perspective on the healthcare bill

I just caught a part of Laura Ingraham’s interview with Dick Morris, and he had an interesting perspective on the healthcare bill in the house. He says that the Democrats are engaged in

“healthcare redistribution … It takes $500 billion in healthcare from the elderly in medicare and medicaid (and medicaid funds nursing homes), and takes it and gives it to people who are younger, healthier and richer, but at the moment lack health insurance. And that redistribution is something they cannot and will not support… I point out the catastrophe, what a horrible thing this would be to the elderly, what this means in elective surgery being postponed, 8 weeks for cancer radiation that you have in Canada, a 25% higher incidence of colon cancer in Canada because there is an 8 month wait for colonoscopies, and these are all inevitable consequences of expanding the demand by covering 50 million new people, without expanding the supply by creating additional doctors”

Finally, someone who understands a bit of economics. Unfortunately, it is apparent why the Democrats won’t explain it clearly like this. But the people seem to be getting the message anyway. I hope this means there is some innate economic sense … hmmm, we probably haven’t been living in mass society long enough for that to evolve. But how else to explain the strange convergence between tea parties and common sense? Could history be repeating itself?

Imagine, if instead of more government involvement, we simply had the government stop restricting the supply of healthcare. Imagine if instead of spending years teaching nurses that “YOU ARE NOT A DOCTOR”, we spent even less time, teaching them that they were a doctor instead. Voila, we’d have a $25,000 dollar GP instead of a $150,000 GP. The Congressional bill increases the demand for healthcare, perhaps we would do more good increasing the supply. Uh oh, “supply side economics” rears its rational head again.

– Martin L

I just caught a part of Laura Ingraham’s interview with Dick Morris, and he had an interesting perspective on the healthcare bill in the house. He says that the Democrats are engaged in

“healthcare redistribution … It takes $500 billion in healthcare from the elderly in medicare and medicaid (and medicaid funds nursing homes), and takes it and gives it to people who are younger, healthier and richer, but at the moment lack health insurance. And that redistribution is something they cannot and will not support… I point out the catastrophe, what a horrible thing this would be to the elderly, what this means in elective surgery being postponed, 8 weeks for cancer radiation that you have in Canada, a 25% higher incidence of colon cancer in Canada because there is an 8 month wait for colonoscopies, and these are all inevitable consequences of expanding the demand by covering 50 million new people, without expanding the supply by creating additional doctors”

Finally, someone who understands a bit of economics. Unfortunately, it is apparent why the Democrats won’t explain it clearly like this. But the people seem to be getting the message anyway. I hope this means there is some innate economic sense … hmmm, we probably haven’t been living in mass society long enough for that to evolve. But how else to explain the strange convergence between tea parties and common sense? Could history be repeating itself?

Imagine, if instead of more government involvement, we simply had the government stop restricting the supply of healthcare. Imagine if instead of spending years teaching nurses that “YOU ARE NOT A DOCTOR”, we spent even less time, teaching them that they were a doctor instead. Voila, we’d have a $25,000 dollar GP instead of a $150,000 GP. The Congressional bill increases the demand for healthcare, perhaps we would do more good increasing the supply. Uh oh, “supply side economics” rears its rational head again.

[From Rick Marken (2009.09.03.0900)]

···

On Thu, Sep 3, 2009 at 2:45 AM, Martin Lewitt mlewitt@comcast.net wrote:

I just caught a part of Laura Ingraham’s interview with Dick Morris, and he had an interesting perspective on the healthcare bill in the house. He says that the Democrats are engaged in

"healthcare redistribution

Finally, someone who understands a bit of economics.

One of the consistent themes I see in my discussions with conservatives is their deep conviction that they understand economics while liberals like myself don’t. This, despite the fact that, by virtually every measure of economic performance, the data show that conservative economic policies have produced worse results than liberal ones. In my “Got data” post I show that increasing the top marginal tax rate (which redistributes income) is associated with greater growth and lower unemployment. I have shown that increases in investment follow increases in growth (rather than vice versa, as predicted by economic theory). The only place where conservative economic policies do better than liberal ones is in increasing wealth discrepancy.

