The irrationality of the public sector in oil rich countries

Behavioral economist have largely focused on the role of irrationalities in the market. An implicit argument is often that the irrational individual is argument for the rational state to play a larger role.

But the state is not made up of divine angels or logical super-computers, it is also constituted of flawed individuals. Politicians and the public they serve in the political system are as prone to irrationalities as much firms and consumers in the market. Sometimes this problem is even larger, because while in the market the individual get's the entire benefit of making rational decisions, as a voter or member of an interest group they get virtually no personal benefit of being rational, causing a collective action problem in public decision making.

I want to illustrate a clear example of public sector irrationality: the price of gasoline in oil rich countries. What is extra interesting is that these countries are mostly non-democracies. Of course even in non-democracies the state is responsive to public opinion, and it seems in this example that the quality of public and state decision making is particularly low in non-democracies.

The price of gasoline in the U.S in 2009 was about $2.7 per gallon ($0.7 per liter). Since taxes on gas are low in the U.S, this is close to the world market price. In contrast, the cost of gasoline in Saudi-Arabia was $0.6 per gallon, in Kuwait $0.87 per gallon, in Venezuela $0.19 per gallon!

This graph illustrates oil reserves per capita, and the price of gas. What you see is that the more oil you have, the cheaper is gas.


Isn't that what economics predicts? Higher supply means lower price?
No! In a integrated world market, there should be no relationship between quantity in any individual point and price.

This is literary a textbook example from economics of oil rich countries ignoring the opportunity cost.

Oil is traded in the world market, so the cost for oil rich countries in giving cheap oil to their citizens is exactly the same (disregarding minor transportation costs) as countries with no oil. While oil prices are kept up by OPEC, the opportunity cost of giving the oil cheap to your population is waiting and selling it expensive to other countries.

Of course part of the reason they subsidize oil instead of taxing could be that they are richer. I did a regression controlling for per capita income: the result is the same. The more oil you produce and the more oil reserves you have, the lower the price of gasoline.

What happens in these countries, such as Iran, is that the public gets angry if gas is expensive "we have so much oil, why should we pay for it?".

The big exception that proves the rule is rational and well-functioning Norway. Norway has as high, or even higher taxes on gas as other countries. Furthermore, unlike most oil rich nations, Norway saves it's revenues from oil and gas, rather than spending most of it.

The artificially low price of gas in Iran is part the reason they need to import refined petrol. Read this excellent report by Atta Tarki if you want a good case illustrating gas subsidies.

Ironically Ahmadinejad is finally reforming its massive and hugely wasteful gas subsidies as part of the military confrontation with the U.S.

Score one more for Chicago economics

His policies may not reflect it, but President Obama privately took Milton Friedman's life cycle-Permanent Income theories seriously.

http://www.nydailynews.com/opinions/2009/05/04/2009-05-04_president_obamas_mantra_in_debt_we_trust.html#ixzz0FLMXeUT6&B

“A close examination of their finances shows that the Obamas were living off lines of credit along with other income for several years until 2005, when Obama's book royalties came through and Michelle received her 260% pay raise at the University of Chicago.”

When their future income went up, but before their current income went up, the Obamas increased their consumption. Textbook case of Milton Friedman's permanent income hypothesis.

The Super-Economy in one picture

We noticed earlier that when Sweden is compared to the U.S, Sweden does better for some part of the income distribution. However, if we want to compare systems, we should look at comparable population groups. For this reason I have graphed Swedish income compared with income of Americans with Swedish ancestry.

Read my previous post on my method, which combines GDP per capita data with income distribution data. Read an earlier post on the rationale of using Swedish-Americans rather than all Americans as the comparison group to Sweden.

The graph is income per capita in Sweden and the U.S (for Americans with Swedish ancestry) for 10 income groups, based on official Swedish statistics and census data. I define you as American with Swedish ancestry if the main ancestry group is Swedish. Americans with Swedish ancestry have a 55.8% advantage in income compared to people in Sweden; very close to the figures I estimated using similar underlying numbers and a somewhat different methodology (is a good sign).

The results of the comparison is striking.



Swedes under the American small-government system beat Swedes in the Swedish welfare system for almost 90% of the income distribution. Among the first 10th percentile the Swedes in Sweden do better. By the 15th percentile or so the Swedes in the U.S have caught up, and vastly outperform Swedes in Sweden for the rest of the income distribution.

