Evidence for State Capacity Libertarianism

The plots do not support the hypothesis that small government produces either greater prosperity or greater freedom. (In reading the charts, remember that the SGOV index is constructed so that 0 indicates the largest government and 10 the smallest government.) Instead, smaller government tends to be associated with less prosperity and less freedom. Both relationships are statistically significant, with correlations of 0.43 for prosperity and 0.35 for freedom.

Using SoG, the Cato measure of size of government, instead of SGOV, the IMF measure, does not help. The correlations turn out still to be negative and statistically significant, although slightly weaker.

Let’s turn now to the alternative hypothesis that quality of government, rather than size, is what counts for prosperity and freedom. Here are those scatterplots:

This time, both relationships are positive. High quality of government is strongly associated both with greater human prosperity and greater human freedom. Furthermore, the correlations are much stronger than those for the size of government.

That is all by Ed Dolan, recommended, and by the way smaller governments are not correlated with higher quality governments.


There's a serious measurement problem here. Take government in current prices as a share of current GDP, gummint will grow like dropsy on account the relative price of its services has risen. Try it in real terms, and it turns out service prices are hard to measure.

Looking at the links in the link, government quality seems to be largely measured by freedom indices of various sorts. There's a bit about health and stuff. Be nice to know how that's measured.

So we have: Freedom good. Surprise?

I'm shocked that there might be something manipulable about the measure of "quality of government"!

You need to read the original, more detailed version of this material linked in the shorter Medium version. Here is where to start: https://niskanencenter.org/blog/freedom-government-part-one/

You will find that two different size-of-government measures were used, which deal with your objection in completely different ways. Neither shows that small government contributes to freedom or prosperity.

Quality of government is not measured by "freedom indexes, health, and stuff." It is measured by Legatum's governance index, Cato's rule-of-law index, and Fraser's "legal system and property rights" index.

@Dismalist - good points, and I hope of course they normalized "big government" as "% of GDP".
Essentially as Dismalist implies or says, the correlation of "good" or "big" government and "sound economy" is backwards: as a good ex ante economy grows, it can afford a 'bigger' government, and since the prices of government services get inflated due to the lack of productivity in the service sector (Baumol's cost disease), which TC has spoken about (https://www.bloomberg.com/opinion/articles/2017-01-18/this-economic-phenomenon-is-making-government-sick)

Essentially, the above study is a statistical misunderstanding. Proof: in parts of Somalia there is no government, and, before religion got involved, the economy was doing just fine. Further, note "high tax" states like California in the USA have higher growth than say a low tax state like Florida or South Dakota. Does that mean higher taxes are better? Or that the stationary bandit centered in Silicon Valley can extract more money out of the private sector than the stationary bandit centered in a farm belt or retiree geography? I'm betting on the former, as I'm not a fan of government (being in the 1%).

Bonus trivia: https://www2.deloitte.com/us/en/insights/economy/issues-by-the-numbers/trade-in-services-economy-growth.html
July 12, 2018 · The relationship between services growth and overall economic growth has become stronger in the past two decades as services’ average contribution to GDP and value added has increased. In 2015, services’ value added accounted for 74 percent of GDP in high-income countries, up from 69 percent in 1997.

So we are supposed to embrace larger more authoritarian governments simply because someone arranged a chart to prove their point???

"@Dismalist - good points, and I hope of course they normalized "big government" as "% of GDP"."

+1, And how does Cuba and North Korea work out? They both are very high on size of government.

"...and by the way smaller governments are not correlated with higher quality governments." But smaller state size is. See the appendices to FH Buckley' s excellent new book on secession.

The United States is a great country. Giving up on it is a sad and pathetic response. We'll muddle through. Thanks for the unhelpful suggestion though.

Succinct and accurate, +5 i.p.

Where were you when those ungrateful colonists decided to secede from my united kingdom?

Probably working on his rebel yell.

@Donald Pretari
But you could push more down to the state level. The federal government main comparative advantages are in defence and collecting taxes, let them focus mostly on that and send most of the money that they collect to the states.

For example on healthcare, the feds should send all the money that they would have spent on HC to the states on an age adjusted per capita basis, but to get the money the state must cover at least the old and poor.

The benefit of smaller governments when they are local is that they are more responsive to local interests. The detriment of smaller governments when they are local is that they are more responsive to local interests.
I am making the assumption that if a national government is small it will also be weak and some or many of its functions will subsequently be taken over by local governments, including gangs and warlords.
As often is the case, the Goldilocks principle applies. As the historians say, context matters: Context, context, context.

Local governments may have bad procurement policies (or none); involve local contractors or developers who can exercise more influence over local officials than they could over state or national contract administrators.