I think conservatives have managed to con people into thinking that they are the “grown ups” when it comes to economics. In fact, I think it’s clear that they are the Elmer Gantrys of economics, playing on people’s (possibly as well as their own) fears and predjudices to enrich themselves.

I don’t think anyone, liberal or conservative, understands economics particularly well. And they won’t start understanding it until they start building plausible models that accurately predict the data. But until then I think we should evaluate economic policies based on data not on beliefs, no matter how firmly held the latter. And the data show clearly that conservative economic policies have produced far worse results (slower growth, higher unemployment, stagnant wages, etc) than liberal ones (unless one’s goal is enriching the rich and impoverishing the middle class).

Best

Rick

PS. I’m glad to see, Martin L, that you are “pro-choice”; what could be a more horrendous governmental intrusion on one’s autonomy than to prevent you from doing what you feel is necessary with your own body. But I thought I detected the typical conservative hedging about this when you said that you thought it should be left to the states. That doesn’t sound very pro-choice to me. What about the states that would outlaw abortion? This is like saying you are anti-slavery but that it should be left up to the states. Would you be comfortable about leaving gun rights up to the states?


Richard S. Marken PhD
rsmarken@gmail.com
www.mindreadings.com

Rick,

How can you characterize and reach conclusions about conservative economic policy, when we haven’t had it for over a century? Reagan’s tenure was dominated by the Volcker induced recession that started him off deep in the economic and budgetary hole and when Reagan proposed eliminating the double tax on dividends, he was greeted by a chorus of class warfare rhetoric from the Democrats and the media. His budget chief also worked internally to oppose implementation of supply side economics. GW Bush stuck to his guns better on elimination of the double tax on dividends, but even he only achieved a reduction in it by half, so equity financing was still taxed 50% greater than debt financing. Bush inherited the dot.com bubble from Clinton. Although Bush tried to reign in the government sponsored enterprises, FANNIE and FREDDIE, he wasn’t willing to expend much of his political capital on it in the face of corrupt Democratic opposition. Neither president put the money supply on a sound basis, nor did a good job of balancing the budget. Neither president achieved the type of major reform of the tax code, that would tax consumption and not savings.

My conclusion that abortion should be left to the states, is based upon application of constitutional law, that even some liberal jurists admit should have been the result of Roe v. Wade. Whether one is pro-choice or not shouldn’t influence his assessment of correct application of the law. Of course, I would hope that every state would decide that abortion should be legal, passing statutes or amending their constitutions as necessary.

The right to keep and bear arms on the other hand, is in the constitution, and it is clear that the main authors considered an armed populace as the ultimate check on the government, should the other checks and balances fail.

regards,

Martin

···

----- Original Message -----
From: “Richard Marken” rsmarken@GMAIL.COM
To: CSGNET@LISTSERV.ILLINOIS.EDU
Sent: Thursday, September 3, 2009 10:01:46 AM GMT -07:00 US/Canada Mountain
Subject: Re: perspective on the healthcare bill

[From Rick Marken (2009.09.03.0900)]

On Thu, Sep 3, 2009 at 2:45 AM, Martin Lewitt mlewitt@comcast.net wrote:

I just caught a part of Laura Ingraham’s interview with Dick Morris, and he had an interesting perspective on the healthcare bill in the house. He says that the Democrats are engaged in

"healthcare redistribution

Finally, someone who understands a bit of economics.

One of the consistent themes I see in my discussions with conservatives is their deep conviction that they understand economics while liberals like myself don’t. This, despite the fact that, by virtually every measure of economic performance, the data show that conservative economic policies have produced worse results than liberal ones. In my “Got data” post I show that increasing the top marginal tax rate (which redistributes income) is associated with greater growth and lower unemployment. I have shown that increases in investment follow increases in growth (rather than vice versa, as predicted by economic theory). The only place where conservative economic policies do better than liberal ones is in increasing wealth discrepancy.

I think conservatives have managed to con people into thinking that they are the “grown ups” when it comes to economics. In fact, I think it’s clear that they are the Elmer Gantrys of economics, playing on people’s (possibly as well as their own) fears and predjudices to enrich themselves.

