In some of my earliest articles, from two years ago, I provided various methods of looking at the relationship between taxes and the economy. One of the more interesting themes I kept seeing from conservatives in response was this: if this is true, why on earth haven’t we heard it from more authoritative sources? I suggested that, until recently, it was hard to do the analysis, simply because of access to the data and the tools necessary to analyze them.
Regardless of the reasons, this Business Insider article corroborates much of what I was saying back then. Apparently, the non-partisan Congressional Research Service agrees with me.
For what it’s worth.
From the article (emphasis mine):
“In fact, although correlation is not causation, when you compare economic growth in periods with declining tax rates versus periods with high tax rates, there seems to be evidence that tax cuts might hurt growth. ”
It is so noted that empirical evidence is accepted as sufficient, without a factual, demonstrable rebuttal delinking that evidence from the conclusion.
What strikes me most about that article, Michael, is that this is stuff most of us have been saying for years, even while conservatives go on and on about the evils of taxes and how tax increases are always bad.
In point of fact, tax rates significantly higher than what we have today — especially top marginal tax rates — seem to be much healthier for the economy.
I’m beginning to think we shouldn’t call this point of view “conservative” any more. It’s simply a bit of partisan dogma. It’s almost religious in the amount of fervent faith-despite-evidence attached to it. Maybe what we’re talking about is simply Antitaxism, or Taxphobia, or perhaps Revenue Denialism. There’s nothing particularly “conservative” about it, any more than a belief in any other religious dogma is “conservative”. Faith transcends the liberal-conservative dualism.
DC,
Yes, but it’s one thing to theorize about it. It’s another to quantitatively analyze it to provide evidence to back it up. That was the reason I was particularly intrigued by the analyses I did. And now I have some vindication in the form of corroboration from the CRS.
Michael, it would seem that tax cuts create what I’ve heard called a Midas Trap where capital is extracted from the Market and sequestered into a less efficient location than it might otherwise be if ‘redistributed’.
If you view the economy though a monetary lens, looking at M1, M2, and M3 variation over time leads to additional data to factor in. An interesting data site.
http://www.shadowstats.com/charts/monetary-base-money-supply
Has anyone thought that maybe since gutting taxes raises income for those higher earners that there’s _less_ incentive for them to hire more people and grow their business? Why hire and pay people when you’ve gotten a free raise?
As astounding as it may sound, Romney only leads Obama by 2 points in AZ in today’s state by state polls. If true, it’s a GOPer armageddon.
Never doubted your math for a minute, Michael. At least, not once I double-checked it for myself.
Must feel nice to get some “official” confirmation.
cc,
I have more to say about this in tomorrow’s (really long) Reëlection Watch.
mclever,
It does. I read the article and exclaimed, “Ha! Told ya so!”
Arizona like a number of red states has a strong agricultural base. Well a number of states period of both colors but some red ones are very farm orientated. Now it looks as if a large number of them are looking at losses in the billions because all of the Republican generated activities has driven off the Latino farm labor they needed to survive. That it is quite likely that many of those same farmers and agribusiness men that are now watching their crops rot in the field are Republicans and helped elect the people that created the problem for them it does look like a serious case of unintended consequences. While it will hurt all of us through higher food prices one has to wonder what is being spoken to big ag’s former pet politicians.
I’ve heard Peter King has problems in the NY 2nd as well.
@Michael,
Congratulations on getting some confirmation. Well done.
I have a few observations that I would like to get some feedback on, however.
I notice that the graphs start in the year 1945. I’m mean, really? As I’ve said a thousand times, we had just annihilated the rest of the civilized world. Americans had put away a lot of savings. There was pent up demand. Our factories hadn’t been blown up. We were able to produce products for the rest of the world that lay in ruin. Marginal tax rates could have reached 100% and the the American economy would have boomed. The highest marginal tax rates could have been 10% and the economy would have boomed.
To a much different degree, the same can be said about other decades since. There waere a lot of factors in play in the 90’s that drove the economy other than tax policy. We had a tech and dot com boom. A housing boom. They all ended in busts. Borrowing hit levels never seen before. Saving rates plummeted. Home values skyrocketed leading to equity to borrow against.
We paid for that in the 2000’s.
I’m not sure how a conclusion can be made when comparing time periods that are so drastically different. I would like to see a similar study with “All things being equal” .
Another thing. Has Keynesian counter-cyclical policy been wrong all these years to suggest lowering taxes to jump start an economy and raising taxes to cool down an economy?
GROG,
Thank you.
