Comparative Government Efficiency

This article discusses the fact that government is usually less efficient that the free market and estimates the inefficiency factor. We then correct a classic view of measuring cost to society of government intrusion into a free market by introducing the government inefficiency factor to a standard measurement model.

Government inefficiency is a hot topic on both sides of the political spectrum. This article discusses some of the theoretical reasons government isn’t terribly efficient, and it also highlights the fact that improved government efficiency does not lead to reduced taxation or market intrusion.

In my view there are a few additional reasons for government inefficiency other than what are discussed in that article. I focus on the keys to free markets and the lack of such features inside a traditional government structure:

  1. Traditional government structures do not exhibit free flows of information. Instead, information is often “silod,” or restricted and/or fragmented.
  2. Traditional government structures do not exhibit perfect competition. Perfect competition involves a lack of barriers to entry, an infinite number of players and more, although those are the two sub-characteristics I am most interested in. An infinite number of players results in minimally biased information and valuations, like a Wikipedia entry on some topic is likely to be less biased than the individual accounts of various parties or subgroups. Competitive markets are important in economics because they allow minimally biased pricing, and they are equally important in social systems because they allow the development of a minimally biased equilibrium of information, perception and norms which competitively represents all interested parties. Much like barriers to entry contribute to the establishment of entrenched economic elites, barriers to entry in politics allow for an entrenched political elite. Both of these entrenched elites have to worry much less about high competition costs then they would in a system with no barrier to entry; a classically free system in the economic sense.
  3. The bureaucracy has been made such that it is hard to fire so that political interests cannot engage in a spoils system, but this has made it sometimes easier for workers to shirk.
  4. Government faces a relative calculation problem. Where a fully Communist or centrally planned economy faces a classic communist calculation problem of the complete absence of market price information, almost any form of government faces a much weaker, but still important, form of the same calculation problem called a relative calculation problem. In these semi-market systems there are prices which the government has access to, but these prices and their contained information are processed more slowly and with distorting bias when compared to a true free market. Furthermore, the prices themselves may be manipulated prices due to government intrusion. Finally, these two effects may create a further synergistic distortion when fed back into each other.

To be fair, there are also real cases where the government has outperformed the expected result in the free market, but these are the exception and not the rule.

These effects of inefficiency are not traditionally accounted for in cost-benefit analyses for government budgets. I wonder why?  😉

Anyway, let’s fix that.

In modern, mainstream neoclassical cost-benefit models of government intrusion into an otherwise free market, the dead-weight loss to society is calculated this way:

(Untreated consumer surplus + untreated producer surplus in free market) – (treated CS + treated PS + government revenue) = dead-weight loss, where “treatment” is the government modification of the market.

This model makes the mistake of presuming that a unit of government revenue is equal to a unit of benefit to society. However, because government spending is comparatively less efficient than untreated market spending, there is the need to introduce an inefficiency multiplier to correctly gauge the real benefit to society of each unit of government revenue.

This inefficiency occurs in at least four ways, some of which we can somewhat readily measure:

  1. Government will chose to spend on projects that an untreated market would not select. The markets will forego the opportunity cost of preferred resource allocation.
  2. Government will spend more on the particular project than an untreated market would need to spend. This is the one we can somewhat easily measure. This involves, but is not the same as nor limited to, overhead or administrative cost differences.
  3. Government may contribute to crowding out that would not occur without treatment, or it might under-crowd other markets, resulting in somewhat of a multiplier effect of inefficiency.

#1 is a huge deal. Think about Solyndra. Think about all the times that a politician sells a big contract for a political favor. Think about when Obama let his personal friends code the Obamacare website. While it’s not an example of spending, think about Obama using the IRS as a tool to silence conservatives. In short, #1 is the cost of corruption.

#3 is also a big deal, but it is very hard to measure. Crowding out in charity can happen, for example, when government substitutes out charity. If government is feeding the homeless why should I donate to feed them? Instead I may choose to donate less and spend my money in other, perhaps less optimal, ways.