I don’t think anyone, liberal or conservative, understands economics particularly well. And they won’t start understanding it until they start building plausible models that accurately predict the data. But until then I think we should evaluate economic policies based on data not on beliefs, no matter how firmly held the latter. And the data show clearly that conservative economic policies have produced far worse results (slower growth, higher unemployment, stagnant wages, etc) than liberal ones (unless one’s goal is enriching the rich and impoverishing the middle class).

Best

Rick

PS. I’m glad to see, Martin L, that you are “pro-choice”; what could be a more horrendous governmental intrusion on one’s autonomy than to prevent you from doing what you feel is necessary with your own body. But I thought I detected the typical conservative hedging about this when you said that you thought it should be left to the states. That doesn’t sound very pro-choice to me. What about the states that would outlaw abortion? This is like saying you are anti-slavery but that it should be left up to the states. Would you be comfortable about leaving gun rights up to the states?


Richard S. Marken PhD
rsmarken@gmail.com
www.mindreadings.com

[From Rick Marken (2009.09.03.1030)]

···

On Thu, Sep 3, 2009 at 9:54 AM, Martin Lewitt mlewitt@comcast.net wrote:

Rick,

How can you characterize and reach conclusions about conservative economic policy, when we haven’t had it for over a century?

If that’s true, then I can’t. But I’m just going by what people say. Republicans characterize themselves as conservative so I take Republication policies as tending toward the conservative. And Democrats are always tagged as liberal (although they really haven’t been all that liberal since FDR). So I assume that, to the extent that Democratic policies are implemented they are more liberal policies. Blaming Reagan’s awful results on Volker is the reason why I look at data over many years. Fed policies are one of those confounding variables I mentioned and I presume that, to some exten the effects of those variables are averaged out when looking at the same or similar policies at several different time periods.

But if conservative policies have’t really not been implemented in over 100 years, then how do you know they work better than liberal policies?

Best

Rick


Richard S. Marken PhD
rsmarken@gmail.com
www.mindreadings.com

[Martin Taylor 2009.09.03.13.50]

[From Rick Marken (2009.09.03.1030)]

Rick,

How can you characterize and reach conclusions about
conservative economic policy, when we haven’t had it for over a
century?

If that’s true, then I can’t. But I’m just going by what people
say. Republicans characterize themselves as conservative so I take
Republication policies as tending toward the conservative. And
Democrats are always tagged as liberal (although they really haven’t
been all that liberal since FDR). So I assume that, to the extent that
Democratic policies are implemented they are more liberal policies.
Blaming Reagan’s awful results on Volker is the reason why I look at
data over many years. Fed policies are one of those confounding
variables I mentioned and I presume that, to some exten the effects of
those variables are averaged out when looking at the same or similar
policies at several different time periods.

But if conservative policies have’t really not been implemented in over
100 years, then how do you know they work better than liberal
policies?

Don’t you smell a troll, Rick? I find it hard to believe that anybody
who thinks seriously about these matters could hold the opinions he
claims to hold. The other clue to his probable trollhood is that he
resolutely abstains from using the CSGnet convention of
self-identifying and time-stamping his messages. If he really intended
to be considered a serious CSGnet discussant, he would at least try to
assist in back-referencing, once he had been informed of the convention.

Maybe I’m wrong, and he really is an economic flat-earther, but he
seems otherwise intelligent, so … But even if he is, why would he be
so adamant in refusing the CSGnet posting conventions?

Martin

···

On Thu, Sep 3, 2009 at 9:54 AM, Martin > Lewitt mlewitt@comcast.net > wrote:

From: “Richard Marken” rsmarken@GMAIL.COM
Sent: Thursday, September 3, 2009 11:29:08 AM GMT -07:00 US/Canada Mountain

*** snipped ***

But if conservative policies have’t really not been implemented in over 100 years, then how do you know they work better than liberal policies?