Here is the same exact picture with income differences filled in.


The median is 42% higher for the Americans compared to the Swedes.
Overall, the middle 60% of the population earn 46% more in the American Super-Economy versus welfare state Sweden.

Now, these figures are slightly unfair to Sweden, because we are comparing ethnic Swedes in the U.S with total Sweden, which includes immigrants (many of whom have not done well in the job-market). For the aggregate figures this is not important, since immigrants are few. Non-immigrants earn 3% more than average of Sweden, simply since most of the average is made up of non-immigrants. The poor in Sweden are disproportionally immigrant however and I would guess Sweden would do even better in the first and second decile if we excluded immigrants. Nevertheless, the advantage in earnings for Americans is massive, indicating that there are large losses in income associated with having a welfare state.

The claims of the left, such as the claim that welfare state is better for ordinary people, are the results of incorrect comparisons. Sweden is homogeneous, whereas the U.S is not. Second, Swedes are a uniquely productive people (as indicated by their performance under the U.S system, and Sweden’s performance before the expansion of the welfare state), which should be taken into account.

The fact that the aggregate results are not even better is due to homogeneous Sweden having innate advantageous in income distribution. Most of the problems of the U.S are concentrated a few distinct populations and geographic areas (such as Appalachia). When we make apples to apples comparisons, it is clear that the American system is better for almost everyone, and almost as good even for the very poor.

The Swedish system is a losing proposal for almost all the population. Unless someone spends their entire life in the bottom 10% of the population, they are better off under the American system.

Here are the lessons from this comparison:

* The U.S would be even richer and have much less poverty if it was made up entirely of Swedes.

* Sweden would be richer than the U.S if Sweden adopted American institutions.

* A large (Swedish-size) welfare state with an American population would do much worse than Sweden.

If you want one picture to demonstrate that free markets and limited government are better for almost everyone in society, this is it.

Income distribution in the U.S and Sweden

We know Americans do better on average. But how about the distribution? After all, the left like to claim that in the U.S only a few hedge fund managers and the like are well off, while the masses live in poverty.

I used the Census American Consumer Survey to make American income data comparable to Swedish income data. Swedish income distribution data adjusts household income to household size (with children having less weight). I further use OECD data to adjust for purchasing power. The data for Sweden is the latest available year, 2007, whereas the U.S data is for 2006-2008.

I think this is a pretty good, but of course not perfect, comparison. You should think about it as comparing GDP per capita taking income distribution into account. More about the method in the end.


Up until around the 45th percentile, Sweden does better than the U.S. After this the U.S opens up a substantial advantage. It is clearly better to be poor in Sweden compared to the U.S, and obviously to be rich in America compared to Sweden.

What matters most is that this graph illustrates that it is better to be middle class in America. The 60% in the middle earn 20% more in the U.S than they do in Sweden, even taking government purchases crudely into account. It is a myth that only a few at the top do better in the American system compared to even arguably the most successful of the European welfare states.


Here is the same data with the differences filled in.


This comparison up until now was between Sweden and all the US. That is not the correct comparison. A better comparison, if we think culture and norms matter, is between Swedes in the Swedish system and people of Swedish ancestry in the American system. I will make those comparisons in the next post.




Method: I combine differences in GDP with distribution data, in terms of differences in income for 10 deciles. Swedish income distribution data adjusts household income to household size (with children having less weight), so I did the same for the American data, giving children the weight 0.35. I further use OECD data to adjust for purchasing power. The data for Sweden is the latest available year, 2007, whereas the U.S data is for 2006-2008.

The comparison is close, but not perfect. Fringe benefits are lacking. The U.S data does not include income taxes and fringe benefits, and many government transfers. The Swedish data is with capital gains. The American income data is top coded. Most importantly, the Swedish data does not include the value of government purchases. The average advantage in household income is 42.6% for the U.S, compared to a 24.3% advantage in GDP.

For a thing such as health care, the bias goes both ways, since American worker’s health care is largely paid for by the employer and not included in income. This effects the distribution of income however, since poor workers in the U.S get less health care than rich workers, whereas they get equal or more in Sweden.

