Thomas Friedman points out that a disproportional number of the finalists of 2010 Intel Science Talent Search are of immigrant origin. He puts up these names:
"Linda Zhou, Alice Wei Zhao, Lori Ying, Angela Yu-Yun Yeung, Lynnelle Lin Ye, Kevin Young Xu, Benjamin Chang Sun, Jane Yoonhae Suh, Katheryn Cheng Shi, Sunanda Sharma, Sarine Gayaneh Shahmirian, Arjun Ranganath Puranik, Raman Venkat Nelakant, Akhil Mathew, Paul Masih Das, David Chienyun Liu, Elisa Bisi Lin, Yifan Li, Lanair Amaad Lett, Ruoyi Jiang, Otana Agape Jakpor, Peter Danming Hu, Yale Wang Fan, Yuval Yaacov Calev, Levent Alpoge, John Vincenzo Capodilupo and Namrata Anand."
and concludes:
"Indeed, if you need any more convincing about the virtues of immigration, just come to the Intel science finals. I am a pro-immigration fanatic. I think keeping a constant flow of legal immigrants into our country — whether they wear blue collars or lab coats — is the key to keeping us ahead of China."
Is Friedman's list a logical argument in favor of current American immigration policy?
Well, let us note that fully 54% of the foreign born population of the U.S is from Mexico and the rest of Latin America. Only 9% are from India and China. Even if you don't ignore illegal immigrants, there are several times as many legal immigrants from Latin America as there are from India and China.
Yet, of the 40 finalists, not a single one seems to be from Latin America! (correct me if I am wrong. John Vincenzo Capodilupo sounds Italian). His list is almost entirely made up of Indians and Chinese kids.
In 2008 only 15% of legal immigration was based on employment or skills, the remaining 85% is skill unrelated things such as having relatives in U.S. This is Americas current immigration policy: take a few high skilled people and masses of lower skilled immigrants.
Thomas Friedman presents us with a very valuable piece of information, which is that high-skill immigrants are very innovative. He inadvertently illustrated that low-skilled immigrants are not very innovative. Americas current immigration policy is taking mostly low-skilled immigrants, and few high skilled immigrants.
Yet instead of drawing the logical conclusion from it, the NYT farcically wants to use the success of the high skilled as an argument in favor or continuing the current American policy of mass-low skill immigrant.
Monday, March 22, 2010
Sunday, March 21, 2010
Obama-care bad for America, good for republicans
Some people on the left have claimed passing health care will not hurt them in the 2010 midterms, because the damage in public opinion was already done. The prediction markets seem to disagree.
As the probability of health care passing went from 40% in the beginning March to close to 100%, the probability of Republican taking congress went from 42% in to 48% as of writing. Since the market's already calculated some probability of health care passing at this point, the marginal increase from the bill is 6/0.6=10% points.
Especially on Sunday there was heavy trading in the probability of republicans re-taking the house around the time the Stupak deal was announced.
Going by the results on Sunday, a 15% increase in the probability of heath care passing translated into a 2% increase in the probability of Republicans winning the house, implying that health care led to a 2/0.15=13% higher chance of Republicans taking over the house.
Some admit that passing health care was bad for congressional Democrats, but necessary for Obama. However the probability that Obama is re-elected did not increase from March to now, nor did it change on Sunday. If anything it went down a little.
The Democrats are still favored to hold the house. But if we believe the prediction markets (which I always do), passing health care has made it around 10% points more likely that republicans will take over the house.
As the probability of health care passing went from 40% in the beginning March to close to 100%, the probability of Republican taking congress went from 42% in to 48% as of writing. Since the market's already calculated some probability of health care passing at this point, the marginal increase from the bill is 6/0.6=10% points.
Especially on Sunday there was heavy trading in the probability of republicans re-taking the house around the time the Stupak deal was announced.
Going by the results on Sunday, a 15% increase in the probability of heath care passing translated into a 2% increase in the probability of Republicans winning the house, implying that health care led to a 2/0.15=13% higher chance of Republicans taking over the house.
Some admit that passing health care was bad for congressional Democrats, but necessary for Obama. However the probability that Obama is re-elected did not increase from March to now, nor did it change on Sunday. If anything it went down a little.
The Democrats are still favored to hold the house. But if we believe the prediction markets (which I always do), passing health care has made it around 10% points more likely that republicans will take over the house.
The ethnic divide in health care opinion
Some very interesting data in a new poll by Pew.