Yes, really. That’s about the right time to start an analysis of federal income tax in the US. One can argue about extending it back to 1918, but the federal income tax between 1918 and 1945 was so small relative to the overall economy as to have no correlation at all. Would I like to have more years to analyze? Sure. But you go to war with the army you have.
And, with the army we have, there are enough years to produce statistically significant results. Not ideal, but significant enough to start drawing solid inferences through statistical analysis. The time graphs are pretty, but you really should look at the statistics (and statistical graphs) instead. They tell the story I was telling.
You did. It’s the same one we’re both talking about. That’s what regression analyses do.
If your question is a binary one, then yes, it has been wrong. But that’s more because the signal to noise ratio in a simple raise/lower all taxes model is horrid. This is why I delved into the quintile-based model in Take 2. At that point, things became much clearer…it depends on whose taxes you are raising or lowering. So, no, the fundamental theory isn’t wrong, but it needs to be applied in a more targeted fashion to really be effective.
Incidentally, that’s also probably true for another aspect of Keynesian countercyclical economics. Not all government spending is created equal from a countercyclical perspective. Merely paying people to dig and fill holes won’t really do much. The money needs to be spent in a way that creates economic value for the nation. Infrastructure is a great way to do that.
If regression analysis can make the 1945 post war economy equal to the 2001 economy, then I have such little understanding of it to speak intelligently about it.
GROG,
Perhaps so. I’ll take a stab at explaining it, though. But first, I want to make it clear that it doesn’t actually make them both equal, per se. What it does is look for similar first-derivative behaviors over time. The idea is that you’re going to have a lot of random things happen at random times (noise), so if you have enough instances of your input signal and enough data points, you can separate the output signal from the noise.
Things start out so blurry that you can’t see anything, but the more data points you add, the more sharply into focus the signal becomes. In that sense, it’s like the distinction between weather and climate. Weather is, say, the temperature in Cleveland today. Climate would then be the collective temperatures on this day (or week or month, depending on desired level of granularity) of the year in Cleveland, over time. Just because the temperature averages 75 degrees on September 22 doesn’t mean you won’t have a snowstorm and a temperature of 28, simply because it’s September 22. It does mean, however, that the temperature is far more likely to be in the mid-70s than in the upper 20s.
This is what statistical analyses do for us. So in this case, it’s far more likely that tax increases on the wealthiest Americans lead to greater economic prosperity…within the bounds that we’ve taxed them since the end of WWII. And, moreover, from the more detailed analysis I did, it seems that the greatest benefit comes when the increase leads to a marginal rate somewhere just below 50%. To some degree, that’s hard to tell with absolute authority, since I’m having to trade fewer data points (and thus fuzzier results) for a narrower range of marginal rates (and thus an otherwise stronger signal).
Does this make sense to you? Does it help clarify what’s going on with this analysis?
Michael,
Just out of curiosity, is there any reason you didn’t link to the CRS study directly? Why The Business Insider article instead?
Does this make sense to you? Does it help clarify what’s going on with this analysis?
No. (My fault, not yours.)
Things start out so blurry that you can’t see anything, but the more data points you add, the more sharply into focus the signal becomes.
How does the resgression correct for the difference between 1946 when the U.S. had virtually no economic competition, and the 2000’s when shipping costs from China became so low that we could no longer compete with them in the manufacture of thousands of goods?
Where did it correct for 911? Where did it correct for post 1945 when Americans where stashing away 20% of their income that was unleashed after the war was over?
In that sense, it’s like the distinction between weather and climate.
First of all, you have hundreds of September 22’s to draw data from. There have only been a handful of tax rate changes over the past 65 years, all instituted during vastly different circumstances which may or may not have had huge impacts on economic growth.
Secondly, it seems you could simply take the average temperature on September 22 over the past 300 years and have a pretty good idea that there’s a better chance of it being 70 deg today than 20. The question of how tax rates affects the economy seems to much more complex than that.
GROG,
I ran across the BI article in one of my feeds, so it was the quick link to provide. Nothing more than that.
The only reason I asked is because I didn’t see a link to it in the BI article and I had a heck of a time finding it elsewhere. But I finally did here.
GROG,
The answer is easier to understand when we look at this:
Excellent observation. And, in economics, we have hundreds of months, or dozens of quarters or years. So, to understand this better, let’s start with a binary input, of two possible tax rates. You’d be able to compare the average under Rate 1 to the average under Rate 2. It works better if you have a mix of times for each of the two rates (Rate 1, then Rate 2, then Rate 1, then Rate 2…etc), because that reduces the likelihood that there’s a time-based trend coinciding with the rate changes.
