EconLog small logo  

For Individual Liberty, Size Does Not Matter - or So It Seems

The Case Against Education<... Whereof One Cannot Speak, Ther...

by Pierre Lemieux

"... let's have a look at the empirical evidence. Do available data show a correlation between individual liberty and the size of a country? A positive or a negative one?"

Catalonia.jpg In a recent Econlog post, Alberto Mingardi provided an interesting reflection on the Catalonia secession attempt. Since then, a secessionist government has been reelected in the region, and the future is uncertain.

More generally, Alberto suggested that, if states are unavoidable, having more of them is better than fewer. One reason is that individuals will have more choices as to which state to live under according to their preferences. A related reason is that competing states will have an incentive to satisfy the preferences of their clienteles. More numerous and homogeneous states will thus offer public services more in line with what citizens want. Alberto Alesina and Enrico Spolaore make related arguments in their book The Size of Nations.

However, there are also good arguments for larger, more cosmopolitan societies. What Friedrich Hayek, in his trilogy Law, Legislation and Liberty, calls the Great Society, i.e. the abstract classical-liberal order, is more likely to flourish in a large, cosmopolitan society than under small-group tyranny. In a large and diversified society, Leviathan may be less effective at oppression. At any rate, smaller, homogeneous nation-states will typically not allow individuals to move freely between themselves.

Alesina and Spolaore argue that, ceteris paribus, Leviathan prefers to rule over a large country because he can extract a larger rent. However, he may find it easier to take over a small country and will be able exploit the minorities there more harshly. There is no totally homogeneous country, and Leviathan will always find minorities to exploit in favor of its favored clientele, which is often the dominating cultural group. In this perspective, it is not surprising that most secessionist movements are run by socialists. (In Québec, it started with rightist, nativist organizations, but was taken over by socialists in the 1960s.)

Political scientist Karl Deutsch was on to something when, in Nationalism and its Alternatives (as quoted by Alesina and Spolaore), he defined a nation as "a group of people united by a common error about their ancestry and a common dislike of their neighbors."

The issue of whether large or small countries offer better chances for individual liberty is not simple, and the two opposite theories seem defendable. So let's have a look at the empirical evidence. Do available data show a correlation between individual liberty and the size of a country? A positive or a negative one?

The cloud of data points on my scatter diagram below suggests that there is no correlation between the size of a country, measured by its population, and its degree of liberty, as measured by the Fraser Institute's Economic Freedom of the World index. My data incorporates the 160 countries included in that index. The variation of the scores (where 10 represents the highest level of economic freedom and 0 the lowest) reminds us that some small countries are relatively free (Hong Kong and Switzerland are examples) and that in others, as Émile Faguet would have said, people "are more oppressed than in Turkey" (Venezuela or the Central African Republic, for example). Same for large countries: some (consider the United States) are relatively free, others much less free (India or China).

Freedom population 2 smaller.jpg A simple regression analysis confirms the absence of statistical correlation between country size and economic freedom. The R-square--a measure of correlation that indicates the proportion of the variation in freedom explained by population size--is tiny: 0.004. The slope of the regression line, which suggests that the freedom index score decreases by 0.00038 for every increase in 1,000,000 inhabitants, is not statistically significant: the P value is 0.42. In other words, there is a 42% probability that the apparent negative effect (already tiny) of size on freedom is due to random factors.

The statistical conclusion does not change if we replace the countries' population by their size in square miles; or if we exclude China and India as outliers; or if we redo the analysis with both population and geographical size as variables (a multivariate regression).

In case the problem lies in the economic freedom index, I re-ran the simple regression using instead Freedom House's Freedom in the World index (for 194 countries). This index measures mostly political freedom, that is, such things as free elections, government openness, the rule of law, free speech, and other civil rights. The Freedom House index is slightly biased towards positive, as opposed to negative, freedoms. The results of the statistical analysis using this index are similar: the R-square is 0.005 while the P value goes down to 0.30.

It is true that correlation is not causation, but non-correlation is even less so. According to my simple statistical analysis, whatever the theoretical arguments are for the outlook of liberty in small versus large countries, there is no supporting evidence one way or another.

As my economist friend Sinclair Davidson pointed out to me, this does not prove that secession from a larger country (or for that matter, integration into a larger one) will have no effect on freedom. Statistical analysis is a tough master. We would need to compare freedom levels before and after secession (or before and after union) in a representative sample of countries having experienced such a move, but such a database is not available.

It does not follow that liberty never justifies secession. In particular cases, it may very well do so. Breaking a Leviathan's stranglehold on the seceding region would be a libertarian justification, assuming that a worse Leviathan is not likely to take over the "emancipated" region. The argument against a world government is closely related: under the reign of a single Leviathan without competition, there would be no way to escape into a separate space of liberty. So some government diversity is required as an insurance policy against monopolistic Leviathan.

In other words, it must be possible to break away from an existing state, even if the required majority is in many ways arbitrary. As Alberto argues, it should be more than 50% plus one, and there should be an accepted process, some agreed-upon rules, to realize the secession. Indeed, states that prohibit secession, like the Spanish state seems to be doing, thereby provide an argument to secede, as opposed to such states as the Canadian or the British states which would (from what we know) recognize secession under some democratic rules. Note that the European Union, a sort of state, also accepts secession.