Last time I was in DC I think there was a scandal in Prince Georges county, but, then again, there always is.

Um, correlation doesn’t test a hypothesis about causality. If more prosperous societies tend to build bigger governments, even if bigger governments are an expense rather than an asset, there will be a positive correlation. If Nigeria increased taxes and spending, I don’t think this will make it freer or more prosperous, but less so; not would Switzerland become less free or prosperous if it reduced taxes, spending, and regulation.

For Dolan to test the hypothesis he’s pretending to test, he’s have to look at what happens or doesn’t happens when countries of a given quality of government increase or decrease the size of their states. Or perhaps for his next post he can show, by correlation study, that buying a luxury car greatly improves your financial well-being.

+1, very good point.

Of interest is the high ratings of the scandinavian countries. I don't know where Canada fits. But all the northern democracies dramatically restructured their governments starting in the mid 80's, mostly due to the high interest rates at the time as well as a number of high debt countries defaulting. Government spending patterns changed, deficits were cut and the way government operated were reformed.

How did these countries change in the ratings from say 1960-2000, going from socialist or very close to socialist governments to center right regimes?

Here are the ten smallest governments as a share of GDP (the second measure used--the first isn't easily findable). Yeah, it's not good to have a dysfunctional government that is unable to effectively collect taxes or provide services. However, that doesn't say a lot about whether middle to high income countries with relatively functional governments should aim for a bigger or smaller government.

Central African Republic
Democratic Republic of the Congo

Part of the problem here is that the relationship between government size and prosperity is obviously not linear. Having no government is very bad. Having a government that consumes all of your GDP is also obviously bad. The optimum size of government is somewhere in between. Applying a linear model to this dataset is not really informative.

Also, there is a survivorship bias problem here. We are looking at the governments that have been successful enough not to collapse or get overthrown. There are very poor governments both at the bottom and top of this spectrum, but it's plausible that the failure modes are different for tiny and huge governments. If (for example) small dysfunctional governments last a long time, but huge dysfunctional governments collapse, this would bias the sample.

Yes, Mark Z, you are right, multivariate analysis is called for to untangle the causal relationships. You will find the answer to your question about what happens or doesn’t happens when countries of a given quality of government increase or decrease the size of their states in the more detailed version of the research posted at NiskanenCenter.org. Here is the link: https://www.niskanencenter.org/quality-government-not-size-key-freedom-prosperity/

Here's what Thoreau said...

"I heartily accept the motto, "That government is best which governs least"; and I should like to see it acted up to more rapidly and systematically. Carried out, it finally amounts to this, which also I believe — "That government is best which governs not at all"; and when men are prepared for it, that will be the kind of government which they will have."

What kind of men will be prepared for it?

"I am as desirous of being a good neighbor as I am of being a bad subject."

In the meantime...

"To speak practically and as a citizen, unlike those who call themselves no-government men, I ask for, not at once no government, but at once a better government."

Seems fine to me.

Something unmentioned on Mr. Dolan post is the scope of the government: local, regional, national/central.

The IMF warns about the source data on Size of Government "These estimates have to be interpreted with caution, since central government expenditure provides a somewhat distorted picture of total public spending, particularly in federal countries with large sub-national governments. As we discuss in the data quality section, it is unfortunately hard to find reliable cross-country data on expenditure below the central level" : https://ourworldindata.org/government-spending#total-government-spending

A extreme case: the Swiss federal (central budget) is only ~30% of total "government" budget.

In our big data era it should be possible to overcome that limitation on the cross-country data of government expenses below central level....albeit, we keep arguing in the dark ;)

I love the smell of cognitive dissonance in the morning.

GMU professor Steven Pearlstein on the cost of a broken government: https://www.washingtonpost.com/business/economy/the-us-economy-is-booming-our-broken-politics-are-going-to-wreck-it/2020/01/23/21c812ac-3d24-11ea-8872-5df698785a4e_story.html

I've always been puzzled by some Libertarians to think (and Republicans who say, though their actions show that they do not believe any such thing) that lower spending by the Federal government would lead to greater freedom and prosperity. [See the very odd piece by John Taylor in Project Syndicate https://www.project-syndicate.org/commentary/restoring-fiscal-order-in-united-states-by-john-taylor-5-2020-01 who advocates shrinking the share of federal spending with no discussion of by what criteria expenditures should be reduced.]

Taxes have a very high deadweight loss, so it makes complete sense that things outside of public goods are not provided by the government.

Maybe Somalia is not so great as Libertarians think after all.

Good point, Mr. Reynolds.

Excuse me, Mr. Halls. I am a Doctor.