1. Obama has an approval rating of 46% and a disapproval of 43% (+3%) among all adults. Among registered voters, it is 45-45 (+-0%).
2. The support for Obama based on income is U-shaped. Those earning more than $100.000 and less than $30.000 support him the most, the middle class supports him the least.
3. There are massive racial gaps in Obama's support. Only 35% of non-Hispanic whites approve of the job Obama is doing as President, compared to 85% of African Americans (not surprising), and 61% of Hispanics.
I exclude those who do not answer and compare this with Obama's results in the exit poll reported by CNN:
Loss of support:
Non-Hispanic Whites: -3.2%
African Americans and Hispanics: -2.2%
What is remarkable is if we don't calculate percentage points, but percentage.
Obama started with a paltry 43.9% among non-Hispanic whites. He is down 3.2% points, which is 7.3% of 43.9%.
Loss of Obama support among whites: -7.3%
Loss of Obama support among Hispanics: -5.1%
Loss of Obama support among African Americans: -1.6%
Let me remind you once again: these are percentage of total votes, not percentage points which are usually reported.
This confirms with the theory that Hispanic voters on the margin follow the same trends as white voters, just with a much higher mean for Democrats.
I also predict that regardless of what happens during the Obama-presidency, his support among African Americans in 2012 exit polls will not go below 90%.
4. The support for the health care bill is 38% among all adults, and only 48% among the uninsured. This last figure is really remarkable. Even the constituency Obama is trying to bribe do not love his bill.
5. There are staggering racial gaps regarding the health care bill. Only 30% of non-Hispanic white voters support the Democrat health care bill. Among African Americans the support is 67%, and among Hispanics 59%.
While net support is -28% for whites, it is +33% for Hispanics. Fully 30% of the uninsured are Hispanic. The demographic transformation is driving the U.S towards the left. This is both through making the electorate ideologically more leftist, and through increasing groups in need for government assistance.
My interpretation is that if it would have not been for immigration policy the last 40 years, the takeover of the health care sector voted for today would not have take place.
This is an important lesson for libertarians who support open borders. You have to choose: if you want left-libertarian immigration politics you have to accept the United States slowly turning into a social-democratic welfare state.
This dilemma is part of reality and must therefore be confronted, it is not something you can wish or assume away in a theoretical discussion.
1. Obama has an approval rating of 46% and a disapproval of 43% (+3%) among all adults. Among registered voters, it is 45-45 (+-0%).
2. The support for Obama based on income is U-shaped. Those earning more than $100.000 and less than $30.000 support him the most, the middle class supports him the least.
3. There are massive racial gaps in Obama's support. Only 35% of non-Hispanic whites approve of the job Obama is doing as President, compared to 85% of African Americans (not surprising), and 61% of Hispanics.
I exclude those who do not answer and compare this with Obama's results in the exit poll reported by CNN:
Loss of support:
Non-Hispanic Whites: -3.2%
African Americans and Hispanics: -2.2%
What is remarkable is if we don't calculate percentage points, but percentage.
Obama started with a paltry 43.9% among non-Hispanic whites. He is down 3.2% points, which is 7.3% of 43.9%.
Loss of Obama support among whites: -7.3%
Loss of Obama support among Hispanics: -5.1%
Loss of Obama support among African Americans: -1.6%
Let me remind you once again: these are percentage of total votes, not percentage points which are usually reported.
This confirms with the theory that Hispanic voters on the margin follow the same trends as white voters, just with a much higher mean for Democrats.
I also predict that regardless of what happens during the Obama-presidency, his support among African Americans in 2012 exit polls will not go below 90%.
4. The support for the health care bill is 38% among all adults, and only 48% among the uninsured. This last figure is really remarkable. Even the constituency Obama is trying to bribe do not love his bill.
5. There are staggering racial gaps regarding the health care bill. Only 30% of non-Hispanic white voters support the Democrat health care bill. Among African Americans the support is 67%, and among Hispanics 59%.
While net support is -28% for whites, it is +33% for Hispanics. Fully 30% of the uninsured are Hispanic. The demographic transformation is driving the U.S towards the left. This is both through making the electorate ideologically more leftist, and through increasing groups in need for government assistance.
My interpretation is that if it would have not been for immigration policy the last 40 years, the takeover of the health care sector voted for today would not have take place.
This is an important lesson for libertarians who support open borders. You have to choose: if you want left-libertarian immigration politics you have to accept the United States slowly turning into a social-democratic welfare state.