It’s more complicated when you have a handful of different rates, but the concept is basically the same.
You’d think so, but it isn’t really. It’s just that we need to acknowledge that we have a bunch of inputs, and tax rates are but one of them. So we can say that the rate will influence the economy, but won’t likely be the primary (and certainly won’t be the sole) input.
Michael,
It’s inconceivable to me how it’s possible to correct for such things as economic, political, and technological differences between 1945 and subsequent decades. Not to mention the fact that taxes are lowered during down economic and taxes are raised when they’re good. Why is five years a reasonble future date to compare? I’d like to see the study that led to that determination.
I’ll have to study up on it some more.
Grog,
As I understand it, what the regression analysis is giving us can maybe be called the direction of influence, not so much the absolute rate of change or the absolute value of the change or even the target — that is, what it is going to change to. In other words, we can say that pretty much regardless of economic conditions, doing “A” will push things in “X” direction.
If things have pretty much always been pushed in “X” direction when you do “A”, you can pretty much guarantee that if you do “A” then things will be pushed in “X”.
Sometimes in the past, when “A” was tried, things were already moving in the opposite direction, “Y”. Doing “A” might not suddenly reverse the direction of things and force them back into direction “X”. But if things slowed down, if the rush toward “Y” was slowed, we can then say that yes indeed, even in this situation, doing “A” forced an “X” influence. Looking at the change in the rate of movement will reveal these influences, even if movement continues in the wrong direction. (That’s what the “first derivative” gives us, if I’m not mistaken.)
The whole point of doing regression analysis is to reveal these influences in order to show what effect doing “A” has, regardless of the presence of other factors, and regardless of whether things were already moving in direction “X” or “Y” or some other direction “Z”.
Does that help?
By the way, Grog, we have often discussed that a) not all taxes are equal, and not all have the same effect, and b) not all spending is equal, not all spending has the same effect.
Why do you keep asking the same question, “Has Keynesian counter-cyclical policy been wrong all these years to suggest lowering taxes to jump start a economy and raising taxes to cool down an economy?” You’ve asked that before, and the answer is always the same. Keynes didn’t really differentiate between types of taxes, or where the tax rates were, or who got taxed. We have more information now, and can perform analyses at greater granularity than he could. Every time you ask that question, you get the same answer. Do you expect the answer to change?
GROG,
It’s inconceivable in the same way that it’s inconceivable that we can account for the various effects of shifting oceanic currents in predicting likely weather. By having enough inputs that are not directly in sync, one can tease out the statistical correlations. They break down when the inputs change at precisely the same time, because it is then impossible to separate them. A long as there are lags among them, the picture becomes possible to discern.
Why five years? My guess is that they tried other lags and found much lower statistical correlation.
I don’t want to beat a dead horse here, so I’ll be brief.
It appears that I misunderstood the BI article and thought the CRS used a 5 year lag. It appears they actually used a 1 year lag instead of taking a long term look at tax policy. I don’t know of anyone who claims tax cuts will have significant affects on economic growth just one year after implementation.
Also, the CRS study basically says that nothing has any affect on economic growth, including population growth, education, and government spending. In my opinion, despite the regression analysis that is supposed to seperate the signal from the noise and all that, the CRS methodolgy has no way of dealing with the complexities and differences in the economies over time and factors that are so difficult to measure, that’s why they found nothing has any affect on economic growth. Not exactly a “Bombshell”.
I’ll bow out now.
Grog,
As the BI article and the CRS study still show “the bottom line appears to be that low taxes do not spur economic growth and DO cause greater economic inequality.”
There’s actual, you know, data, to support that statement.
You are certainly entitled to your opinion. There’s no evidence to support that opinion, however.
In fact, the evidence rebuts that opinion. GROG has not rebutted the evidence, simply voiced his opinion.
Our friend GROG is simply unwilling to accept evidence counter to his initial belief.
Max,
Our friend GROG is simply unwilling to accept evidence counter to his initial belief.
I could trudge through life like an unthinking, intellectually bankrupt zombie who questions nothing and believes everything that is spoon fed to me. But what fun would that be?
GROG,
It is perfectly FINE to question!!! I strongly encourage it. I beg people to rebut MY arguments.
But your quote: “Our friend GROG is simply unwilling to accept evidence counter to his initial belief.” without including my preceding statement takes what I said totally out of context.
