I used to think Narwhals were fictional animals. And people telling me they were real were just joking. It wasn’t until HS that I was convinced they actually existed. My mental process was like “No way are there aqua unicorns.”
Semi-political: I used to believe the correlation between economic freedom and economic growth was much stronger than it is. (I know there is no canonical choice of measurement for either variable). This realization had pretty important consequences for me.
My estimates of public opinion surveys were totally wrong. On almost every issue (sexuality, morality, politics, etc) I was completely wrong about the distribution of beliefs. Given my history of failure in this domain I no longer really on my own “intuitive” estimates of the distribution of group beliefs. Instead I seek explicit surveys.
The top 10 are all very prosperous countries. In particular, Hong Kong and Singapore are both much richer than surrounding areas. Chile is conspicuously richer than other South American countries. Mauritius is conspicuously richer than other African countries. Ireland is one of the wealthiest countries in Europe.
The top 10 are all very prosperous countries. In particular, [...]
Princess_Stargirl is talking about the correlation with economic growth, which is not going to be the same as the correlation with economic prosperity. This looks like a confusion of a variable with its rate of change.
It is informative to see what happens if I (1) correct this by correlating the economic freedom index with GDP growth, and (2) use as big a sample as is available instead of focusing on particular cases. I copied the freedom ratings from that Heritage web page and real GDP growth rates (“estimates are for the year 2013 unless otherwise indicated”) from Wikipedia. The correlation between the freedom index and real GDP growth turns out to be negative: the Pearson correlation coefficient is −0.19 and Spearman’s rank correlation coefficient (which allows for nonlinearity) is −0.35. Plotting the data and a loess curve with R’s default settings:
Some outliers are evident. Perhaps they’re disproportionately skewing the freedom-growth correlation? I take out North Korea (the far left point) and the two lowest points, Cyprus and Central African Republic:
and in fact the freedom-growth correlation sinks further, to −0.38 (Spearman’s rank) or −0.30 (Pearson). The main mass of nations hovers around the part of the loess curve which slopes downward.
At this point in time, the correlation between the Heritage Foundation’s index of economic freedom and economic growth is unambiguously negative. I suspect this is because being a poor country is associated with low economic freedom but high catch-up growth. What happens if I correlate the freedom index with annualized GDP growth over a longer period, 1990-2007, for which catch-up growth is probably less important?
I now wind up with positive correlations (+0.25 for Spearman, +0.20 for Pearson). Now I throw out the outlying North Korea (the leftmost point), Zimbabwe (the bottommost), and Equatorial Guinea (the topmost).
This has no meaningful effect on the correlations, which become +0.24 (Spearman) and +0.21 (Pearson). Overall, by switching from a more up-to-date growth statistic to a more long-term (and pre-Great Recession and mostly post-Soviet) statistic, I change the sign of the correlation between growth and the Heritage Foundation’s assessment of economic freedom.
It’s not immediately obvious to me which GDP growth statistic is more appropriate. 2013 growth has the advantages of being more up-to-date and better matching when the Index of Economic Freedom was calculated, but is less representative of each country’s long-term economic trajectory. 1990-2007 growth is more representative but also involves comparing 24-year-old data to a recent index; the proper thing to do here would be to use a similarly long-term average of the Index of Economic Freedom, but that’s too much like real work.
There is also the question of how to operationalize “economic freedom”, but this comment is long enough. I evade that problem here by simply taking the data tables suggested upthread as given, and after doing so the basic conclusion seems to be that the correlation between “economic freedom” and “economic growth” is modest, with its sign sensitive to how one operationalizes growth. (Check my work with my data file.)
[Edited 19⁄10 to change “Gunea” to “Guinea”, and add “after doing so”.]
At this point in time, the correlation between the Heritage Foundation’s index of economic freedom and economic growth is unambiguously negative. I suspect this is because being a poor country is associated with low economic freedom but high catch-up growth.
To test this hypothesis, I did a linear regression of overall score, each of the ten subscores and 2013 real GDP growth against the log of 2013 GDP per capita (at parity). I then took the correlation between the residual of 2013 real GDP growth and the residual for each of the scores. Here are the results:
overall score −0.04
property rights −0.15
freedom from corruption −0.11
fiscal freedom 0.25
government spending 0.18
business freedom −0.03
labor freedom 0.05
monetary freedom −0.12
trade freedom 0.01
investment freedom −0.18
financial freedom −0.09
These results were approximately opposite of what I expected (I expected minimal correlation for fiscal freedom and government spending and generally positive correlations for everything else). While I’m only somewhat surprised by the government spending and fiscal freedom results, I find the others very confusing. Does anyone have any idea what might be going on?
At this point in time, the correlation between the Heritage Foundation’s index of economic freedom and economic growth is unambiguously negative. I suspect this is because being a poor country is associated with low economic freedom but high catch-up growth.
