Thursday, July 29, 2004

previous entry | main | next entry | TrackBack (4)

Does a fear of hell lead to economic growth?

Timothy Perry links to a paper by two Federal Reserve Bank of St. Louis economists suggesting that religious piety (operationalized as a fear of hell) could contribute to economic growth. The key section:

There might, therefore, be two parts to the link between religion and economic growth: a belief in hell tends to mean less corruption, and less corruption tends to mean a higher per capita income. The first part of the link is illustrated by the first chart below. It uses the 1990-1993 World Values Survey, which asked people in 35 countries whether they believed in hell, and the Corruption Perceptions Index produced by Transparency International, which surveyed many countries’ residents about corruption.9 The first chart plots the rankings of 35 countries’ percentages of people who believe in hell against the rankings of the countries’ perceived levels of corruption. As the chart shows, there is a tendency for countries in which a larger percentage of the population believes in hell to have lower levels of corruption.

The second part of the link is illustrated by the second chart, which plots the GDP-per-capita rankings of the 35 countries against their corruption rankings.10 It shows a strong tendency for countries with relatively low levels of corruption to have relatively high levels of per capita GDP.

Combining the stories from the two charts suggests that, all else constant, the more religious a country, the less corruption it will have, and the higher its per capita income will be. Of course, these charts are only suggestive. However, they are nonetheless consistent with Weber’s argument and the Barro and McCleary result that religious beliefs can influence economic outcomes.



The graphs would seem to be convincing -- except for the fact that the authors omitted a discussion of any direct correlation between a fear of hell and per capita income in their data. There's a good reason for that -- when you crunch the numbers, it turns out there's a correlation coefficient of -.21 between the two variables, which means there's a very weak negative correlation between a fear of hell and income status.

The authors' hypotheses might be correct, because this kind of correlation is not a ceteris paribus test. But the aggregate effect would seem to be pretty weak.

Another thing -- for a paper concerned with economic growth, it's odd that they're using GDP per capita instead.

Readers are invited to suggest alternative ways to test this hypothesis.

UPDATE: Interesting -- it looks like the authors have eliminated all the graphical evidence. And now there's an editor's note that explains:

It is the second revision that has been posted. In both the original version and the first revision, the article ended with a discussion of simple correlations between countries’ religiosity, levels of corruption and per capita incomes....

Thanks to the keen eyes of a number of readers, however, we have discovered that the charts used in both of these versions of the article contained errors. Consequently, the version below does not include discussions of the correlations between religiosity, corruption and per capita income. It is important to note that this has no bearing on the results in the literature that are discussed in the article.

Kevin Drum is less kind than the editor: "In other words: this was just simplistic crap and it wasn't even computed correctly at that."

This has not stopped media coverage of the paper. Greg Saitz wrote it up in the Newark Star-Ledger, but bless his heart, he was smart enough to ask some atheists about it:

"I cannot imagine what the belief in mythological beings or things that don't exist can do for business," said Ellen Johnson, president of Cranford-based American Atheists. "What about the pornographic industry? That is probably very good for growth."

Of course, Glenn Reynolds would reply that the consumption of pornography does not necessarily lead to antisocial behavior.

[You started with piety and ended with porn -- you are so going to hell!!--ed.]

posted by Dan on 07.29.04 at 12:48 PM


The link is indirect, but fairly clear.
Commerce of any kind, requires a certain level of trust between the participants. Moral behavior engenders trust. More trust, more commerce.

Now, I'm sure I'll hear arguemnts that religion is not the only source of morality, and that's arguably true. But I don't think many will argue they are not symbiotes.

posted by: Bithead on 07.29.04 at 12:48 PM [permalink]

Hummm. I remember a book by R.H. Tawney, "Religion and the Rise of Capitalism" (1922) that argues the case much more broadly than "fear of hell."

