February 17, 2004

Now back to the usual...

I'm glad to write that the domain name trouble has been fixed. Service should be back to normal.

The core problem was the inflexibility of transferring domain names coupled with the even more inflexible Swedish association system. It suggests an interesting research question: how to construct distributed domain name management without loss of property rights?

Posted by Anders at 02:28 PM | Comments (5)

February 10, 2004

More Fun With Statistics

As a continuation of my previous experiments with the freedom indices, I plotted some other national data.

Besides GDP life expectancy is a good indicator of quality of life. Here is the life expectancy at birth vs log GDP:

There seems to be a "main sequence" where life expectancy increases roughly with the logarithm of GDP for most somewhat well-off nations. Below a certain cut-off things rapidly turn far worse, most likely because basic health services and sanitation are absent. There is also a cloud of nations in the low middle with far lower life expectancy than their GDP suggest. These nations are the main victims of the AIDS epidemics - this diagram shows the enormous cost in terms of life expectancy.

So, let's start looking at how economy and freedom interacts in a few other ways.

The Freedom House data is wonderful because it contains historical entries from the early 70's. One simple way of visualising it is to plot the sum of the changes for all nations.

Two peaks of freedom can be noticed - the de-colonizations in Africa in the early 70's, and the big one: the fall of the eastern block. It is interesting to see the oscillations preceeding it, including sharp declines in 1975 and 1988 and how increases in civil liberties appears to preceede increases in political freedoms. There is also a sad dip in civil liberties in 92, likely due to the emergence of many new not-very democratic states.

Another fun way of visualizing what is going on is to plot it as color (red=little freedom, blue=much freedom, white=no data). Starting in 1972, one can either sort the countries after how free they became or how free they were when they started:

[Political freedom and civil liberties added together]

The first plot shows where they ended up. We can clearly see that democracies are rather stable, while there is plenty of things going on in the intermediate range. The worst tyrranies do tend to remain despotic. The color fields to the right of the figure shows the relative sizes of different freedoms: democracy is exapanding, but so is the orange-red area.

The second plot shows how they started. At the top it is fairly clear when new states emerged (former colonies in the 70's, the USSR breakup 90). One can see how the new states of the 70's either went democratic right away, or spent the time up to 90's with serious lack of freedom (lots of African communist-despotic states here) and still have not gotten halfway to maximal freedom. It is also clear that democracies remain democratic, but even the worst dictatorships can change.

I also tried plotting everything. Log GDP, absolute change in GDP per capita and economic freedom index, with colors denoting life expectancy and sizes corresponding to freedom house's index.

This is a good example why plotting too many parameters is not always good visualisation :-)

Some interesting features: most states lie along a "main sequence" or rather a slightly curved surface of increasing GDP, growth and freedom. The flatness for poor countries is of course partially due to using absolute GDP change rather than relative; the spread shifts the other way if relative change is used. The wounded economies of Venezuela, Argentina and Uruguay are clearly visible as one branch, while the growing anglosphere is expanding in the other direction led by Ireland. There might be a third one, involving Iceland, Kuwait, Japan and some other nations that are not doing too well (hard to see, even when you can rotate the diagram freely). Libya is a suspicious outlier - people get too old and it has too much wealth to fit the low freedom indices. The AIDS-victims of southern africa are visible as blue dots in the middle of the middle red.

Finally, just a "phase space plot" of different countries and how their GDP and freedom index are changing:

[Version with directions color marked to the right]

Each line shows how the nations would change if they kept up their present trend (current GDP change, freedom index weighted with an IIR alpha=0.5). The usual stable, well-off corner down to the right is there. A few nations like France and Spain are moving the last bit to maximally free in the index. Up to the left of them there is a group of nations that have become rather democratic and now are moving up mainly economically: Latvia, Lithuania, Grenada, Poland, Costa Rica, South Africa, Estonia, Slovak Republic, Hungary and the Czech republic.
Overall, wealth seems to stabilize political systems. The wild shifts up and down exist in the poor left half of the diagram, while the right is far calmer.