To test this hypothesis, I did a linear regression of overall score, each of the ten subscores and 2013 real GDP growth against the log of 2013 GDP per capita (at parity). I then took the correlation between the residual of 2013 real GDP growth and the residual for each of the scores.
The analysis I’d do would be simpler. Compute the correlation of log GDP per capita with the freedom index (or its subscales); if I’m right it should be substantially positive. Then correlate log GDP per capita with GDP growth; the result should be substantially negative. Taking correlations of residuals addresses the different question of whether unusually high growth for a country’s income level correlates with unusually high freedom indices for a country’s income level.
I did the simpler analysis first and all the correlations between log GDP per capita and all the economic freedom index subscores were pretty negative (as was the correlation between log GDP per capita and GDP growth). Log GDP per capita was positively correlated with economic freedom subscores.
The comment above yours was not very clear. I have edited it for clarity.
There is in fact a positive correlation between the economic freedom index and log GDP per capita.
I’m more confused now. The parent comment says the EF index correlates positively with log GDP per capita, while the edited comment says the EF index subscores correlate both negatively and positively with log GDP per capita. I don’t understand how that can all be true simultaneously...
Your charts graph 1990-2007 economic growth as a function of 2014 economic freedom, 1990 economic freedom, so assuming that correlation is causation here (almost always a dubious assumption), this would indicate that economic growth leads to economic freedom, not the other way around.
That site has a nice slightly-interactive map where you can pick out individual components of their “freedom index”. Mostly they correlate with prosperity (I have no idea what the actual causal relationships are) … until you click on “Government Spending” and suddenly it goes exactly the other way round—the allegedly-worst government spending figures are for the US, Canada and Western Europe, and the allegedly-best are for severely messed up central African countries and China (!) and India.
If they stopped counting government spending as opposed to freedom—it seems to me only marginally a matter of freedom—the correlation between “freedom” and prosperity would become even more impressive.
(Note 1. The cynic in me says: Of course that’s out of the question because a central part of the reason why the Heritage Foundation exists is to argue for lower government spending and hence lower taxes. If it advocated less forcefully for that, it would become less useful to those who fund it.)
(Note 2. It seems like there are lots of other things that could go into a “freedom index” with about as much reason as government spending. Two examples: longer working hours mean less freedom to do as you please with your time; stronger IP law means less freedom to start a technology-based business, to do as you please with the books and music and software you own, etc. Again, the absence of these things from the Heritage Foundation’s “freedom index” seems adequately explained by the interests of the organizations that provide its funding.)
I used to think Narwhals were fictional animals. And people telling me they were real were just joking. It wasn’t until HS that I was convinced they actually existed. My mental process was like “No way are there aqua unicorns.”
Semi-political: I used to believe the correlation between economic freedom and economic growth was much stronger than it is. (I know there is no canonical choice of measurement for either variable). This realization had pretty important consequences for me.
My estimates of public opinion surveys were totally wrong. On almost every issue (sexuality, morality, politics, etc) I was completely wrong about the distribution of beliefs. Given my history of failure in this domain I no longer really on my own “intuitive” estimates of the distribution of group beliefs. Instead I seek explicit surveys.
On a similar note to narwhals, for a while I assumed that fan death was just a meta-urban legend.
You were not alone in your narwhal-doubting.
Here is a list of countries ranked by economic freedom:
http://www.heritage.org/index/
The top 10 are all very prosperous countries. In particular, Hong Kong and Singapore are both much richer than surrounding areas. Chile is conspicuously richer than other South American countries. Mauritius is conspicuously richer than other African countries. Ireland is one of the wealthiest countries in Europe.
Princess_Stargirl is talking about the correlation with economic growth, which is not going to be the same as the correlation with economic prosperity. This looks like a confusion of a variable with its rate of change.
It is informative to see what happens if I (1) correct this by correlating the economic freedom index with GDP growth, and (2) use as big a sample as is available instead of focusing on particular cases. I copied the freedom ratings from that Heritage web page and real GDP growth rates (“estimates are for the year 2013 unless otherwise indicated”) from Wikipedia. The correlation between the freedom index and real GDP growth turns out to be negative: the Pearson correlation coefficient is −0.19 and Spearman’s rank correlation coefficient (which allows for nonlinearity) is −0.35. Plotting the data and a loess curve with R’s default settings:
Some outliers are evident. Perhaps they’re disproportionately skewing the freedom-growth correlation? I take out North Korea (the far left point) and the two lowest points, Cyprus and Central African Republic:
and in fact the freedom-growth correlation sinks further, to −0.38 (Spearman’s rank) or −0.30 (Pearson). The main mass of nations hovers around the part of the loess curve which slopes downward.
