An Underpants Gnomes theory of the world

I was reading this article on Vox by Ezra Klein and I was absolutely floored by this line:

A lot of pundits…have a sort of Underpants Gnomes theory of Marco Rubio’s chances. Step one is Rubio is the only acceptable nominee to Republican elites. Step two is ... something. And step three is Rubio wins the nomination.
Why one political scientist thinks Donald Trump might actually win, Ezra Klein

If you’re not familiar with the Underpants Gnomes, they were invented by the creators of South Park, and appeared in the episode Gnomes; which is helpfully summarised on the Wikipedia page.

In the episode the Underpants Gnomes have a plan to make profit which consists of three phases, set out in the screenshot below:

"Gnomes plan". Via Wikipedia -

"Gnomes plan". Via Wikipedia -

If you can't see the image above it says:

Phase 1: Collect underpants

Phase 2: ?

Phase 3: Profit

Klein’s usage in his Vox article was the first time I had seen reference to an Underpants Gnomes theory of something, but, apparently its use in analysis of policy and political economy goes back quite a way. Forbes has a nice article that explains its applications to political economy.

Basically, the Underpants Gnomes have a view of the world that links the collection of underpants to profit via some mechanism that they can’t explain, and likely doesn’t exist. When I saw Klein use it, I was struck by how powerful and widely applicable the idea can be.

For example, in Indonesian infrastructure you see this kind of thinking.

Phase 1: Make a speech about how Indonesia needs USD 450 billion of infrastructure investment between 2015-2019.

Phase 2: ?

Phase 3: Indonesia gets all the infrastructure investment it needs

We have been in Phase 1 for the last three years, and haven’t yet seen any evidence that infrastructure investment is accelerating at the rate needed.

Another example could be from PPP.

Phase 1: Create PPP policy framework

Phase 2: ?

Phase 3: Indonesia has dozens of PPP projects

We have been in various stages of Phase 1 for a decade now, give or take…

I certainly don’t want to give the impression that Phase 2 in either of the above plans is easy! Most of my career has been spent trying to figure out exactly what they involve (after a decade, I do have some partial theories…).

The general point is that a plan needs to do more than identify a problem, and a desired outcome. To really effect change, you have to follow the thinking process all the way through.

I don’t know about you, but I plan to add this idea to my policy analysis toolkit! Now I just have to think about how I translate it into Indonesian for the first time I pull it out in a meeting with my government counterparts…

What is the government's job in infrastructure?

The government’s job in infrastructure is to ensure the right infrastructure is delivered by the right party at the lowest cost.

I chose each of the underlined words quite carefully, so in the rest of the article I'm going to explain what I mean by each. 

The right infrastructure

The right infrastructure means that it is the right stuff in the right places and built to the right design standards. By right, I don’t mean some sort of “international best practice” right, I mean right according to the society that the government represents. 

If the society wants a road, and they want it enough to forgo the other things that they might otherwise spend the money on, then the government should give them a road that meets their needs (subject to their budget constraint).

While the road should meet society’s needs, it’s also important that it’s not over-engineered, or “gold plated.” In Australia that might mean the seats on a bus will be padded, and wide enough for an average Australian rear end, while in Indonesia, it might mean they are hard plastic and wide enough for an average Indonesian rear end. Giving everyone first-class aeroplane seats on a commuter bus might be comfortable for the passengers, but it most likely wouldn’t represent the best use of society’s resources.


Societies are made of up people with different preferences, and some might want fancy chairs, while others would prefer a cheaper solution and are happy for everyone to ride on hard wooden benches to get it. Picking a level of service is always going to upset someone, but generally, when governments do it, they are trying to achieve some sort of middle ground that makes the least number of people (or the least number of people they care about) unhappy*. That’s generally as close as you can get to the ideal of the right infrastructure.

The right party

In a modern, open economy, people get their services through a range of delivery modalities, including:

  • fully government, in which the government is wholly responsible for the delivery of the service. Imagine a government building a road with equipment and materials it owns directly using its own employees. A more common example would be the core functions of the military.
  • government contracted, in which the government contracts for the delivery of the services. This could be a government hiring a contractor to improve a road on a 6 month contract (a common type is an EPC contract), or one to build, operate then transfer a port as part of a PPP on a 50 year concession.
  • government regulated, in which the government doesn’t own or contract anything directly, but generally oversees the provider, or providers of the service. This encompasses a huge range of models. Some examples are below:
    • Heavily regulated: A fairly closely regulated example is in the New Zealand power sector where the Electricity Authority creates the market rules, enforces them, and monitors performance. 
    • Loosely regulated: Businesses like management consulting firms are generally quite loosely regulated. They still need to comply with laws and regulations surrounding the general practice of business, hiring, firing, workplace health and safety, taxes and so on, but there aren’t many specific regulations targeting the industry. Note that this can vary depending on the kind of advice being provided, for example, tax and financial advisers have more stringent guidelines than strategy advisors (or economists).
    • (almost) no regulation: Businesses like kaki lima (roaming street food vendors) in countries like Indonesia. In Indonesia, technically they’re meant to have licenses issued by the local Camat, but many don’t. Even if they do get licenses, they’re still exempt from many other regulations by virtue of their size. And, even if they’re not exempt, in practice, no one is going to check whether a kaki lima vendor is paying her employees properly**. Even so, it's not strictly accurate to say that they are not regulated.
A lightly regulated industry  Image source:  Wikimedia

A lightly regulated industry

Image source: Wikimedia

These delivery mechanisms are not mutually exclusive. There are plenty of industries where the government uses all three delivery mechanisms. In Australia, local councils own childcare centres, contract private parties to deliver childcare services, and oversee compliance with regulations for fully private providers. In Indonesia the government owns and operates hotels, and regulates private providers within the hotel industry.