Forget about testing so narrow a hypothesis. Focus instead on the broader framing in which such a hypothesis might be put forward. Then you might find it is folly to even seek empirical evidence for such.. dave.

posted by: Dabbler Dave on 07.29.04 at 12:48 PM [permalink]

Hey, lets try this on individual states and see what happens - thats a good test. Heh.. I wonder what that outcome will be?

Just eyeballing that first graph will tell you this conclusion is very likely BS. I'd like to see an ANOVA table on this.

posted by: mickslam on 07.29.04 at 12:48 PM [permalink]


posted by: Carleton on 07.29.04 at 12:48 PM [permalink]

This kind of approach requires a mediation analysis, in order to make that kind of inference:

"Generally speaking, mediation can be said to occur when (1) the IV significantly affects the mediator, (2) the IV significantly affects the DV in the absence of the mediator, (3) the mediator has a significant unique effect on the DV, and (4) the effect of the IV on the DV shrinks upon the addition of the mediator to the model. These criteria can be used to informally judge whether or not mediation is occurring, but MacKinnon & Dwyer (1993) and MacKinnon, Warsi, & Dwyer (1995) have popularized statistically based methods by which mediation may be formally assessed." (see link for source)

And that's not even getting into the issues of causality...clearly declining economic prospects could just as easily lead to increased religious attendance (e.g. belief in a better afterlife).

Seems like very junky science, with a religious agenda behind it.

IMHO, a real study would look at the impact of personal ethics on corruption level and then disentangle all the sources of personal ethics (education, religion, enforcement, history, professionalization of government, etc...) I don't think an aggregate level model with 35 countries will tell us much of value.

posted by: pgm on 07.29.04 at 12:48 PM [permalink]

I heard there was a correlation between confucian values and high growth rates, over the past 20 years.

posted by: praktike on 07.29.04 at 12:48 PM [permalink]

I'm no stats nerd so let us test the hypothesis removing modern day variables.

It could be argued that fear of hell was at its peak during the middle ages, hardly a time of soaring productivity and GNP growth. Whereas regions like Asia (where hell was not a concept) experienced a golden age by comparasion.

posted by: Fledermaus on 07.29.04 at 12:48 PM [permalink]

What do you suppose the 'fear of hell' quotient is in Iran, or Pakistan? Gee, I wonder if those two states have as much corruption as secular France. Even more, you think? 'Fear of hell on earth' certainly seems to correlate positively with corruption.

I would suspect that corruption thrives most where the actions of the powerful can escape question behind a shield of superior wisdom and pure intentions (if not just superior firepower), whether that monopoly of power is justified in religious or secular terms.

posted by: Joel on 07.29.04 at 12:48 PM [permalink]

Sorry, but this study you link to is probably the most ridiculous thing I've read in a while.

Pretty much ALL the countries on the "belief in hell" scale that have a low belief in hell AND a relatively low GDP per capita are formerly communist countries.

If we didn't know it already, we could perhaps use this study to postulate that communism and a lack of belief in hell are somehow correlated. Gee, thanks for that deep insight.

We know that communism is fiercely anti-religious. We know that Poland remained a rather strictly Catholic country as pretty much the only exception from that rule. We also know that communism generally leads to a low GDP per capita. Which explains that part of the data.

For the other side of the data, we know that religious belief is generally stronger in developing countries than in developed countries - with the notable exception of the USA and strongly Catholic countries like Poland and Ireland.

Apart from all that I found it curious that they used the "ranks" rather than the actual values to draw their correlation diagrams. The selection of countries also seems rather arbitrary - they could have included quite a few more ex-communist countries, which would have strengthened the correlation between little fear of hell and low GDP per capita even more. Or they could have included dozens more African countries like Nigeria with a relatively high belief in hell and also a rather low GDP per capita to prove the opposite point.

posted by: gw on 07.29.04 at 12:48 PM [permalink]

Hey, lets try this on individual states and see what happens - thats a good test. Heh.. I wonder what that outcome will be?

Have you looked at California, of late? If I'm not mistaken, they're in a business and fiscal crisis out there these days. I think you'll be able to find some who can point to a death of morality out there of late.