So am I, Doctor Reynolds.

More important than size of the government is the source of its revenues. The more of these that are paid directly by identified citizens, the more must the government care about the citizens, and so the better will the quality of the government be.

Are countries that care only about resident citizens and not foreign resident corporations and citizens always the best outcome? If Amazon.com is responsible for prosperity in a nation, then perhaps you want government to care about them (while remaining alert to disrupting fair and openly competitive markets), and so, following this logic, you want them to pay fairly high taxes.

It seems like the correlations should have been run on subsets of countries, or maybe just a simple OLS regression (with all its issues) showing the regression line between size and other variables. When thinking about the other policy areas, standard good practice is to compare like with like. So show the correlations controlling for GDP per capita or just comparing OECD countries etc. These are simple things to do, and the fact the author didn't do them makes me think he's out of his depth on simple statistical analysis or the results from better analysis aren't as striking. It's not clear to me that there will be such a strong correlation between size and freedom or size and quality. Just eyeballing the graph in the article with some labeled country data points, there's probably a flat to slightly negative correlation between size and quality. I think Cowen is... how would he put it... mood affiliating?

JFA -- Could you please read the original long-form articles, linked at the end of the short Medium version? You will find many of the statistical tests you request there. (The two-part series starts here: https://www.niskanencenter.org/freedom-government-part-one/

Furthermore, I did many things such as testing for OECD numbers only, throwing petro-states in or out of the sample, and so on, too many to report every result in the paper.

If you are interested in getting into the weeds with a serious dialog on statistical methods, please contact me directly. I am sure I have a lot to learn. But, on the other hand, understand that the Medium post only gives highlights.

Worth reading this short piece from Tim Besley "Debating the Size of Government is a Distraction"

"For a market economy to flourish, government needs to be effective and constrained but it need not be small. Casual empiricism should make this claim self-evident. Among the richest economies in the world are the Nordics (Denmark, Norway, and Sweden) which seem to thrive (on a wide range of metrics) in spite of their apparent preference for high taxes and public spending...There is no paradox to explain. Countries which have large governments, measured by share of taxes or spending, also by-and-large have effective governments. So the real focus should be on why this happens in some places and not others – focusing on government size is not only a distraction, it diverts focus from what matters."

Nice sentences.

If you look only at Western countries (i.e. countries that have had 200 years of industrialization and marketization and no history of colonial exploitation), then you'd get the opposite results; the countries with smaller government are more prosperous (although their government quality is not necessarily higher). Some poor countries could indeed benefit from a stronger government (assuming it was of decent quality) because they lack any meaningful private sector wealth that could be used to fuel market-led growth. However, that does not apply to developed countries that do have significant private sector wealth and investment.

I think the point about the old saying is useful -- smaller on its own is hardly what is needed. Seems like an error similar to I think commonly occurs with the application of Ochamm's Razor. Was it Einstein that made the statement - keep things as simple as possible but no simpler?

Maybe the same applies to the claims about small government. But that doesn't solve any definitional or measurement aspects.

The criteria for what constitutes quality of government seem weak in terms of this overall goal. The question is really what size the government needs to produce the "social services" or "public goods" mentioned: rule of law, limited corruption and the like. Moreover, it is not clear the production is purely within government or if it needs to be.

I also wonder if the initial analysis showing the negative relationship is complete. Looking at the scatter plots there seems to be some internal factors that maybe need exploring before we simply apply the linear analysis to show a relationship. To the extent the simple regression points to asking questions that is good but I don't think it really has fully established the claimed relationship.

I feel like I've been calling for quality of government, effective governance, for some time.

+1 Quality and effectiveness, irregardless of size. Measure the output and pick the delivery vehicle that maximizes benefit and minimizes costs.

Mostly you have been railing against the conspiracy of Republicans who deliberately sabotage government quality. To you, quality governance = Democrats. If you have actual ideas then you have not deigned to share them.

Wow, someone from the Niskanen Center found "evidence" for a strong central government? I mean the Niskanen Center, of all places! I did not see that coming! Next you're gonna tell me that they like UBIs and Elizabeth Warren!

RP long: don't forget about universal catastrophic health coverage. If you are going to have a free-market welfare state, you have to touch all bases.

I can see where state capacity is indicative of a country that can be a success, but do they have to use the capacity?

I think that if the 90%+ of Swedes knew how much taxes they pay and what they get for their tax money, they'd cut the size of Government.

Singapore has the most effective government in the world, and its pretty small for a first world country.

The "big government" Scandanvian countries actually stated doing better after they got smaller and move to the right in the 1980s.

