at a similar goal is are found in the "policy markets" suggested by Robin Hanson. However, a "policy market" is actually a betting market for speculating on the results of a policy, not a market for policies themselves. While betting markets may provide better information on expected outcomes under certain conditions, nobody has yet found a high-quality way to include them in the policy-making processes. And finally, policy markets continue the emphasis on policy over administration.

You can also contrast the "administration market" approach I'm advocating with the sort of faux-privatizations championed in the West during the 80s to mid-00s, in which an inefficient unitary bureaucracy is replaced by an inefficient, unitary, unaccountable, privately-owned bureaucracy. I don't mean to say that privatization is never an appropriate solution; only that it's unlikely to succeed in certain cases, such as:

Where competition or benchmarking is impossible either across regions or time, making the vendor no better than a unitary state bureaucracy

Or where it's intractable for efficiency reasons, as

The list of actual failures in such privatizations is numerous and well documented, and I think that for some of these two cases, the proper choice is to allow competition between agencies in nearby states.

Logistics of Administration Markets

Administration markets present obvious logistical challenges—ensuring the bureaucracy is neutral between the parties it serves, and harmonizing standards to allow cross-state work—but this isn't wildly different than the logistical challenges behind a privatization. Routine commerce relies on a wide variety of theoretically-insoluble but practically-solvable agreements—trade deals, shared dispute resolution mechanisms, and so forth. I believe an administration market's challenges should be qualitatively similar, and that the benefits should be large enough that all parties will be incentivized to find solutions.

I write as if this is a novel approach, but there are many great successes that already follow similar logics. The state of Delaware, for example, is in large part funded by its role as the efficient legal arena of choice for a wide variety of companies—hence, the "Delaware Corporation." Many standards are already outsourced—local equivalents outsource their drug certification to the FDA, or their airplane certification to the FAA, by allowing reciprocal approvals. It is merely an expansion of this logic to proceed fully to government-to-government sales of administrative capacity that I am arguing for.

The obviousness of such an expansion begs the question of why it hasn't happened much yet. I submit only that, due to the greater risks involved, governments evolve at a much slower rate than corporations, and so innovations that allowed distributed workforces in the corporate world decades ago are only now being adopted in earnest by governments.

Further, while I've focused on the case for agglomerating local bureaucracies into larger ones, there are likely similar benefits going the other way—when a job is too big for one state, it doesn't always need to be forced into total centralization. Many bureaucracies that today exist on a unitary federal- or international- level could easily instead be several competing regimes operating under a membership model. For example—if the US Department of Education, instead of switching hands every between administrations and suffering ongoing bureaucratic whiplash, instead consisted of several organizations arranged by the states, which individual governors could enter their states into or leave on an ongoing basis, we would surely have better side-by-side data on outcomes.

Administration Markets and Talent

One of the clearest benefits in scaling up bureaucracies and making them more dynamic is that it makes them more attractive workplaces for skilled talent. An organization whose leadership faces competitive winds is less likely to turn to petty tyranny, and allowing skilled practitioners the ability to share their expertise beyond their local area should be broadly appealing.

As globalization has progressed, local governance has suffered from an evaporative cooling effect, as bright and ambitious people have increasingly moved to large cities and centers of industry. This has facilitated decades of massive economic growth, but it's also exacerbated the never-stellar record of local governments, especially in smaller towns. While some reactionary localists argue the only fix is to revert to a more local world, there would be a huge human cost.

Instead, administration markets should make it possible to agglomerate and spread local agencies over wider areas. Much as SaaS startups have reduced the headcount necessary to start a company, a semi-delocalization of local government could reduce the total number of people necessary to run a town while increasing its capacity for governance and its conformance to best practices. Given the small number of potential candidates for office in any small town, shrinking the number of local seats would increase responsiveness, as it would become easier to vote people out. Dissatisfaction with individual agencies would be translated into a supplier decision, rather than an extended saga of bureaucratic reform. And finally, towns with particularly good institutions (which surely exist) would then be able to extend those further afield, increasing revenue and retention.

COVID-19 and the Case for Administration

When I wrote a first draft of this piece in the Summer of 2019, my belief was that there was a general crisis in the administrative capacity of the American state, and likely similar ones across much of the West. This is not a novel thesis, but it has been somewhat fringe for some time.

The actual response of the US government to recent circumstances has been so laughably bad that I see no need to summarize it. I had expected that while the state was largely in autopilot, it was capable of mobilizing to a competent wartime governance in a case of emergency. This was wrong.

If there is hope for an improvement in the state's condition, it will involve more thought towards the question of building a robust American administrative state. The US has been angst-ridden about the role of its bureaucracies for centuries due to distinctively American concerns about their lack of an effective theory of representation, the courtly character of bureaucracy, and efficiency. This has resulted in an emaciated state that struggles to perform its basic duties. My belief is that whether under the banner of administration markets or G2G sales, the solutions I've outlined are distinctively suited to addressing those concerns.

I've tried to enumerate the many immediate ways administration markets strengthen the administrative state—among them, by increasing efficiency, producing economies of scale, promoting alignment of interest in contracting, enabling competition where it would otherwise be impossible, providing an experimental basis for bureaucracy, making possible changes in the scale of governance, and facilitating the attraction of high quality talent. The form however, strikes me as a sort of structural organizational change that arises only very seldom, and leads to major—and likely positive—changes in the structure of a society.

  1. While my examples focus mostly on an American context, I think this is a beneficial model both sub-nationally and internationally. ↩︎

  2. It's worth comparing this approach to the theoretical basis for experimental governance used in China. While actual practice varies widely, an uncritical textbook account might go something like this: “In the Chinese system, legalistic policy is largely subservient to the bureaucratic entities enforcing it, rather than the reverse. While bureaucracies themselves don't scale up, the set of goals for different provinces' local governments are centrally set, and personal advancement for leading cadres are tied to realizing them. Benchmarking is used as a substitute for true competition, and profits are returned as political profits, rather than monetary ones—but the principle of using competition between peer states to encourage bureaucratic performance. Exemplary bureaucratic performances are documented by the central government, and promulgated to peer provinces, building a widespread knowledge of functional practices.” ↩︎