What I think my reflection suggests is that, in regard to liberty, the size of a country does not matter per se.

COMMENTS (15 to date)
Jon Murphy writes:

Interesting stuff.

I doubt anything will happen, but do the results change if you use log population?

Hazel Meade writes:

I wonder if anything would pop out if you plotted this against the log of the population. or did the regression against the log (base 10) of the population.

Pierre Lemieux writes:

@Jon and @Hazel: I wondered about that myself, but I didn't try it because the data and the residuals don't lend any visual support to the idea.

It seems to me that the size of a jurisdiction ought to be weighted according to available forms of transportation, and to location of the population relative to geographic center. If the typical denizen can get to the border in a matter of hours, her country is effectively smaller than if it would take her days, regardless of physical geography. Of course, arriving at appropriate measures and measurements seems a dishearteningly difficult task.

It seems to me (as to many before me) that one of the reasons for the rise of the West and for the decline of the Islamic world in the millennium after about AD 900 was exactly that the West was politically fractured, allowing scientists and philosophers more opportunity to escape from one jurisdiction to another, whereas the geniuses of Islam were trapped.

But this theory brings to light a new complication. Even while the Islamic world was more integrated and centralized, in theory a Muslim near the border might flee, but he would have had then to flee to a non-Islamic nation, whereas a Christian could flee and yet remain within Christendom (and within a world in which intellectuals spoke Latin). Now-a-days, I could potentially flee to many other Western nations, and to many jurisdictions in which I would not even need to learn a new language (though I'd have to endure still more people who say ‘whilst’). Other people lack these privileges. Again, arriving at appropriate measures and measurements seems a dreadful task.

Pierre Lemieux writes:

@Daniel: Thanks for reminding me, in your second paragraph, of a fact that should be included in any discussion of this topic, and which should have been mentioned in my post. Your other critique is relevant if one thinks there are good reasons to define size in geographical terms--which I did only incidentally when I reran my regressions to check if I was missing something important. Perhaps physical size would be better for what we really want to measure, that is, heterogeneity; but as you mention, there are many problems with straight geographical size. Note that there is a highly significant correlation (R-square) of 0.195 between population and straight geographical size.

Mark writes:

I think quality of governance would depend more on how many neighbors (or culturally similar neighbors) than size per se. For example, if Belgium mistreats its citizens too badly, they could move to the Netherlands, France, or Luxembourg, and it wouldn't be terribly inconvenient. However, a comparably sized state that is an island in the middle of an ocean with no close cultural cousins nearby enjoys much more of a 'monopoly' on its citizens.

In other words, as with analysis of private industries, people too often mistake firm size (or market share) as a stand in for competitiveness. But it's really the presence of alternatives that matter, and that just happens to sometimes correlate with average firm size.

Martin writes:

Pierre, I have two comments: One regarding your conclusion from the freedom data, and one regarding your statistical analysis.

Firstly, the higher dispersion (variability) in smaller countries is precisely what is desirable! If two subunits of a state split up into separate states, one of these subunits are likely to contain a more liberty-oriented population and government, and one is likely to contain less liberty-oriented ones. Those who prefer the other regime can move! With smaller states, the sociocultural and economic cost of migration is lower. That is great!

Secondly, it is not accurate to conclude that there is a 42% chance that the effects are from random factors. There's a 42% probability to observe data at least as "extreme" as we did, assuming that there is no true effect. Very common mistake to make regarding P-values. Inverting the conditioning (computing probability(effect given data) instead of probability(data given effect)) cannot be done without assuming some underlying a priori probability of the effect.

Misthiocracy writes:

I would like to see a similar regression analysis done not for population, but rather for square mileage.

How does individual liberty fare when the jurisdiction becomes physically larger, and therefore (arguably) more difficult to govern?

A third regression analysis based on population density might also be interesting.

James Pass writes:

Mark said "it's really the presence of alternatives that matter." But isn't having alternatives just one aspect of individual liberty?

Mark also mentions an "island in the middle of the ocean," in other words, geographical isolation. But we see countries like East Germany (before 1989) and North Korea that have a pretty strong monopoly on their citizens in spite of their proximity to "close cultural cousins."

Arnold Kling writes:

Linear regression is a horrible way to approach this data. All you are doing is comparing China and India to everyone else. If you use more sensible methods, you will see that small countries do well.

Mr. Econotarian writes:

The good news is that the Socialists' Party of Catalonia has dropped down from 37.9% of votes for the Parliament of Catalonia in 1999 to just 13.9% today.

Pierre Lemieux writes:

Thanks for your comment, @Arnold. Of course, I would't claim that my little statistical analysis is the end of the matter. However, as I mention in my post, "The statistical conclusion does not change if ... we exclude China and India as outliers."

Pierre Lemieux writes:

@Misthiocracy: I haven't done a density regression, but I suspect that the results would not be much different since there is a statistically significant correlation between population and physical size. As for the physical dimension as an independent variable, I think that my post answered your objection: "The statistical conclusion does not change if we replace the countries' population by their size in square miles."

Pierre Lemieux writes:

@Martin: Is your second criticism a Bayesian point?

Pierre Lemieux writes:
Comments for this entry have been closed
Return to top