We get it. Basket case third world hellholes have small governments for the same reason they have small everything else, they are too low IQ to make anything work in the public or private sectors, and there is little economic surplus for the government to confiscate.

Dolan lists his indices in an image on his blogpost here (https://www.niskanencenter.org/quality-government-not-size-key-freedom-prosperity/), so by the magic of optical character recognition, we can extract the "Size Of Government" index and plot it against some other measures.

1) ILO Public Share of Employment vs Dolan's "Size of Government - https://i.imgur.com/KlF5UhO.png

There's some relationship, but quite a few of the states Dolan's measure identifies as small (high scoring on his index!) actually have high levels of Public Sector Employment, particularly Singapore.

The relationship is present, but somewhat loose. Dolan's measure would have Netherlands and Norway as having about the same state size, but measured by employment Netherlands is rather low, while Norway is quite high.

2) Expenditure per Capita as a % of Household Final Income vs Dolan's "Size of Government - https://i.imgur.com/9fHg6Zi.png

This relationship is a lot stronger, though there are some outliers. Particularly the "Tigers" (Asian and Celtic) are measured as having smaller government size by Dolan's measure as compared to normalizing government expenditure to private consumption.

3) Dolan's quality of governance is unrelated to ILO public share of employment - https://i.imgur.com/AdEH8X2.png

4) Legatum's Prosperity Index has a reasonable relationship with Household Final Consumption - https://i.imgur.com/vM0YH04.png ... Except that the United States has a far lower score under Legatum's index that its very high consumption standards would indicate.

If anyone would like a closer look at Dolan's data, here is a pastebin of it in .csv format - https://pastebin.com/e1XY8FPb

Is either smaller or larger government related to Quality of Government once we regress out the correlation of Household Income with Quality of Government?

Regress Dolan's Quality of Governance measure on National Household Income Per Capita - https://imgur.com/a/wYELXrs

That should account for any systematic "richer countries are better governed effect".

So will the remaining variance be related to size of government? Let's test it using three different measures of size of government; Public Sector Share of Employment, Dolan's Size of Government and Public Expenditure as a % of Household Income Per Capita: https://imgur.com/a/RhsXye8

Seems like nothing.

Regress x on noisy x, find spectacular fit... Need to rethink method.

"compile such a measure by extracting and combining relevant indicators from the LPI and HFI to construct a Quality of Government Index (QGOV)"

So what is "QGOV"? See here: https://www.niskanencenter.org/quality-government-not-size-key-freedom-prosperity/

"For purposes of this post, I combine the Fraser [Economic Freedom Index], Cato [Personal Freedom Index], and Legatum [Legatum Prosperity Index] indicators into an equally weighted average that I will refer to as the quality of government indicator, or QGOV"

So basically it is 2 economic freedom measures (although note that those consider government as percent of GDP as a negative) and an index that is a weird mixture of economic freedom (like "market access", "investment environment") with other result-oriented metrics like "Safety & Security" and "Living Conditions".

Probably he could have just said "economic freedom" instead of "government quality" and obtained the same results (see economic freedom literature).


A somewhat unrelated technical question: Is it standard in economics to describe such bivariate distributions as "[horizontal axis] vs [vertical axis]"? This is reverse of the convention in the physical sciences.
Over the last 10 years of teaching college physics, I've noticed more and more students being very confused about this (i.e., doing it the way that it is represented in these graphs, claiming that's they way they were always taught). I've polled colleagues in Chemistry, Biology, and Mathematics to try to find out who is misinforming our students, but it seems like the culprit might be the social/behavioral sciences???

You can draw a similar plot of software that runs on various kinds of devices, but it would be folly to conclude afterwards that writing a more bloated and resource intensive program will lead to the device it's running on to have more cpu cycles, ram, or storage. More likely that government, like software, tends to consume available resources.

This is a weak analysis. The "small" government list just includes poverty stricken countries with no ability to provide even basic services. This should pick a band of gdp per capita (e.g. 10-20k, 20-30, 30+) and then try to see if size vs outcomes holds.

Common sense: I'm not sure if you have read the long version of my research, published at Niskanen Center, starting here: https://www.niskanencenter.org/freedom-government-part-one/ It has a lot more than the shorter summary at Medium.

To answer your comment, yes, I ran the regressions on many such subsets: Top GDP quartile, OECD only, rich countries less petro-states and so on. I could not include every such test in the published version but generally speaking, they confirmed the robustness of my main conclusions.

If you, or anyone with similar questions, really wants to get into this, please email me via the contact tab at Ed Dolan's Econ Blog. I still have the original data set and some (I'm not sure all) of the intermediate results, so I can send you something to play around with.

Comments for this post are closed