Because they reverse the tax penalty on investment and capital formation, lowering the cost of capital and changing the mix of capital towards equity which allows more flexibility than debt capital, puts the economy on a less speculative footing by reducing leverage. Current policy results in uneconomic retention of earnings because even though the returns will be lower when invested where they are they are taxed only once, whereas moving the capital to where the returns will be higher incurrs a tax penalty, that the higher returns seldom can overcome. The overall efficiency of the economy is lower when capital can’t flow to its most economic use.

regards,

Martin L

mlewitt@comcast.net

···

----- Original Message -----

Best

Rick

[From Rick Marken (2009.09.03.1105)]

[Martin Taylor 2009.09.03.13.50]

Don’t you smell a troll, Rick?

I sure smell something;-)

I gotta get to work – on moving to Canada. This place is nuts. I don’t think even Obama can save us. And with global warming, Canada’s looking more and more attractive;-)

Love

Rick

···


Richard S. Marken PhD
rsmarken@gmail.com
www.mindreadings.com

From: “Martin Taylor” mmt-csg@MMTAYLOR.NET
Sent: Thursday, September 3, 2009 11:57:34 AM GMT -07:00 US/Canada Mountain

Maybe I’m wrong, and he really is an economic flat-earther, but he seems otherwise intelligent, so … But even if he is, why would he be so adamant in refusing the CSGnet posting conventions?

I am supplying the requested information in the most cost effective way in terms of my time. Note, that I included the “Sent” line that has the time requested. My own email will presumably have a “Sent” line as well. Why are you insisting on a pedantic format that doesn’t add any information and is troublesome toboth remember and type? I don’t see any advantage or value.

Martin

Martin L

···

----- Original Message -----

[From Rick Marken (2009.09.03.1150)]

···

On Thu, Sep 3, 2009 at 11:01 AM, Martin Lewitt mlewitt@comcast.net wrote:

----- Original Message -----
From: “Richard Marken” rsmarken@GMAIL.COM

But if conservative policies have’t really not been implemented in over 100 years, then how do you know they work better than liberal policies?

Because they reverse the tax penalty on investment and capital formation, lowering the cost of capital and changing the mix of capital towards equity which allows more flexibility than debt capital…

This is all theory. I will no longer accept theory as a basis of policy when the data is inconsistent with the theory. The theory you have described here is belied by the fact that there is a strong positive correlation between capital gains tax rate and growth. There is also a strong negative correlation between private investment and growth when the private investment measures precede the growth measures.

I can can make up economic stories that are consistent with these facts. But I won’t believe even my stories until I have a working model that predicts these facts perfectly.

Unless you are willing to let your economic theories be tempered by data I really am not interested in hearing them.

Best

Rick


Richard S. Marken PhD
rsmarken@gmail.com
www.mindreadings.com

From: “Richard Marken” rsmarken@GMAIL.COM
Sent: Thursday, September 3, 2009 12:52:23 PM GMT -07:00 US/Canada Mountain

*** snip ***

I can can make up economic stories that are consistent with these facts. But I won’t believe even my stories until I have a working model that predicts these facts perfectly.

Commendable skepticism, but that leaves you unable to justify any policy changes. I’m not sure where it leaves you vis’a’vis continuation of existing policies.

Unless you are willing to let your economic theories be tempered by data I really am not interested in hearing them.

Fortunately, I am willing to let them be tempered by data. But I also consider theory when the data is unavailable. Given the dynamic nonlinear complexity of the economy, theories are usually tested by perturbations of idealized models. The “story” I put forward is within the mainstream of the Austrian and Chicago schools of economics.

I hope your commendable skepticism, and insistance on perfect models and presumably also perfect data, also extends to other fields, such as the anthropogenic global warming hypothesis.

Best

Rick

regards,

Martin L

mlewitt@comcast.net

···

----- Original Message -----

[From Rick Marken (2009.09.03.1730)]

From: “Richard Marken” rsmarken@GMAIL.CO

I can can make up economic stories that are consistent with these facts. But I won’t believe even my stories until I have a working model that predicts these facts perfectly.

Commendable skepticism, but that leaves you unable to justify any policy changes. I’m not sure where it leaves you vis’a’vis continuation of existing policies.