What do you suppose the 'fear of hell' quotient is in Iran, or Pakistan?

while what exists is done in the name of 'religion', I'm not convinced many would call it 'moral', even those involved directly in that culture.

posted by: Bithead on 07.29.04 at 12:48 PM [permalink]

If you bias your sample, you can prove anything. Their sample includes only two Asian countries (Japan, India) and two African countries. None from the Middle East. If there were as many Middle Eastern countries on that list as there European countries, results would probably have been completely the opposite.

posted by: Mike K on 07.29.04 at 12:48 PM [permalink]

Is it too late to invest in Haiti, given this news?

posted by: Greg Wythe on 07.29.04 at 12:48 PM [permalink]

Bithead writes: "Commerce of any kind, requires a certain level of trust between the participants. Moral behavior engenders trust. More trust, more commerce."


Pray tell how you measure the moral behavior of a corporation without being privy to all it policies and practices?

Lets try self-interest and profit:

and to echo a dead president's theme:

Trust and verify!


Caveat Emptor!


Trust but only with a multi-page contract and the threat of litigation.

On the whole "business" is more amoral than moral.

posted by: j swift on 07.29.04 at 12:48 PM [permalink]

Rebuttle: The Philippines.

90% Christian, 70% members of the Catholic Church and a near destitute 3rd world country.

In fact the Philippines proves just the opposite, they are all afraid of going to hell and populate like rabbits because of that fear. All of which removes any possibility of the government making things better.

posted by: Marc on 07.29.04 at 12:48 PM [permalink]

I agree there are too many confounding factors to get much out of this. GW's example of ex-communist countries is one excellent example.

Because of my own experience in parts of Eastern Europe, the premise does not seem absolutely ridiculous to me. Romania was and is a corrupt society. The fact that the Orthodox Church was heavily infiltrated with spies and shills certainly gums up the data whether religious teaching had much sincerity behind it, hence it is hard to know how "religious" the country is (Even if we knew what we meant by "religious.")

The smaller and persecuted denominations, however, tend to have lower rates of incarceration now that they are legal. Is this because they were more sincerely religious, because they now have something to prove, or an emotional response to increased freedom? I don't see how we measure that.

Yet the fact remains. They have a lower rate of incarceration. Surely it signifies something.

Measuring changes over time might be more fruitful, but even at that, if there is a delayed effect, we will be unable to see it clearly. That is often a problem in measuring economic impact. It is likely that changes in education policy, tax policy, trade barriers, or gov't job creation affect the economy on different time scales. Do we measure the effect of a tarriff change in years 1-3 or in years 2-6? Dunno. Measuring religious changes, corruption changes, and economic changes, are all going to have these conflicting delays built into them.

And does it matter which religion? Ooh, it's just to messy to get one's mind around.

posted by: Assistant Village Idiot on 07.29.04 at 12:48 PM [permalink]

The Assistant Village Idiot writes, apropos of incarceration rate variations: "Surely it signifies something."

Well, not necessarily. In any statistical analysis, we deal with the probability of accepting false hypotheses or rejecting true hypotheses. Which means it might not signify anything. And, anyway, as some religious propaganda or other says, "Time and chance happeneth to us all."

posted by: Donald A. Coffin on 07.29.04 at 12:48 PM [permalink]

If you break down GDP / head by state (not counting D.C.), you get:

- Top 5:
New Jersey
New England

- Bottom 5:
New Mexico
West Virginia

'Nuff said...


posted by: babel on 07.29.04 at 12:48 PM [permalink]

Coffin says, "In any statistical analysis, we deal with the probability of accepting false hypotheses or rejecting true hypotheses." I find that economists often fall into a completely different category of error:

"Back in 1968, the statistician Howard Raiffa proposed a third type of error, the Type III. A Type III Error is solving the wrong problem precisely."

posted by: Dabbler Dave on 07.29.04 at 12:48 PM [permalink]

Obviously the "editor" believes in hell - raise his salary.

posted by: Jody on 07.29.04 at 12:48 PM [permalink]

Pray tell how you measure the moral behavior of a corporation without being privy to all it policies and practices?