This dilemma is part of reality and must therefore be confronted, it is not something you can wish or assume away in a theoretical discussion.
Friday, March 19, 2010
Tea Party protesters do not seem to suffer from tax illusion, but Bruce Bartlett does
Fiscal illusion is a economic term for when the public has systematic misconceptions about the size of government, because of information costs. The error can go both ways, Anthony Downs believed it would lead voters to underestimate the benefits of government, whereas Amilcare Puviani predicted that taxes would be hidden and the burden of taxation underestimated.
For Sweden, more than half of all taxes are hidden (payroll taxes, VAT), and a study I conducted with Björn Wallace showed on a sample of about 1000 people that Swedes underestimate the tax burden on a typical worker by about 1 third (they believed 40% whereas at the time it was about 63%).
FrumForum conducted a survey of 57 tea party protesters, and asked them about the tax burden.
I would have guessed they would strongly over-estimate taxes, given their ideological bias. In fact the tea party people have a pretty good handle on taxes.
1. They were asked:
"What Does the Typical Family Earning $50KPay In Taxes?"
The mean answer was $12.710., or 25.4%.
Fromfurom and former conservative Bruce Bartlett wrongly claim that this figure is incorrect. In fact it is very close to the likely truth.
Fromforum absurdly claims that a typical family earning $50.000 only pays 15% in taxes, because they are in the 15% federal tax bracket. This claim is wrong on several levels. First of all, the questions asks about how much you pay, the average tax ("how much do you typically pay"), not the marginal tax ("how much of your income increase would you pay in taxes"). Second, they ask generally about all taxes, not just the federal income tax.
We are at a point where the biggest tax burden on low and middle income American's are payroll taxes, no longer the federal income tax. This is particularly if you include, and of course you must include, the half of the payroll tax that is nominally "paid" by the firm, but whose burden is on the worker. Lastly Fromforum may be interested to learn that there is such a thing called state and local taxes.
According to the IRS, the average amount of taxes ultimately paid by households earning 40-50.000 was 6.7% of adjusted gross income, and for households earning 50-75.000 8.1%. Let us say 7%. (Bartlett claims it is 1.7%, which I find too low).
In addition to this, someone earning 50.000 will directly pay 7.65% in Social Security and Medicare taxes, and indirectly another 7.65% in Social Security and Medicare taxes "paid" by the employer. It is unclear if Bartlett takes into account the payroll taxes nominally on the employer whose incidence is on the worker. He should.
Lastly we have state and local taxes, which Bartlett completely ignores, even though the question was on "taxes", not "federal taxes".
It is very hard to know what a "typical" American pays in state and local taxes. According to the Census Bureau. in 2008 state and local individual income and sales taxes were $636 billion dollars. Since these are mostly flat, and since pre-tax labor income in 2008 was about 10 trillion dollars, these taxes amount to another 6% on someone making 50.000.
The total taxes are (3500+3830+3830+3000)/(50000+3830)=26.3%
This is very close to the estimate made by Tea Party protesters, Bruce Bartlett is wrong is saying that the Tea Partiers don't know how much taxes they pay. It is he, the intellectual, who underestimates taxes and underestimates the Tea Partiers.
The public is generally not very good at estimating statistics. For example according to Slemrod and Bakijas 2004 textbook while 2% of millionaires do not pay any taxes, the American public had estimated this at 45%. On issues close to them however, the Tea Partiers have a surprisingly good gauge.
That is presumably why they know approximately what the tax burden on someone earning 50.000 and Bartlett doesn't: Many of them actually earn around this, whereas it is just an intellectual game for Bartlett.
It would be very interesting to see the same poll on the general public.
2. FromForum and Bruce Bartlett are also petty in pretending that the Tea Party people are way off on two other questions, taxes as a share of GDP and whether Obama wants to raise taxes.
An extremely important fact about interpreting polls, an insight made by Paul Grice, is that people do not answer what you literary said, they answer what they think you are asking, their best guess of what your question really implies.
So when you ask a Tea Protester if Obama raised taxes, he will not interpret you literary (so far not), but think about the most salient truth: Obama has promised to repeal large parts of the Bush tax cuts, raise indirect taxes on carbon emissions, and is currently in the process of raising taxes through his health care bill.
3. Also the question about "federal taxes as a share of GDP" is very dishonest. Most economist I know would think of total revenue as a share of GDP, not just the federal part, unless you remind them explicitly that you ONLY (why?) want part of the state.