Let’s be generous. Let’s just look at #2. First, this number is arguably easiest to measure without need for speculation. Secondly, if we can establish this number we can have a confident and conservative estimate, whereas if I tried to incorporate the others I might not be confident or I might overestimate.

The logic behind point 2 has two sides: a consumer and a producer side. On the consumer side, the logic is that when I spend a dollar I receive more than the government receives when it spends a dollar. On the producer side, the logic is that when I am paid a dollar, or when most private-sector business are paid a dollar, they output more for than the government would output for an equal amount. It’s the old idea that you should donate to Red Cross rather than contribute to the food stamp program.

First, let’s be clear that the relative superiority of free markets is a normative thing, not a certainty. This article, for example, makes it clear that government sometimes outperforms charities in terms of overhead or administrative costs as a % of total expenditure. Saying medicare has lower administrative costs as a percent of expenditure compared to private insurance is usually true, but quite a red herring. Efficiency, in my view, should be about two things:

  1. Output (quality and quantity) per input, or cost (total, absolute input, not % input and not component input, in absolute or % – That is some Krugman level n00b. Instead of that, try a healthy dose of Mankiw-nage)
  2. Compare the result with the next best (marginal advantage) or best possible (total advantage) scenarios. I don’t care if A is better than B if you completely forgot to admit the best choice, C. I am more impressed if A is 2xB than 

I would be happy with a 99% overhead rate from the government if the next best and best possible choice is an untreated market with a 99.99% overhead, for the same quality and so forth. I also am not impressed if government overhead is only 2% while the market overhead norm is 20%, if the market is producing an overall better good, or if other kinds of costs for the market are such that the total market price is lower for an equivalent quality.

This published journal article (2007) remarkably calculates a cross-program government average and a similar one for charities. They find 30% of the welfare budget is actual spent on the poor. Private companies, instead, flip that: about 70% of services on average are delivered. A notorious case where this does not hold is Medicare, because a nonvariable administrative cost is spread among a larger overall cost in an older population of health product consumers. Medicare actually has a substantially lower administrative cost as a %, and it demonstrates why the administrative cost as a % is not helpful.

Instead, we should look at the total end-cost differential for similar goods. While I don’t know the full cost comparison, we will use administrative costs – absolute, not % – as a proxy. Medicare: $509. Private industry: $453. The Government Inefficiency Factor is 453/509 ~= .89. I could say government is about 11% less efficient, but I am leaving the factor to be expressed this way on purpose as you will see. Please note, this is as conservative as I can be because there are 2 other large contributing factors which advantage the private sector which I entirely ignored, and I am using the poster boy of Government efficiency as my reference for comparison. Imagine if I had used a really inefficient program to peg this to! Many other conservatives will argue for the 70% inefficiency based on the article I mentioned. So I am being very conservative (as in, I am being particularly generous to Neokeynesians, not as in taking a hard right-of center political stance) here!!

Here’s how the new, extended dead-weight model of government intrusion looks:

(Untreated consumer surplus + untreated producer surplus in free market) – (treated CS + treated PS + (government revenue)*(government inefficiency factor)) = dead-weight loss, where “treatment” is the government modification of the market.

While the government inefficiency factor varies from program to program, ceteris paribus it is expected to be at most .89 and probably closer to .43 (= (.3/.7), the more general case).

As you can see, this model will predict that government intrusion is much worse for the economy than most people currently recognize. You will also see that in the extreme, this model indicates that government revenue is ideally = 0. Government revenue = 0 would imply a fully untreated, anarchic market. Stay tuned next time when I discuss untreated markets vs free markets! I also plan to talk about an objection to this model which I anticipate, which would be the Modern Monetary Theory, Neochartalin, Neokeynesian, and otherwise left-economic objection. Even neoclassicals might make this objection too. The objection would be that government need not obtain revenue to spend.

They might also respond that revenue does not imply spending – it could be a tax cut, etc – but this was dealt with by the article at the beginning which I will link again here. As the article clarifies, while it’s true in theory that government could shrink itself, in practice it never happens.


2 thoughts on “Comparative Government Efficiency”

Leave a Comment