To account for all these problems and in order not to bias the comparison against Sweden, I did a crude adjustment of increasing the Swedish average data to reflect government services, so that the average difference is now equal to average difference in GDP. What is great about GDP is that it includes almost everything. Now, for a perfect comparison for the U.S and Sweden we would need to know exactly how much government cervices each income groups gets, data that is very hard to calculate. Another problem is that this assumes that 1$ spent by the state is worth 1$ spent by the individual, where I believe it is worth somewhat less, something that benefits Sweden in the comparison. But since much of government spending has already been taken into account, and since most of the remainder does not differ significantly in the U.S and Sweden in terms of who it is targeted towards (in both countries the poor get more benefits than the rich), the comparison is likely to be a reasonable reflection of the economic status of various groups.

Mormons have higher I.Q than average

As I noted, there is a discussion regarding if atheists are smarter than people from other religions. From the NLY97 I have now included how people with different religions score on aptitude tests, often used by researchers as a proxy for I.Q.

(Of course these are just test score, neither I nor anyone else can guarantee that they are not biased for any religion. Take them for what they are worth.)


It is well known that Jewish people score very high in I.Q tests. What is less well known is that Episcopalians also get astronomical I.Q scors.

Note again that new age types with "personal philosophies" rather than pure atheism are not smarter than average.

The most important thing I note is that Mormons are smarter than average. If you are going to go around saying that atheists are smarter than average, you should know that the same tests show Mormons to be smarter than average as well.

Mormonism is not just a religion. It is a cohesive and self-replicating cultural identity, with strong positive effects on behavior and outcome for its members.

Does the lack of economic freedom kill you?

A regression of life expectancy on Gini was, as we established, not statistically significant for the 28 advanced OECD countries, or the 21 (or sometimes 23) that The Spirit Level uses.

But why get bugled up in country selection? Let us include all countries the U.N gives data for! The only problem is that some countries have short life expectancy because they are poor. In order to correct for this, I control for income, which the UN conveniently provides.

In a regression of all 140 countries with data the coefficient for Gini is not statistically significant (p 0.169). GDP is statistically significant, (p-value 0.000).

I also did the same regression, now looking at the association between life expectancy, per capita income, and the Heritage Index of economic freedom.

In a regression of the 169 countries for which there is data, both per capita income AND The Index of economic freedom are statistically significant (p value for both 0.000).

There are more countries with missing data for Gini than the index of economic freedom. If we only look at the 140 countries where we have data for Gini the exact same results hold (pvalue for the Index of Economic freedom now 0.002).

So unlike Gini, the Index of Economic freedom has a statistically significant association with life expectancy, controlling for per capita income, in the full UN sample.

Can we now claim that we have proven that free-market policies increase life expectancy? Or that the lack of economic freedom murder people?

OF COURSE NOT. We only have a correlation, not a causation. There are a million reasons why these two variables could be related. Maybe more democratic countries have better health policies, and also more free-market policies. Maybe more ethnically homogenous countries have less health problems and more free institutions. Maybe there is reverse causality, with more healthy people having better institution.

If you go around with weak evidence, not having established causality, and screaming that you have proven that the lack of Economic freedom kills people, you are an ideologue, and not a scientist.

This example illustrates the two problems of the Spirit Level. First, their basic correlations are much weaker than they give the impression of, and often not even statistically significant. Second, they have not a shred of evidence in terms of inequality actually causing bad outcomes.

Are atheists smarter than average?

There is a discussion about religion and IQ, specifically that of atheists. Are atheists smarter than others?

As luck would have it there are great datasets that can answer questions like this. The NLSY79 and NLYS97 have one of the most accurate measures of general intelligence for large datasets (the surveys are designed so that we can get nationally representative results from them).

From the NLSY97 I have the IQ, and religion as measured in 2005 (where most of the sample are in their 20s).

Atheists and Agnostics have far higher IQ than the average. On average atheists/agnostics have 8 I.Q points higher than the national average, which is half a standard deviation.

However, and on closer inspection not surprisingly, people who don’t believe in god and who identify as instead having “a personal philosophy”, are around average in intelligence. This group is much larger than atheists and agnostics.


Obviously the fact that smart people don’t believe in god does not prove that god does not exist either way.