Not true. I would keep policies that seem to be working and change those that don’t; kind of an e. coli approach to policy, except that the changes need not be completely random. When change seems needed we can change back toward a policy that seemed to work in the past but is no longer in effect. If the result of the change (after some reasonable period of observation, say two years) is not obviously better economic performance (given some agreed on criteria for good economic performance, which is not necessarily an easy agreement to reach) then change again. I’m rather conservative that way; I believe in trying policies that have seemed to work in the past, based on the historical data. So, for example, I would have changed my policy from tax cuts for the rich to tax increases for the rich as soon as I saw that that policy was not working. The historical data shows that economic growth has been greatest and unemployment lowest when the top marginal tax rate was high. So that would be an obvious policy change I would make right now, for example. I think the economy will not really get back up on it’s feed until the top marginal tax rate is raised to 1950s (Eisenhower) levels.

Unless you are willing to let your economic theories be tempered by data I really am not interested in hearing them.

Fortunately, I am willing to let them be tempered by data. But I also consider theory when the data is unavailable. Given the dynamic nonlinear complexity of the economy, theories are usually tested by perturbations of idealized models. The “story” I put forward is within the mainstream of the Austrian and Chicago schools of economics.

Which is useless unless they tested their theories against data. If their theories predict, for example, that taxes are recessionary then we know that they are wrong.

I hope your commendable skepticism, and insistance on perfect models and presumably also perfect data, also extends to other fields, such as the anthropogenic global warming hypothesis.

Sure. Actually, I have no skin in that game. I think we have to start making a serious effort to start shifting to non-fossil energy sources simply because fossil resources are obviously limited. If making this shift turns out to be irrelevant to global warminyong, then we’ve still gained big time by having an energy source that may last as long as people do (assuming you think people will last beyond the Apocalyse, which will occur when the Jews return to Israel – which is why those fundamentalists are not getting me to move there;-)

Best

Rick

···

On Thu, Sep 3, 2009 at 12:30 PM, Martin Lewitt mlewitt@comcast.net wrote:


Richard S. Marken PhD
rsmarken@gmail.com
www.mindreadings.com

From: “Richard Marken” rsmarken@GMAIL.COM
Sent: Thursday, September 3, 2009 6:32:04 PM GMT -07:00 US/Canada Mountain

*** snip ***

The historical data shows that economic growth has been greatest and unemployment lowest when the top marginal tax rate was high. So that would be an obvious policy change I would make right now, for example. I think the economy will not really get back up on it’s feed until the top marginal tax rate is raised to 1950s (Eisenhower) levels.

I went back and looked at your original posting, because these things are often sensitive to which starting years are selected for analysis. I suspect your conclusions are largely sensitive to those post war and Eisenhower years, when the US was essentially the only functioning industrial economy in the world and experienced rapid growth. It would be interesting to see how robust your results are starting after those years. I also suspect that your analytic method is based upon the assumption that the years are independent variables. I think you will find that years in the time series are correlated, and that would mean that your statistical significance is much lower than you are assuming.

*** snip ***

Best

Rick

regards,

Martin L

mlewitt@comcast.net

···

----- Original Message -----

[From Rick Marken (2009.09.04.0910)]

···

On Fri, Sep 4, 2009 at 6:37 AM, Martin Lewitt mlewitt@comcast.net wrote:

From: “Richard Marken” rsmarken@GMAIL.COM

The historical data shows that economic growth has been greatest and unemployment lowest when the top marginal tax rate was high. So that would be an obvious policy change I would make right now, for example. I think the economy will not really get back up on it’s feed until the top marginal tax rate is raised to 1950s (Eisenhower) levels.

I went back and looked at your original posting, because these things are often sensitive to which starting years are selected for analysis. I suspect your conclusions are largely sensitive to those post war and Eisenhower years

Well, it certainly changes things. When you look at the tax-growth correlation for only the years 1960-2008 the correlation goes from .18 (which it was for the 1947-2008 range) to .45, which is statistically significant (two-tailed, .05) which means only that the correlation can be considered non-zero. If we do a lagged correlation for those years, with taxes leading growth by a year, the correlation goes even higher, .48.

The tax-unemployment correlation for the years 1960-2008, which is -.23 for the period 1947-2008, goes to -.02; the one year lagged correlation goes to .04. These correlations are, of course, not significant which means that we cannot reject the hypothesis that the true correlation between taxes and unemployment is 0.0.