What do we expect of the public, who may only have a vague idea of what even GDP means, let alone taxes as a share of GDP? They will probably just tell you how large they think the government it.
Also, are we to expect the public to take into account that tax revenue as a share of GDP collapses during recessions, even when tax rates stay constant? The prediction from economic science is that the correct burden of government is the tax *rates*, not the tax *revenue*.
Lastly, should they include non-tax government revenue or not? Are we for example suppose to pretend that a future cap-and trade and similar current sources of revenue are not taxes?
The last year before the crisis U.S tax revenue as a share of GDP was 28.3%. As a share of national income, it was 32% (basically you remove destroyed capital from GDP to get national income). Government spending in 2010 is 44.5% of GDP and about 50% of national income.
Milton Friedman usually advocated including deficits as a tax (on the future), and using national income, rather than GDP (excluding depreciations). (To his credit Bartlet mentions the point about the future, although in a snide way). That is why Milton Friedman in 2003 said that government was 40% of the economy.
Who do you think had thought more deeply about the burden of government, Milton Friedman or Bruce Bartlett?
Was Milton Friedman also a know-nothing crazed populist who merely imagined that government was too large, because he was fooled by Fox news?
This is why people reject intellectuals like Bruce Bartlett: the quality of the thinking of the current bunch of intellectuals America has is sometimes simply too low relative to their pretensions (and incidentally relative to the instincts of the public).
I am convinced it is not intellectualism itself the market has rejected, but the actual lazy and left leaning intellectuals America has to offer.
Grice, P., 1975. Logic and conversation. In: Cole, P., Morgan, J. (Eds.), Syntax and Semantics: Speech Acts, vol. 3. Academic Press, New York, pp. 41–58.
Update:
See comments. I missed that in the appendix it is specified that it is the federal government that they refer to. Excluding state and local taxes someone earning 50.000 pays about 21%, if payroll taxes are included.
For Sweden, more than half of all taxes are hidden (payroll taxes, VAT), and a study I conducted with Björn Wallace showed on a sample of about 1000 people that Swedes underestimate the tax burden on a typical worker by about 1 third (they believed 40% whereas at the time it was about 63%).
FrumForum conducted a survey of 57 tea party protesters, and asked them about the tax burden.
I would have guessed they would strongly over-estimate taxes, given their ideological bias. In fact the tea party people have a pretty good handle on taxes.
1. They were asked:
"What Does the Typical Family Earning $50KPay In Taxes?"
The mean answer was $12.710., or 25.4%.
Fromfurom and former conservative Bruce Bartlett wrongly claim that this figure is incorrect. In fact it is very close to the likely truth.
Fromforum absurdly claims that a typical family earning $50.000 only pays 15% in taxes, because they are in the 15% federal tax bracket. This claim is wrong on several levels. First of all, the questions asks about how much you pay, the average tax ("how much do you typically pay"), not the marginal tax ("how much of your income increase would you pay in taxes"). Second, they ask generally about all taxes, not just the federal income tax.
We are at a point where the biggest tax burden on low and middle income American's are payroll taxes, no longer the federal income tax. This is particularly if you include, and of course you must include, the half of the payroll tax that is nominally "paid" by the firm, but whose burden is on the worker. Lastly Fromforum may be interested to learn that there is such a thing called state and local taxes.
According to the IRS, the average amount of taxes ultimately paid by households earning 40-50.000 was 6.7% of adjusted gross income, and for households earning 50-75.000 8.1%. Let us say 7%. (Bartlett claims it is 1.7%, which I find too low).
In addition to this, someone earning 50.000 will directly pay 7.65% in Social Security and Medicare taxes, and indirectly another 7.65% in Social Security and Medicare taxes "paid" by the employer. It is unclear if Bartlett takes into account the payroll taxes nominally on the employer whose incidence is on the worker. He should.
Lastly we have state and local taxes, which Bartlett completely ignores, even though the question was on "taxes", not "federal taxes".
It is very hard to know what a "typical" American pays in state and local taxes. According to the Census Bureau. in 2008 state and local individual income and sales taxes were $636 billion dollars. Since these are mostly flat, and since pre-tax labor income in 2008 was about 10 trillion dollars, these taxes amount to another 6% on someone making 50.000.