At this point in time, the correlation between the Heritage Foundation’s index of economic freedom and economic growth is unambiguously negative. I suspect this is because being a poor country is associated with low economic freedom but high catch-up growth. What happens if I correlate the freedom index with annualized GDP growth over a longer period, 1990-2007, for which catch-up growth is probably less important?
I now wind up with positive correlations (+0.25 for Spearman, +0.20 for Pearson). Now I throw out the outlying North Korea (the leftmost point), Zimbabwe (the bottommost), and Equatorial Guinea (the topmost).
This has no meaningful effect on the correlations, which become +0.24 (Spearman) and +0.21 (Pearson). Overall, by switching from a more up-to-date growth statistic to a more long-term (and pre-Great Recession and mostly post-Soviet) statistic, I change the sign of the correlation between growth and the Heritage Foundation’s assessment of economic freedom.
It’s not immediately obvious to me which GDP growth statistic is more appropriate. 2013 growth has the advantages of being more up-to-date and better matching when the Index of Economic Freedom was calculated, but is less representative of each country’s long-term economic trajectory. 1990-2007 growth is more representative but also involves comparing 24-year-old data to a recent index; the proper thing to do here would be to use a similarly long-term average of the Index of Economic Freedom, but that’s too much like real work.
There is also the question of how to operationalize “economic freedom”, but this comment is long enough. I evade that problem here by simply taking the data tables suggested upthread as given, and after doing so the basic conclusion seems to be that the correlation between “economic freedom” and “economic growth” is modest, with its sign sensitive to how one operationalizes growth. (Check my work with my data file.)
[Edited 19⁄10 to change “Gunea” to “Guinea”, and add “after doing so”.]
To test this hypothesis, I did a linear regression of overall score, each of the ten subscores and 2013 real GDP growth against the log of 2013 GDP per capita (at parity). I then took the correlation between the residual of 2013 real GDP growth and the residual for each of the scores. Here are the results: overall score −0.04 property rights −0.15 freedom from corruption −0.11 fiscal freedom 0.25 government spending 0.18 business freedom −0.03 labor freedom 0.05 monetary freedom −0.12 trade freedom 0.01 investment freedom −0.18 financial freedom −0.09
These results were approximately opposite of what I expected (I expected minimal correlation for fiscal freedom and government spending and generally positive correlations for everything else). While I’m only somewhat surprised by the government spending and fiscal freedom results, I find the others very confusing. Does anyone have any idea what might be going on?
The analysis I’d do would be simpler. Compute the correlation of log GDP per capita with the freedom index (or its subscales); if I’m right it should be substantially positive. Then correlate log GDP per capita with GDP growth; the result should be substantially negative. Taking correlations of residuals addresses the different question of whether unusually high growth for a country’s income level correlates with unusually high freedom indices for a country’s income level.
I did the simpler analysis first and all the correlations between log GDP per capita and all the economic freedom index subscores were pretty negative (as was the correlation between log GDP per capita and GDP growth). Log GDP per capita was positively correlated with economic freedom subscores.
Edit: clarity
Thanks. A negative correlation between log GDP per capita and the freedom index surprises me; that falsifies my “poor country” confounder speculation.
The comment above yours was not very clear. I have edited it for clarity. There is in fact a positive correlation between the economic freedom index and log GDP per capita.
I’m more confused now. The parent comment says the EF index correlates positively with log GDP per capita, while the edited comment says the EF index subscores correlate both negatively and positively with log GDP per capita. I don’t understand how that can all be true simultaneously...
Your charts graph 1990-2007 economic growth as a function of 2014 economic freedom, 1990 economic freedom, so assuming that correlation is causation here (almost always a dubious assumption), this would indicate that economic growth leads to economic freedom, not the other way around.
Is 1990 economic freedom data available?
My bad. The oldest data is from 1995.
That site has a nice slightly-interactive map where you can pick out individual components of their “freedom index”. Mostly they correlate with prosperity (I have no idea what the actual causal relationships are) … until you click on “Government Spending” and suddenly it goes exactly the other way round—the allegedly-worst government spending figures are for the US, Canada and Western Europe, and the allegedly-best are for severely messed up central African countries and China (!) and India.
If they stopped counting government spending as opposed to freedom—it seems to me only marginally a matter of freedom—the correlation between “freedom” and prosperity would become even more impressive.
(Note 1. The cynic in me says: Of course that’s out of the question because a central part of the reason why the Heritage Foundation exists is to argue for lower government spending and hence lower taxes. If it advocated less forcefully for that, it would become less useful to those who fund it.)
(Note 2. It seems like there are lots of other things that could go into a “freedom index” with about as much reason as government spending. Two examples: longer working hours mean less freedom to do as you please with your time; stronger IP law means less freedom to start a technology-based business, to do as you please with the books and music and software you own, etc. Again, the absence of these things from the Heritage Foundation’s “freedom index” seems adequately explained by the interests of the organizations that provide its funding.)