I think it’s best to use as much of the time series data as we can in attempting to determine whether or not there is a relationship between variables. This is not only because the more df we have the more power we have for the statistical test; but it also is better quasi-experimental design; we have a better chance of canceling out the effect of “extraneous” variables (like variations in fed rates, trade policies, technological developments, etc) that surely have some confounding influence on the variables of interest. But the results of all my analyses suggest one general conclusion; increases in the top marginal tax rate either have no relationship at all to growth and unemployment (both findings inconsistent with current economic dogma) or, their relationship is the opposite of what is predicted by current economic theory (the significant positive relationship between taxes and growth and the nearly significant negative relationship between taxes and unemployment).

I think there is really only one reasonable thing for an economist to do in the face of these data: start all over again from scratch, just as psychologists should do on the basis of the data produced in studies of perceptual control. I’m not holding my breath, though;-)

Best

Rick


Richard S. Marken PhD
rsmarken@gmail.com
www.mindreadings.com

Still with the year to year correlation your effective number of samples would be perhaps one-fith to an order of magnitude smaller. I wonder what statistical significance the result actually has. – Martin

···

----- Original Message -----
From: “Richard Marken” rsmarken@GMAIL.COM
To: CSGNET@LISTSERV.ILLINOIS.EDU
Sent: Friday, September 4, 2009 10:09:18 AM GMT -07:00 US/Canada Mountain
Subject: Re: perspective on the healthcare bill

[From Rick Marken (2009.09.04.0910)]

On Fri, Sep 4, 2009 at 6:37 AM, Martin Lewitt mlewitt@comcast.net wrote:

From: “Richard Marken” rsmarken@GMAIL.COM

The historical data shows that economic growth has been greatest and unemployment lowest when the top marginal tax rate was high. So that would be an obvious policy change I would make right now, for example. I think the economy will not really get back up on it’s feed until the top marginal tax rate is raised to 1950s (Eisenhower) levels.

I went back and looked at your original posting, because these things are often sensitive to which starting years are selected for analysis. I suspect your conclusions are largely sensitive to those post war and Eisenhower years

Well, it certainly changes things. When you look at the tax-growth correlation for only the years 1960-2008 the correlation goes from .18 (which it was for the 1947-2008 range) to .45, which is statistically significant (two-tailed, .05) which means only that the correlation can be considered non-zero. If we do a lagged correlation for those years, with taxes leading growth by a year, the correlation goes even higher, .48.

The tax-unemployment correlation for the years 1960-2008, which is -.23 for the period 1947-2008, goes to -.02; the one year lagged correlation goes to .04. These correlations are, of course, not significant which means that we cannot reject the hypothesis that the true correlation between taxes and unemployment is 0.0.

I think it’s best to use as much of the time series data as we can in attempting to determine whether or not there is a relationship between variables. This is not only because the more df we have the more power we have for the statistical test; but it also is better quasi-experimental design; we have a better chance of canceling out the effect of “extraneous” variables (like variations in fed rates, trade policies, technological developments, etc) that surely have some confounding influence on the variables of interest. But the results of all my analyses suggest one general conclusion; increases in the top marginal tax rate either have no relationship at all to growth and unemployment (both findings inconsistent with current economic dogma) or, their relationship is the opposite of what is predicted by current economic theory (the significant positive relationship between taxes and growth and the nearly significant negative relationship between taxes and unemployment).

I think there is really only one reasonable thing for an economist to do in the face of these data: start all over again from scratch, just as psychologists should do on the basis of the data produced in studies of perceptual control. I’m not holding my breath, though;-)

Best

Rick


Richard S. Marken PhD
rsmarken@gmail.com
www.mindreadings.com

[From Rick Marken (2009.09.04.1400)]

···

On Fri, Sep 4, 2009 at 9:34 AM, Martin Lewitt mlewitt@comcast.net wrote:

Still with the year to year correlation your effective number of samples would be perhaps one-fith to an order of magnitude smaller. I wonder what statistical significance the result actually has. – Martin