The total taxes are (3500+3830+3830+3000)/(50000+3830)=26.3%
This is very close to the estimate made by Tea Party protesters, Bruce Bartlett is wrong is saying that the Tea Partiers don't know how much taxes they pay. It is he, the intellectual, who underestimates taxes and underestimates the Tea Partiers.
The public is generally not very good at estimating statistics. For example according to Slemrod and Bakijas 2004 textbook while 2% of millionaires do not pay any taxes, the American public had estimated this at 45%. On issues close to them however, the Tea Partiers have a surprisingly good gauge.
That is presumably why they know approximately what the tax burden on someone earning 50.000 and Bartlett doesn't: Many of them actually earn around this, whereas it is just an intellectual game for Bartlett.
It would be very interesting to see the same poll on the general public.
2. FromForum and Bruce Bartlett are also petty in pretending that the Tea Party people are way off on two other questions, taxes as a share of GDP and whether Obama wants to raise taxes.
An extremely important fact about interpreting polls, an insight made by Paul Grice, is that people do not answer what you literary said, they answer what they think you are asking, their best guess of what your question really implies.
So when you ask a Tea Protester if Obama raised taxes, he will not interpret you literary (so far not), but think about the most salient truth: Obama has promised to repeal large parts of the Bush tax cuts, raise indirect taxes on carbon emissions, and is currently in the process of raising taxes through his health care bill.
3. Also the question about "federal taxes as a share of GDP" is very dishonest. Most economist I know would think of total revenue as a share of GDP, not just the federal part, unless you remind them explicitly that you ONLY (why?) want part of the state.
What do we expect of the public, who may only have a vague idea of what even GDP means, let alone taxes as a share of GDP? They will probably just tell you how large they think the government it.
Also, are we to expect the public to take into account that tax revenue as a share of GDP collapses during recessions, even when tax rates stay constant? The prediction from economic science is that the correct burden of government is the tax *rates*, not the tax *revenue*.
Lastly, should they include non-tax government revenue or not? Are we for example suppose to pretend that a future cap-and trade and similar current sources of revenue are not taxes?
The last year before the crisis U.S tax revenue as a share of GDP was 28.3%. As a share of national income, it was 32% (basically you remove destroyed capital from GDP to get national income). Government spending in 2010 is 44.5% of GDP and about 50% of national income.
Milton Friedman usually advocated including deficits as a tax (on the future), and using national income, rather than GDP (excluding depreciations). (To his credit Bartlet mentions the point about the future, although in a snide way). That is why Milton Friedman in 2003 said that government was 40% of the economy.
Who do you think had thought more deeply about the burden of government, Milton Friedman or Bruce Bartlett?
Was Milton Friedman also a know-nothing crazed populist who merely imagined that government was too large, because he was fooled by Fox news?
This is why people reject intellectuals like Bruce Bartlett: the quality of the thinking of the current bunch of intellectuals America has is sometimes simply too low relative to their pretensions (and incidentally relative to the instincts of the public).
I am convinced it is not intellectualism itself the market has rejected, but the actual lazy and left leaning intellectuals America has to offer.
Grice, P., 1975. Logic and conversation. In: Cole, P., Morgan, J. (Eds.), Syntax and Semantics: Speech Acts, vol. 3. Academic Press, New York, pp. 41–58.
Update:
See comments. I missed that in the appendix it is specified that it is the federal government that they refer to. Excluding state and local taxes someone earning 50.000 pays about 21%, if payroll taxes are included.
Wednesday, March 17, 2010
Hayek defeats Marx once again.
Justin Wolfers, a clever up and coming economist, thinks that Friedrich von Hayek was not an important enough economist to be included in the company of Adam Smith, Karl Marx, John Maynard Keynes and Milton Friedman.
He basis this on a search on JSTOR, that shows Hayek to have the same influence as Larry Summers.
But Wolfers methodology is faulty. First of all, it seems to be very sensitive to wording. Second, it does not tell us the influence of the most important idea.
I instead use Google Scholar to look at the number of citations for the 20 most cited works of each economist. Since I can see exactly who wrote what, this does not have the problem associated with Wolfer's method.
The result confirms what (at least my) common sense tells me, Hayek is very influential.
First, number of citations of most cited work (regardless of spelling):
1. Adam Smith (15616).
2. John Maynard Keynes (11445)
3. Friedrich von Hayek (5397)
4. Milton Friedman (3528)
5. Karl Marx (3210)
6. Larry Summers (2082)
Which confirms another point, Hayek's theory on dispersed knowledge is more important than any *single* article Milton Friedman wrote, even though Friedman was a broad guy who made contributions in several parts of economics. Marx does not do as well here, because his followers made their strongest impression conducting revolutions (and more recently analyzing language as a tool of oppression) rather than doing mainstream economics.