What I’m getting at by describing all of these delivery mechanisms is that, in most cases, it shouldn’t matter who it is that provides the service, as long as it gets provided well. Governments should look at all the pros and cons of all of the different delivery models for the service they need and make a choice on that basis.

The lowest cost

By “the lowest cost” I mean that I don’t want to pay any more than I absolutely have to for my right infrastructure***. When I talk about paying, I mean either directly through user charges, or indirectly through government spending money that they might spend on other things that I want for society.

If the government is operating a water network and the private sector could do it to the same or better standards at a lower cost, then I want government to either match that cost, beat it, or hand the project on to the private sector and act as regulator.

Similarly, if a private operator of an airport is abusing their market power to charge exorbitant parking fees, then I want the government to step in to penalise them for it.

When contemplating engaging the private sector in the delivery of infrastructure services, it’s generally considered good practice to develop what’s called a public sector comparator (PSC). A PSC compares the cost of public provision with the cost of private provision, taking into account the upfront and ongoing direct financial costs, economic costs or benefits, risks that could be transferred or retained, and competitive neutrality (which I’ll come back to in a later post). If a private option is cheaper than the government provision, then you go ahead with the private option, if not, stick with government.


So, the governments job is to give us the right services, cheaply, by whatever means creates the most value to society. 

The process that governments go through to determine that they are ensuring the delivery of the right infrastructure at the lowest cost is not a linear one. You can’t just go through and try to answer the question of who should be delivering in isolation of the design and cost, and vice versa. In the delivery of a major project, there are typically years of work going through multiple cost estimates, evaluating different delivery mechanisms and engineering solutions, all feeding in to the coordination process that, ideally, produces an outcome in line with what we want from our governments.

An interesting approach to making the process of delivering infrastructure is more transparent has been pioneered by the Department of Treasury and Finance in the State of Victoria, Australia and is called the Investment Management Standard. If you’re interested, you can read about it more here.

*Considerations of access are an example where it's generally considered good practice to not weight all individuals' preferences equally. Ensuring wheelchair accessibility for things like public transport is a significant additional cost and is only required by a tiny minority of the population. However, as societies get richer, they increasingly feel like this is an appropriate cost burden for society to carry for the sake of this minority.

**It’s not all good by any means. The lack of regulation is accompanied by a lack of protections as well. Informal sector actors have little recourse if they are shaken down by local criminals (or government representatives), or if they get hit by a car, etc.

***My desire for low costs is, of course, in direct conflict with my desire for good services. The best outcome will be an optimisation of both. 

What I talk about when I talk about infrastructure

"Infra-"  means "below" and "-structure" means "structure". So when people talk about infrastructure, they are usually talking about the structures that allow things to operate.

Confusingly, people use the word "infrastructure" for all sorts of purposes. IT infrastructure refers to the "combined set of hardware, software, networks, facilities, etc. (including all of the information technology), in order to develop, test, deliver, monitor, control or support IT services". The World Bank has a project where they talk about "judicial infrastructure" being both the physical buildings like courts and offices where the justice system operates, but also the institutions, policies and procedures that allow it to operate. Some people even talk about "emotional infrastructure" of a family or organisation being the systems that allow people to make their feelings known and have them validated. 

When I (and lots of other people) just say "infrastructure" without further clarifying, we are generally talking about the structures that underly the real economy, and allow it to operate smoothly. This sort of infrastructure is usually differentiated along two different axes: hard to soft, and economic to social.

"Hard" infrastructure usually refers to the "crunchy stuff", roads, buildings, physical assets. While "soft" infrastructure refers to institutions, regulations and systems. "Economic*" infrastructure usually refers to things like roads, power systems, ports, while "social" infrastructure refers to schools and education systems, healthcare, sporting facilities, museums and so on.

Below I have sketched out a table showing a very subjective and not-intended-to-be-perfect categorisation of different kinds of infrastructure according to where they sit on the two continua.

Hard vs. soft, and economic vs. social infrastructure

Hard vs. soft, and economic vs. social infrastructure

So, technically, you can call all of this stuff infrastructure. But, this isn't what most people (including me) mean when they say it. When most people talk about infrastructure, they're talking about the upper-left quadrant of my table above; the water and sanitation, power, transportation and telecommunications systems of a country. In general, the further you get away from hard economic infrastructure, the less likely people are to call it infrastructure.

The next table shows what most people mean when they talk about infrastructure, with darker shading denoting the fact that you're likely to get more strange looks if you call it infrastructure.

What most people mean when they talk about infrastructure

What most people mean when they talk about infrastructure

This blog will follow these conventions in most cases, I'll generally talk more about harder economic infrastructure. This is definitely not because I think hard economic infrastructure is more important than other kinds of infrastructure. It's partly because my experience is mostly in these areas, but it's also because sorts of economic tools and techniques I typically use are more easily and less controversially applied to this sort of infrastructure. I'll talk about the difficulty and controversy in future posts, because it's very interesting!

So, that's what I (and lots of other people) talk about when we talk about infrastructure. I hope you find it interesting! 

* The term "economic infrastructure", as opposed to "social infrastructure" is not universally agreed on. Some people refer to power, water, telecoms and transportation systems just as "infrastructure", while schools et al get bundled into "social infrastructure". Some people also use "economic infrastructure" to refer to the economic systems of a country; the central bank, the banking system, etc. Confusing, no?