I don’t understand what the “number of samples would be one-fifth to an order of magnitude smaller” refers to. The correlations between taxes and growth that I first reported are based on 80 data points: the tax and growth rate each year from 1929 to 2008. The correlations for tax and unemployment rate are based on 62 data points: the tax and unemployment rate for each year from 1947-2008. 62 is not 1/5th of 80. The correlations I just ran for the post Eisenhower years are based on 49 data points (1960-2008). And 49 if not 1/5 of 80. What were you referring to?
The significance of a correlation, be the way, depends on its size, the number of degrees of freedom (df) used to compute it (the number of data points - 2), the significance level desired (traditionally either .01 or .05, which is the probability that the decision to reject the null hypothesis – that the true correlation is 0.0 – is an error) and whether it is a one or two tailed decision. When I say that a correlation is or isn’t significant it is because I have done a conventional test of significance, taking all the factors I just mentioned into account.

Remember, in statistics the word “significant” does not mean that a result is significant, in the sense of being important. It means that there is only a small probability (traditionally .01 or .05, the significance level) that the decision to reject the null hypothesis is an error (a Type I error, to be specific). So when I say that a correlation is “significant” all I’m saying is that I reject the notion that the true correlation is 0.0 and that my decision has only a small chance(.01 or .05) of being a mistake. Similarly, when I say that a correlation is not “significant” all I’m saying is that I cannot reject the possibility that the true correlation is 0.0.

Only a couple of the correlations I reported are significant, and they are significant in the “wrong” direction (higher taxes are “significantly” associated with higher growth rate, for example). But even if all the correlations I reported were not significant, that should still lead economists to revise their theories of the economy if their theories lead them to conclude that there is any relationship between taxes and growth/unemployment. A non-significant correlation between tax rate and growth rate or between tax rate and unemployment rate means that there is no reason to believe that these variables are related, which should be very bad news for economists who believe that taxes are recessionary.

Best

Rick

Richard S. Marken PhD
rsmarken@gmail.com
www.mindreadings.com

[From Bill Powers (2009.09..04.1529 MDT)] --

Rick Marken (2009.09.04.1400) --

RM: The correlations for tax and unemployment rate are based on 62 data points: the tax and unemployment rate for each year from 1947-2008. 62 is not 1/5th of 80. The correlations I just ran for the post Eisenhower years are based on 49 data points (1960-2008). And 49 if not 1/5 of 80. What were you referring to?

I think Martin was saying that each year's data will be correlated with other years (because of seasonal effects?) which reduces the number of independent data points. But of course the yearly variations wouldn't be 100% correlated, so how do you figure the actual meaningful correlation? Is this one of those multiple regression situations? I'm just repeating words I've heard.

Best,

Bill P.

[From Rick Marken (2009.09.04.1510)]

Bill Powers (2009.09…04.1529 MDT) –

I think Martin was saying that each year’s data will be correlated with other years (because of seasonal effects?) which reduces the number of independent data points. But of course the yearly variations wouldn’t be 100% correlated, so how do you figure the actual meaningful correlation? Is this one of those multiple regression situations? I’m just repeating words I’ve heard.

That’s a good point. I checked it out and it seems that the test for the significance of a correlation requires no assumption about the independence of the data points on which the correlation is based. And, non-mathematician that I am, I just ran some simulations with a spreadsheet and it appears to me that sequential dependencies have no effect on the size or sign of an observed correlation. Which is nice since we use correlations all the time when analyzing the results of tracking experiments where there are very strong sequential dependencies between data points.

Best

Rick

···


Richard S. Marken PhD
rsmarken@gmail.com
www.mindreadings.com

[From Bill Powers (2009.09.04.1941 MDT)]

Rick Marken (2009.09.04.1510) --

That's a good point. I checked it out and it seems that the test for the significance of a correlation requires no assumption about the independence of the data points on which the correlation is based. And, non-mathematician that I am, I just ran some simulations with a spreadsheet and it appears to me that sequential dependencies have no effect on the size or sign of an observed correlation. Which is nice since we use correlations all the time when analyzing the results of tracking experiments where there are very strong sequential dependencies between data points.

I think that Martin L. was referring to the significance calculation. Suppose you have two data sets that you're correlating, and you simply duplicate them (place them end to end). If each data set has N elements, now the new data sets have 2N elements. So N is larger, and therefore sigma is smaller, or something like that. Wouldn't that artificially inflate the significance? Why am I telling you -- you're the expert on this.