Oh, please.
With all the liberal Democrats running around, dripping socialist sweat all over every private transaction, looking for something dirty to prove their point?

posted by: Bithead on 07.29.04 at 12:48 PM [permalink]

The (extremely secular-) Scandinavian countries usually end up on top of these kinds of surveys as they have very low levels of corruption.


posted by: Marcus Lindroos on 07.29.04 at 12:48 PM [permalink]

The French are as secular as the Scandinavians and are far more corrupt. The key determinant of corruption is the degree to which patronage, familial and other clan-like structures-- as opposed to individualism-- dominate social and commercial relations.

This explains the apparently odd convergence of Catholic and communist nations on the far end of the corruption scale. Italy, perhaps the most indelibly corrupt society on the planet, represents the perfect fusion of Catholicism's extreme emphasis on paternalistic, hierarchical, family-centered "belongingness" and communism's paternalistic, hierarchical, party-centric "belongingness." Bribery, fraud and kickbacks suffuse most official transactions in Italy. No surprise that the Latin Americans are probably next to the Italians on the corruption scale: look for a latinate Catholic society with very strong trade unions and a patrimonial economic heritage and you'll find massive corruption, without exception.

Spain, interestingly, has become much more individualist and less Catholic during the last 20 years, mainly as a reaction to Franco's brutal and stifling rule; hence one would expect (I don't have any data re Spain) Spain in the age of Almodovar and Aznar to be significantly less corrupt than Italy or Mexico.

The French are somewhat less corrupt than the latins because under French socialism, meritocracy and feminism have made huge inroads on the family. To get ahead in France, it's not enough to be a fils de papa; you also need to get into ENA or HESEC, and test scores determine admission there.

The Irish might be expected to be as corrupt as the Italians, but in point of fact Irish families tend to be far weaker than the latin families because of the high incidence of alcoholism and neglect found among Irish fathers. Paternalistic figures in Ireland command far less sway than they do in Mexico or Brazil or Italy.

Close to the (ex-Spain) latins would be the ex-communist nations, whose political and social structures remain heavily influenced by clans comprising ex-communist pols, mafiosi, legitimate businessmen, and security officials. In Italy, one dentifies oneself as a son of "I'm from Ferrara or Le Marche or Venizia"; in Ukraine, you're a son of Kuchma; in Czech, you're one of Klaus's boys; in Russia, you're either affiliated with Fridman and Aven (Alfa), or Berezovsky and Yeltsin, or Chubais and Yeltsin, or Khodorkovsky, or... you get the idea.

At the far end of the scale are those societies in which individualism weighs more heavily than family ties; in which the economic legacy is one of competition rather than patrimonialism; and in which geographic mobility is very high and access to local power structures matters little to economic outcomes.

In sum, the most corrupt societies are those where people routinely refer to themselves less as individuals forming associations based on common principles than as members of small, clannish groups defined by paternalistic figures hailing from a specific region (Napoli, Rome, Yekaterinburg, Nizhny Novgorod, Arkansas, south side of Boston).

posted by: augustine on 07.29.04 at 12:48 PM [permalink]

[You started with piety and ended with porn -- you are so going to hell!!--

No, Dan... You started as you say with piety, and ended with Glenn Reynolds... and were I Glenn, I'd be wondering about that reletive positioning.


posted by: Bithead on 07.29.04 at 12:48 PM [permalink]

Actually, consumption of porn would probably be very negatively correlated with corruption, as porn consumption seems to be highest in the individualistic, non-patrimonial countries and lowest in the patrimonial, latinate Catholic ones.

Highest porn consumption per capita is certainly the US, which is probably the least corrupt; next highest porn consumption would probably be Germany, perhaps the UK. Lowest porn consumption is probably Italy or France. The East Europeans also tend to have very little interest in porn (though the Czechs especially export a lot of it).