Data and programs

On request, here is the data files and Matlab programs I used to make the illustrations in this and previous post. The data itself of course belongs (?) to the Heritage Foundation, Fraser Institute, Freedom House. I take no responsibility for correctness and suggest that anybody who wants to play seriously with this fun field download it directly from them, since there are many useful comments on methodology at the sites and it is so educational to write one's own code.

Download phaseplot.m
Download index of economic freedom file (Heritage data)
Download GDP/life exp.
Download gdp growth
Download country names for FH historical data
Download poltical freedom
Download 3D plot
Download change plot
Download color order plot
Download freedom house time data
Download fraser-heritage plot
Download loadall.m
Download fraser data
Download correlation matrix plot

Posted by Anders at 07:19 PM | Comments (9)

February 07, 2004

Freedom Benchmarking

Does the rankings of economic freedom from the Fraser Institute and Heritage Foundation/Wall Street Journal point in the same direction? Here is a simple scatterplot I did of one index against another. Click on the thumbnail for a big popup.

As can be seen there is a lot of correlation (0.85). While the ranking order could be changed quite a bit by using one index over the other, the overall estimates appears to be quite consistent. There are no real outliers in the data.

(And where is Sweden? It is hidden by El Salvador, just above Bahrain)

The Heritage index is composed of several factors, and it is interesting to plot the correlations between them and the Fraser index. Below is a correlation matrix plot, where the color of each square denotes how much one factor correlates with another. The hot colors show the strong correlations, the bluish the weakest (or even negative).

[Technical notes: I used the 2002 data, and negated the Fraser index so that it points in the right direction. ]

There is a big group of strongly correlated factors in the corner: respect for property rights, no unnecessary regulation and absence of black markets are good predictors for economic freedom (cause and effect is of course variable; in a free economy there is little competitive advantage for a black market, while less regulation on the other hand implies economic freedom). Another conclusion is that both indices have roughly the same correlations with the factors, which suggests that they have some validity. It is interesting to note (but hard to see) that the Fraser index correlates best of all the other factors with the entire Heritage index: the combination of factors is a better predictor than each in isolation.

Plotting GDP against freedom shows a nice curve:

Having a more free economy correlates with more GDP. Another obvious fact that needs to be stated again and again.

OK, but what about political freedom? I used Freedom House's ratings for civil liberties and political freedom, and plotted the average of them against the Heritage index.

[The small height differences between countries are just a display trick. Since the FH index either is an integer or a half integer there was a great deal of crowding of the text, so I jittered the positions upwards and downwards by +-0.1.]

This scatterplot has some fun aspects. We have a dense cluster of very politically free western countries at the lower left with some of the "well-behaved" South American countries like Chile and Uraguay. Then there is a wide sigmoidal band leading up to the Axis of Evil. Some notable outliers are Saudi Arabia, Bahrain and the Arab Emirate - oil is good for business but bad for democracy. And Singapore is doing its best cyberpunk imitation. Among the 'nice guys' Estonia is notable and should really fix its civil liberties. On the other side of the sigmoid we have Suriname as a surprisingly democratic but economically unfree country (probably because of the focus on Bauxite and the governments economic troubles). But on average, the trend is that political freedom and economical freedom goes hand in hand. The correlation is 0.65 - not bad (if this had been psychology :-).

This kind of indices are of course subjective and prone to biases of various kinds. Still, they seem to provide a reasonably good aggregate sketch of the state of different nations. They don't tell us in themselves why a nation is as it is, but it shows many of the commonalities. The next step in analysis is to look at how they change in time. Does political freedom often cause economical freedom, or vice versa? How does economic freedom changes lead to GDP changes? Lots of fun things to explore with a spreadsheet or Matlab for the armchair economist.

Posted by Anders at 03:04 AM | Comments (4)