Anyway, you've been talking about "strong" correlations of 0.5, which to me indicate that there's no relationship worth paying attention to. The chances of making a correct prediction of the sign of a change (not even the amount) are only 2 out of 3, which means an error rate of 1 in 3, which seems pretty bad to me. Of course I'm always thinking of the next prediction concerning an individual case, whereas prognosticators are more concerned with the track record involving all the cases. If you're right 2 times out of 3, that looks pretty good on your resume, but people involved in the individual case are looking at that 33% chance of a wrong decision which could affect them very seriously. I wouldn't want to make any important decisions for myself or someone else with that large a chance of getting it wrong.

Best,

Bill P.

[From Rick Marken (2009.09.04.2145)]

Bill Powers (2009.09.04.1941 MDT)–

I think that Martin L. was referring to the significance calculation. Suppose you have two data sets that you’re correlating, and you simply duplicate them (place them end to end). If each data set has N elements, now the new data sets have 2N elements. So N is larger, and therefore sigma is smaller, or something like that. Wouldn’t that artificially inflate the significance? Why am I telling you – you’re the expert on this.

Yes, the standard error of the sampling distribution of r goes down with increasing N but I don’t see how that would “artificially” inflate significance (you can’t really inflate significance because it’s an all or none thing). I guess I still don’t understand the problem. But the significance tests are not important to my argument anyway. My argument has actually just been a question: Why do economists think that taxes are recessionary? Is it based only on theory or are there some observations that would lead one to this conclusion. What seems like the most relevant observation to me is the actual relationship between tax rates and growth/unemployment. What I observe are correlations suggest no relationship between taxes and recession (if the correlations are not significant) or that are the opposite of what one would expect if taxes are recessionary. So am I looking at the wrong data to tell whether taxes are recessionary? If so, what data should I be looking at? What data were economists looking at (if any) that led them to this conclusion.

Anyway, you’ve been talking about “strong” correlations of 0.5, which to me indicate that there’s no relationship worth paying attention to. The chances of making a correct prediction of the sign of a change (not even the amount) are only 2 out of 3, which means an error rate of 1 in 3, which seems pretty bad to me. Of course I’m always thinking of the next prediction concerning an individual case, whereas prognosticators are more concerned with the track record involving all the cases. If you’re right 2 times out of 3, that looks pretty good on your resume, but people involved in the individual case are looking at that 33% chance of a wrong decision which could affect them very seriously. I wouldn’t want to make any important decisions for myself or someone else with that large a chance of getting it wrong.

If I said that a correlation of .5 is “strong” then I misspoke; the only thing I should have said is that it is significant (if it is). Read my previous post for the meaning of “significance” in statistics; it does not mean “strong”.

And I am not presenting these correlations in order to predict or recommend what individuals should do in their daily lives. I am simply presenting them as measures of an observed relationship between variables (aggregate variables at that), a relationship that seems pertinent to the question of whether, as economists believe, increased taxation (an aggregate variable) is associated with a decrease in growth (another aggregate variable) or an increase in unemployment (another aggregate variable).

These correlations have nothing to do with making statements about individuals or causality. I see them as equivalent to, say, the correlation between cusor and handle or between the cursor on two trials with identical disturbance in tracking tasks. These correlations are typically on the order of 0.0 to 0.4. These low correlations show that there is something wrong with the idea that the cursor movements guide actions. Of course, what is wrong with that idea is that it leaves out the fact that actions are guiding cursor movements while cursor movements are guiding actions. Maybe the same thing is going on with taxes and growth.

The way we know what’s going on with the cursor-action relationship is through modeling. If that’s how economists know that something like circular causality is involved in the tax-growth relationship then I would just like to see the modeling that reveals it. It would not disturb me at all to find out that there is clear evidence, based on modeling, that taxes really do impede growth. I just want to know how economists know this is true. Because just looking at the data it looks like taxes either have nothing to do with growth (if the correlations are really not significant) or they are positively related to growth.

Best

Rick

···


Richard S. Marken PhD
rsmarken@gmail.com
www.mindreadings.com