Wasn't there a famous court case in Utah in which a video rental store owner used, as his defense, statistical data showing that the highest consumers of adult videos from his store were from ZIP codes which also contained the most deeply religious Utahans? And aren't Mormons the least corrupt of all American social groups?

Is it high corruption that diminishes the need for porn, or high porn consumption that diminishes the need for corruption?

posted by: augustine on 07.29.04 at 12:48 PM [permalink]

"Is it high corruption that diminishes the need for porn, or high porn consumption that diminishes the need for corruption?"

Probably the latter.

If it turns out that a belief in hell does have positive economic effects, I'd figure that that is a special case of long-term thinking winning out over short-term thinking. In the long term, crooks, corrupt politicians, and despots do get their comeuppance; the problem is that too many of them die first.

So a working anti-aging treatment, along with its other salutary effects, would also induce long-term thinking in the population at large and in the leaders, and discourage corruption since most people will live long enough to face the consequences.

posted by: Ken on 07.29.04 at 12:48 PM [permalink]

But in the most corrupt countries, there are no malign consequences to individuals for corruption. In fact, often the opposite is true: if Mexican political and trade union officials were not corrupt, very little official business would get done. Corruption is the glue that holds the Mexican polity together.

Hell and the fear thereof has f-all to do with corruption, folks. Reminder the three p's (and get the correlations straight):

Patrimonial economies produce a legacy of corruption; individualistic capitalist ones produce less corruption.

Paternalstic and clannish social cultures produce high corruption; individualistic ones with strong feminist movements produce less corruption.

Porn consumption fills the aching need to slither in the mud created by the less corrupt, highly individualistic societies.

High papa coefficient --> high corruption + low porn consumption, and vice-versa.

posted by: augustine on 07.29.04 at 12:48 PM [permalink]

For all those commenting on corruption and the fear of hell there is a catch. In the graph, the corruption risk is actually lower as the numbers grow. The so called economist had ranked the highest risk country as #1. The result therefore is there is more corruption if you believe in hell.

posted by: mustafa karhat on 07.29.04 at 12:48 PM [permalink]

If you want to see real predictors of GDP-per-capita, try religiosity, TIMSS scores, or (most importantly) mean IQ (behind subscription wall, but full text from UK Telegraph available here if you scroll down:

A COUNTRY’S prosperity is closely related to the average IQ of its population, according to research that has mapped global intelligence levels.

The study of 60 countries identified a clear correlation between assessments of national mental ability and real gross domestic product, or GDP.

Richard Lynn, Professor Emeritus of Psychology at the University of Ulster, and Tatu Vanhanen, Professor Emeritus of Political Science at the University of Tampere in Finland, tested the non-verbal reasoning abilities of a representative sample of the different populations. They found that the countries of the Pacific Rim had the highest intelligence scores: Japan, South Korea, Taiwan, China, Hong Kong and Singapore averaged IQs of about 105.

The next brightest were the populations of Europe, the United States, Canada, Australia and New Zealand, averaging 100. In South Asia, North Africa and most Latin American countries, the score was an average of about 85, and in sub-Saharan Africa and the Caribbean closer to 70.

Set against international measures of economic performance, the academics concluded that about 58 per cent of the differences in national wealth could be explained by differences in intelligence. Each average IQ point above 70 was worth about £500 in GDP per head of population. The report says that people with high IQs can acquire complex skills to produce goods and services for which there is international demand.

Religiosity is strongly negatively correlated with GDP-per-capita, IQ, and TIMSS scores. The latter three are all positively correlated. And strongly so (whether you use rank or standard correlations).

posted by: gc on 07.29.04 at 12:48 PM [permalink]

The original Fed article was plain wrong - it was based on a simple mistake. They got the "Fear-of-hell" rank back to front.

See my comments posted a few days ago on the Elite Trader web site:

posted by: Garry on 07.29.04 at 12:48 PM [permalink]

Why are catholic countries poor and corrupt?

posted by: Dreams and Visions on 07.29.04 at 12:48 PM [permalink]

Post a Comment:


Email Address:



Remember your info?