Second, number of citations of 20 most cited work:
1. Milton Friedman (35867)
2. Adam Smith (22997)
3. Friedrich von Hayek (22668)
4. John Maynard Keynes (21679)
5. Karl Marx (19695)
6. Larry Summers (13039)
(for those curious, and as a measure of how much a "normal" top economist gets, Wolfers himself has 3170 citations).
By Wolfer's own criteria, a quantitative measure of scholarly influence, Hayek beats or ties with Keynes and with Milton Friedman, and beats Marx in both measures. He is far above Larry Summers (an unfair comparison, since prolific Summers is contemporary, which boosts you in this type of count.).
What Wolfers also fails to take into account is the diversity of the idea. Hayek is extremely original, and his insights about decentralized knowledge and spontaneous order are quite different from mainstream neoclassical arguments for the market.
This is a *plus* for putting him in the textbooks. You don't want a dozen more neoclassical intellectuals who make the exact same argument Adam Smith and Milton Friedman made. Hayek had a unique mind and offered unique insights.
He basis this on a search on JSTOR, that shows Hayek to have the same influence as Larry Summers.
But Wolfers methodology is faulty. First of all, it seems to be very sensitive to wording. Second, it does not tell us the influence of the most important idea.
I instead use Google Scholar to look at the number of citations for the 20 most cited works of each economist. Since I can see exactly who wrote what, this does not have the problem associated with Wolfer's method.
The result confirms what (at least my) common sense tells me, Hayek is very influential.
First, number of citations of most cited work (regardless of spelling):
1. Adam Smith (15616).
2. John Maynard Keynes (11445)
3. Friedrich von Hayek (5397)
4. Milton Friedman (3528)
5. Karl Marx (3210)
6. Larry Summers (2082)
Which confirms another point, Hayek's theory on dispersed knowledge is more important than any *single* article Milton Friedman wrote, even though Friedman was a broad guy who made contributions in several parts of economics. Marx does not do as well here, because his followers made their strongest impression conducting revolutions (and more recently analyzing language as a tool of oppression) rather than doing mainstream economics.
Second, number of citations of 20 most cited work:
1. Milton Friedman (35867)
2. Adam Smith (22997)
3. Friedrich von Hayek (22668)
4. John Maynard Keynes (21679)
5. Karl Marx (19695)
6. Larry Summers (13039)
(for those curious, and as a measure of how much a "normal" top economist gets, Wolfers himself has 3170 citations).
By Wolfer's own criteria, a quantitative measure of scholarly influence, Hayek beats or ties with Keynes and with Milton Friedman, and beats Marx in both measures. He is far above Larry Summers (an unfair comparison, since prolific Summers is contemporary, which boosts you in this type of count.).
What Wolfers also fails to take into account is the diversity of the idea. Hayek is extremely original, and his insights about decentralized knowledge and spontaneous order are quite different from mainstream neoclassical arguments for the market.
This is a *plus* for putting him in the textbooks. You don't want a dozen more neoclassical intellectuals who make the exact same argument Adam Smith and Milton Friedman made. Hayek had a unique mind and offered unique insights.
Tuesday, March 16, 2010
Japan - Defying Gravity
By Claus Vistesen: CopenhagenPopular myth and, allegedly, the laws of aerodynamics have it that the bumblebee should not be able to take flight. Yet still, our good bumblebee refuses to be pulled down by such details and year after year it takes flight as if nothing has happened. This allegory applies, with some imagination, to Japans economy too. Year after year it consequently appears able to
Monday, March 15, 2010
A defence of the Spirit Level that is not really a defence
Sociology-professor Denny Vårgerö today writes (in Swedish) an article about inequality and mortality. He has a graph that links income with mortality, showing that lower income people are more likely to die.
1. First a small point: Vårgerö writes that our critique of the Spirit Level is that the choice of countries and the measure of inequality influences the results.
Our critique is much more powerful than that. In ALL definitions of inequality and REGARDLESS if we looked at 21, 23 or 28 countries, there was no statistically significant relationship at conventional levels between inequality and mortality.
The writes or the Spirit Level are deliberately confusing their readers in PRETENDING that we are just nitpicking about inequality measures.
