State capacity at scale

One Weird Trick for public sector efficiency

Today is day 10 of the 31 Days of Policy. I recommend reading the intro post here if you haven’t already.

In recent years a number of reformers have turned their attention to the comparatively high costs of public sector goods in the US. Jennifer Pahlka, the former CTO for the US Digital Service under the Obama administration, has been banging the drum of improving public sector efficiency for years. The NYU Marron Institute’s Transit Costs Project has highlighted how outrageously expensive new transit lines are in the US, and a number of blogs have sprouted up to investigate why we pay more for less on nearly everything compared to other OECD countries. These good government obsessives have christened themselves the state capacity movement.

Despite the consensus understanding of the problem, there is a dearth of actionable solutions at hand¹. Unlike the YIMBY movement in housing there are no silver bullet reforms that activists can put in front of legislators and yell “Pass this!”. The issue is likely to involve a lot of “slow boring of hard boards” without quick payoff. Reformers shouldn’t be dissuaded, though, as there is at least one way we can improve public administration that doesn’t require clear-cutting the thicket of poor governance incentives as a prerequisite.

It’s a big club…

We’ve previously discussed the role of government in resolving collective action problems but we can generalize this further through the concept of a public good. A public good is any good which is non-excludable (i.e. you can’t prevent someone from using the good) and non-rivalrous (i.e. one person’s consumption of the good doesn’t reduce the amount available for others). Classic examples of public goods include national defense, public health, and rule of law.

Public goods are underprovisioned by private actors because of the free rider problem. Providing for national defense is expensive. If payment for the military was entirely voluntary, there would be a strong incentive to free-ride off the payment of others. Free riding eventually erodes the ability to provide public goods, leading to market failure. The state, with its monopoly on violence, can suppress free riding by forcing all citizens to contribute to public good provision through taxes. Ideally taxes would be based on the marginal benefit that each citizen derives from the public good, and there are a few theoretical mechanisms for achieving that. In practice the price paid through taxes will only imperfectly match each consumer’s marginal benefit.

The theory of public goods is well-developed but much of what a government provides aren’t technically public goods. Many government services are actually club goods. A club good is non-rivalrous (or relatively so) but excludable. Highways are a club good. It’s excludable as you can exclude drivers from accessing it through toll booths. At rush hour the highway is very much rivalrous, as the addition of one user creates congestion for the others, but during most times of day your use of the highway is non-rival.

Club goods with perfect excludability² don’t suffer from free rider issues and are frequently provisioned by the private sector, so the case for government provision is not as obvious. Indeed some states have privately built and operated highways that are operationally profitable. There is no brightline standard for when the public sector should get involved in club goods but there are a few commonalities.

Firstly, if the good has increasing returns to scale, as can happen with strong network effects, then it will be more productively efficient for the good to be produced by one firm. This is a natural monopoly. As discussed in our post on the FTC, monopolies cause problems for allocative efficiency, which in turn justifies some level of public sector involvement. Public utilities managing water, electricity transmission, natural gas, etc. are the paradigmatic example of this.

Secondly, some club goods are vulnerable to hold-up problems. When negotiating over a highly specific asset needed for production (like a piece of land along the path of a high-speed rail line) the asset owner is incentivized to hold out for a much higher price than the asset would actually be worth. The Myerson-Satterthwaite theorem proves that a bilateral trade with one seller and one buyer will in general result in inefficient outcomes, so a land owner holding-out for a higher price on a highly specific asset is not subject to normal market efficiency assumptions. The government has the ability to resolve this issue in favor of the transaction with eminent domain. This is why transportation infrastructure, which technically could be privately built, is usually managed by the public sector.

Finally, some club goods generate positive externalities. We’ll discuss externalities in more detail tomorrow but the gist is that a positive externality is a spillover benefit that the producing entity can’t internalize. As such there is a tendency to underproduce goods with positive externalities. An example would be the statistical data gathered by agencies like the Bureau of Labor Statistics. That data generates benefits many multiples in excess of the benefits that accrue to the government directly. In theory the government could just pay to procure such goods from private firms instead of producing it internally but contracting is not always effective for goods where quality is hard to ascertain or it’s difficult to monitor compliance.

Unlike with public goods, there is no optimality condition for government provision of club goods. If the private sector can provide a club good efficiently then government provision is not ideal as it can crowd out private investment. Nonetheless economists are fairly united in the belief that certain club goods are better served through the public sector than the private sector.

Government, bit by bit

One incredibly valuable club good that governments provide is identification cards. In the US no centralized ID card was ever developed at the federal level so we use a kludge solution of our (highly insecure) Social Security number combined with other indicia like driver’s licenses or birth certificates to serve the same function. A shocking amount of bureaucratic resources in this country goes towards identity verification.

It doesn’t have to be this way. After the fall of the Soviet Union, Estonia decided to redesign their government identification system from the ground up. They created a digital ID card for every resident and developed an open source data exchange layer called X-Road to facilitate interoperability in data transfer. The benefits of Estonia’s e-Government system are mind-boggling: filing taxes takes only a few minutes, voting can be done online with biometric verification, cash transfers are instant, social services are easy to access, transaction costs and acquiring permits are faster across the board, and much more. Citizens have personalized control over their data privacy, corruption is much lower than peer countries, and many tedious paper-pushing roles have been automated away, allowing Estonia to sustain a low tax-to-GDP ratio relative to their public service quality.

The economic impact of Estonia’s digital government is around 2% of GDP. Countries like India are taking notice and are now implementing similar systems as they develop. Since network effects are superadditive, a bigger and richer country like the US stands to gain even more than Estonia. A McKinsey report from a few years ago estimated that the potential benefit to the US would be a staggering 4% GDP boost.

The decentralized nature of the US government makes a single national ID tricky, particularly as it intersects with the immigration debate, but bottom-up reform at the state level is possible. Many states are rolling out digital driver’s licenses with some effort to integrate it with other state-level services. The experience of New South Wales in Australia is precedent that a digital ID at the subnational level can succeed³. The path to a fully electronic bureaucracy in the US likely runs through our laboratories of democracy.

A digital ID with its associated infrastructure is a slam dunk state capacity reform. It has strong network effects, the government can use its substantial power to require firms to adopt its usage, and the positive spillover benefits for the private sector are immense. State capacity reformers looking for an easy win would be well served to pursue it.

1. One issue is that public sector collective bargaining in the US is enterprise-level rather than sectoral as it is in Europe. This makes public sector labor contracts uniquely vulnerable to local rent-seeking. As a result state and local governments farm out internal capacity to nonprofits and contractors to circumvent high FTE costs. This is a tricky political economy issue that state capacity reformers have failed to crack.

2. Many club goods suffer from some degree of free-riding due to perfect excludability being costly, e.g. Netflix password sharing.

3. Australia in general is underrated from a public policy standpoint. They tend to fly under the radar compared to other rich countries.