“GROG has not rebutted the evidence, simply voiced his opinion. ” By NOT rebutting facts in evidence, you really do not “question”. You fail to produce and prove alternate hypothesis that debunks the assertion. That is not “questioning”, it is only a rejection of fact, with a subsequent substitution of opinion. That opinion may, or may not, be true. As I said during the recent discussion on Reagan policy, “all you have to do is give a proof delinking the assertion from the conclusion”. Proof by contradiction.
Context matters.
For most critical thinkers, there comes a point when they stop questioning whether, say, the Earth’s atmosphere is primarily nitrogen, or whether regression analysis does what it says it does.
There are places between “questions nothing” and “accepts nothing”, between “believes everything” and “denies reality.”
DC,
I refuse to accept that.
How about refusing to accept anything but the validity of a regression analysis that finds that nothing has an affect on economic growth?
I don’t think it found that. Can you show me where the paper claims “nothing has an affect on economic growth”?
GROG,
You’re making a mistake common to those who are somewhat new to regression analyses. A strong correlation is good evidence for causality, but a lack of an apparent correlation is not, in fact, good evidence against causality.
Perhaps should have started with ANOVA and worked up from there!
Max,
To ASUPERNOVA.
It’s a bit curious to me that Mr. Hungerford, a financial supporter of Obama, puts out a “non-partisan” study 7 weeks before the election, supporting the idea that tax rates have no affect on the economy.
Am I to assume I won’t be seeing a similar paper analyzing how government spending has affected economic growth, with a 1 year lag, beginning in 1945?
A) His paper didn’t show that. As the BI article summarized, the paper showed “the bottom line appears to be that low taxes do not spur economic growth and DO cause greater economic inequality.” I wouldn’t categorize “greater economic inequality” as being “no effect on the economy.” Further, his paper showed that increasing top marginal tax rates might have the effect of increasing economic growth. I also would not call that “no effect on the economy.”
B) Are you implying something about the timing of the release of the paper? If so, do you have any evidence to support your innuendo?
A) “Further, his paper showed that increasing top marginal tax rates might have the effect of increasing economic growth.”
No, it doesn’t say that at all.
B) Let’s see. An ardent supporter of Obama and the Democratic party. Study is released 7 weeks before the election. He does not appear to have any study prepared comparing such things as government spending and economic growth, or tax rates and revenue. So yes, I would say Mr. Hungerford released this study with an agenda in mind.
Sorry, I misremembered. I stand corrected. The correlation of faster growth is with higher capital gains tax rates. (Again, this seems in contraindication of your contention that the study claimed “nothing has an effect on the economy”). As the BI article said, in describing Figure 5,
Grog,
The study also showed (figure 3, page 7) that private savings tends to rise along with increasing top marginal rates and also along with increasing capital gains rates. You have said increased personal savings is good.
The study also shows that investment rates tend to fall with increasing top marginal rates and increasing cap gains, but you seem to think personal savings is what drives investment — that doesn’t seem to be true.
The study also shows (figure 4, page 8) that labor productivity rises along with increasing top marginal rates. Productivity seems to fall with increasing cap gains rates, but this is a much smaller correlation.
Increasing top marginal rate doesn’t seem to affect GDP growth, but (as I said) there is a weak correlation between increasing GDP and increasing cap gains rates.
And again, the study definitely shows wealth disparity increasing as the top marginal rate and the cap gains rate decreases. The share of wealth owned by “labor” increases (fig 9 page 15) as the top marginal and top cap gains rates increase — though, since this doesn’t hurt GDP growth, one cannot argue that the gains by labor come at the expense of the rest of the economy.
All of which shows that many things affect the economy, and often in ways that differ from conservative economic expectations.
In #34, dc asked “
Can you show me where the paper claims “nothing has an affect on economic growth”?” in response to GROG’s assertion of “
How about refusing to accept anything but the validity of a regression analysis that finds that nothing has an affect on economic growth?“
Instead of answering dc’s question, GROG starts another deflections by making an ad hominem attack on the author of the study with NOTHING to show for it but innuendo.
GROG, such stunts as this is why I feel justified in making charges against your “methods” as I did in #30. You have NO REBUTTAL and then, when you are asked for something, you make an assertion THAT THE FACTS do not support, then when called on THAT, you resort to a baseless ad hominem.
I believe, IMHO, that you have no ground to take insult when so accused, my friend.
GROG,
In this context, Hungerford works for Congress, and thus must have had a request from some Congressional Committee to do the research. Might the Committee members have requested it in the expectation that it would help Obama? Possibly.
Not unless a Congressional Committee asks for it. The analysis I’ve done on this particular relationship has been inconclusive.