Professor Vårgerö: Don't take their word for it, they are fooling you.
2. A larger point:
This graph does not in itself prove that lower inequality would save lives, because of reverse causality. If you are of poor health, you will earn less, and also be more likely to die. This is extremely important for interpreting his graph.
Vårgerö neglects this problem, which I find astonishing. The left and left leaning social scientists in Sweden really needs to learn about causality.
Furthermore, there are missing variables that co-determine life expectancy and income. A particularly important one is education: more educated people take care of themselves better, and earn more. Vårgerö mentions this second point at least.
3. The most important point:
The claim in the Spirit Level is not that low income kills you as Vårgerö seems to think. It is that the high income for the rich *cause* others people in society to die from stress. That is why they do all those cross country comparisons, they believe the income of OTHER PEOPLE is what is the core of the issue, not your own income (as Vårgerö uses).
It is of course far from common sense observations that educated people take better care of themselves and thus live longer. (and people with more self-discipline are more likely to get educated in the first place).
It is also different than Vårgerös point that lower income causes you to die.
If the Spirit Level was just claiming that low income is bad, it would not have gotten so much attention. Everyone already know that (although the causal effect on mortality in countries with Universal Health care have not been understood).
Vårgerö does not seem to realize that they make a much stronger claim: that the income of other people determined my health, and that high income is bad. That is why they are obsessed with inequality measures, rather than income measures.
Does Vårgerö really want to defend the astonishing claim in the Spirit Level? If so, does he have any casual evidence for it? He only has my own income and my chance of dying in the graph, which leads me to believe he has not studied the argument in the Spirit Level carefully.
I already posted this paper by Magnus Johannesson and Ulf-G. Gerdtham about Sweden, but let me post it again:
"We test whether mortality is related to individual income, mean community income, and community income inequality, controlling for initial health status and personal characteristics. The analysis is based on a random sample from the adult Swedish population of more than 40,000 individuals who were followed up for 10–17 years. We find that mortality decreases significantly as individual income increases. For mean community income and community income inequality we cannot, however, reject the null hypothesis of no effect on mortality."
Lower income is associated with mortality (the causality problem is of course still there), but income INEQUALITY is not. According to Johannesson and Gerdtham Vårgerö is right and the Spirit Level is wrong. But Vårgerö does not understand in his SVD article that he and the Spirit Level are making different, in some terms opposite claims.
He is defending people who actually disagree with him. I suppose one reason is that The Spirit Level argument about inequality killing you through stress is hard to defend scientifically.
A simpler reasons seems to be that Vårgerö thinks The Spirit Level just sais inequality=bad, and we are saying inequality=good, so he wants to defend them.
But I don't think inequality is good, and I strongly believe in raising the income of the poor in Sweden, and even stronger in raising their health status. What I oppose to is the bad science in the Spirit Level, and the claim that lowering the level of the educated will magically help the uneducated health status.
4. Look for a moment at this graph from the Swedish National Institute of Public Health. It shows unhealthy habits and education and income. The lower your education and the lower your income, the more likely you are to have unhealthy habits.
5% of Swedish men with high education smoke, compared with 27% of the unemployed and 32% of those on long sick-leaves.
The first question is, why do we think these people are more likely to die? Because they have low income, or because they have bad health habits?
The second question is: do we think they have bad habits because the rich have high incomes? If the poor stayed poor and uneducated, but I raised the marginal tax on the rich, would mortality of the poor go down?
The answer matters enormously for policy. If you believe the junk science of the Spirit Level, the problem in society are the rich and educated. If we target them, and make them poorer, social problems of the poor will automatically solve, because they were caused by envy and relative status.
The Swedish health policy instead sensibly focuses on the behavior and living standard of the poor, and tried to solve them through education, information, taxes on alcohol and tobacco. If the Spirit Level is correct, we have been silly to do this. The problem are not the poor, the problem is the rich! If you see a kid in school smoking, don't focus on him. Find the kid from a good home not smoking and getting straight A:s, and punish him. This will reduce the stress of the kid smoking, and solve his health problems.
Does this sound insane to you? It should. And if you want to make insane claims like this, at least have some empirical evidence, which demonstrates causation.
1. First a small point: Vårgerö writes that our critique of the Spirit Level is that the choice of countries and the measure of inequality influences the results.
Our critique is much more powerful than that. In ALL definitions of inequality and REGARDLESS if we looked at 21, 23 or 28 countries, there was no statistically significant relationship at conventional levels between inequality and mortality.
The writes or the Spirit Level are deliberately confusing their readers in PRETENDING that we are just nitpicking about inequality measures.
Professor Vårgerö: Don't take their word for it, they are fooling you.
2. A larger point:
This graph does not in itself prove that lower inequality would save lives, because of reverse causality. If you are of poor health, you will earn less, and also be more likely to die. This is extremely important for interpreting his graph.
Vårgerö neglects this problem, which I find astonishing. The left and left leaning social scientists in Sweden really needs to learn about causality.
Furthermore, there are missing variables that co-determine life expectancy and income. A particularly important one is education: more educated people take care of themselves better, and earn more. Vårgerö mentions this second point at least.
3. The most important point:
The claim in the Spirit Level is not that low income kills you as Vårgerö seems to think. It is that the high income for the rich *cause* others people in society to die from stress. That is why they do all those cross country comparisons, they believe the income of OTHER PEOPLE is what is the core of the issue, not your own income (as Vårgerö uses).
It is of course far from common sense observations that educated people take better care of themselves and thus live longer. (and people with more self-discipline are more likely to get educated in the first place).
It is also different than Vårgerös point that lower income causes you to die.
If the Spirit Level was just claiming that low income is bad, it would not have gotten so much attention. Everyone already know that (although the causal effect on mortality in countries with Universal Health care have not been understood).
Vårgerö does not seem to realize that they make a much stronger claim: that the income of other people determined my health, and that high income is bad. That is why they are obsessed with inequality measures, rather than income measures.
Does Vårgerö really want to defend the astonishing claim in the Spirit Level? If so, does he have any casual evidence for it? He only has my own income and my chance of dying in the graph, which leads me to believe he has not studied the argument in the Spirit Level carefully.
I already posted this paper by Magnus Johannesson and Ulf-G. Gerdtham about Sweden, but let me post it again:
"We test whether mortality is related to individual income, mean community income, and community income inequality, controlling for initial health status and personal characteristics. The analysis is based on a random sample from the adult Swedish population of more than 40,000 individuals who were followed up for 10–17 years. We find that mortality decreases significantly as individual income increases. For mean community income and community income inequality we cannot, however, reject the null hypothesis of no effect on mortality."
Lower income is associated with mortality (the causality problem is of course still there), but income INEQUALITY is not. According to Johannesson and Gerdtham Vårgerö is right and the Spirit Level is wrong. But Vårgerö does not understand in his SVD article that he and the Spirit Level are making different, in some terms opposite claims.
He is defending people who actually disagree with him. I suppose one reason is that The Spirit Level argument about inequality killing you through stress is hard to defend scientifically.
A simpler reasons seems to be that Vårgerö thinks The Spirit Level just sais inequality=bad, and we are saying inequality=good, so he wants to defend them.
But I don't think inequality is good, and I strongly believe in raising the income of the poor in Sweden, and even stronger in raising their health status. What I oppose to is the bad science in the Spirit Level, and the claim that lowering the level of the educated will magically help the uneducated health status.
4. Look for a moment at this graph from the Swedish National Institute of Public Health. It shows unhealthy habits and education and income. The lower your education and the lower your income, the more likely you are to have unhealthy habits.
5% of Swedish men with high education smoke, compared with 27% of the unemployed and 32% of those on long sick-leaves.
The first question is, why do we think these people are more likely to die? Because they have low income, or because they have bad health habits?
The second question is: do we think they have bad habits because the rich have high incomes? If the poor stayed poor and uneducated, but I raised the marginal tax on the rich, would mortality of the poor go down?
The answer matters enormously for policy. If you believe the junk science of the Spirit Level, the problem in society are the rich and educated. If we target them, and make them poorer, social problems of the poor will automatically solve, because they were caused by envy and relative status.
The Swedish health policy instead sensibly focuses on the behavior and living standard of the poor, and tried to solve them through education, information, taxes on alcohol and tobacco. If the Spirit Level is correct, we have been silly to do this. The problem are not the poor, the problem is the rich! If you see a kid in school smoking, don't focus on him. Find the kid from a good home not smoking and getting straight A:s, and punish him. This will reduce the stress of the kid smoking, and solve his health problems.
Does this sound insane to you? It should. And if you want to make insane claims like this, at least have some empirical evidence, which demonstrates causation.
Subscribe to:
Posts (Atom)