This is turning up all over the place — at Brad DeLong’s, Crooked Timber, and this pair is from Cosmic Variance — it’s the most sublimely, awesomely, wickedly stupid example of fudging a curve ever. The two graphs below have exactly the same data points, and the only difference is the curve that was ‘fit’ to the distribution. Which one looks plausible to you?
The one on the left looks sensible and simple, and looks like it was actually drawn with some consideration of the data. The one on the right … not so much. I have no idea how anyone could think that particular curve belongs in there.
Now guess which one was actually published?
Hint: it was published in the Wall Street Journal editorial pages.
Someday, somebody’s going to write a book about the shenanigans at the WSJ that allows a clown college, the editorial staff, to exist and thrive within the bounds of an otherwise staid and sort of boring, but respectable, newspaper. That they could actually publish something like the ridiculous abomination on the right and no one said “Wait! What about our credibility?” is merely symptomatic of some really interesting pathology going on there.
How about neither?
The right hand graph is of course a joke. But the straight line on the left depends for most of its significance on two points, at least one of which is debatable. Does the UAE really have tax revenues as 0% of GDP?
I bet it was fit to all points by the method of least squares of error.
According to the chart it has no taxes on corporations, so I think that’s plausible.
I bet it was fit to all points by the method of least squares of error.
According to the chart it has no taxes on corporations, so I think that’s plausible.
What is especially odd is that one could draw a downturning line into the data that would still be reasonable, since the “U.S., Germany group” of data looks to be a little bit below the “U.K. group”. So even if he wanted to force the line to support his theory, he could have done it in a plausible way. Instead he went out out of his way to make sure that everyone could instantly understand he is an idiot. If nothing else, this shows more consideration for the readers’ time than most idiots have. Perhaps we should thank him for that.
One thing to consider: it’s a graph of gross government revenue plotted against only part of its income stream. It’s hard to draw a general conclusion on the basis of that selective fragment.
It doesn’t stop the WSJ though!
The one on the right gives a superb fit to Norway. Maybe the WSJ is looking to improve its Norwegian readership.
I say we all remove our frontal lobes and bask in the rays of specious reasoning. When done, we’ll never question the OP/Ed section of a ‘respectful’ newspaper, ever.
This is an *egregious* example of confirmation bias. Whoever drew the second curve clearly has no trouble believing that black is white, and can find evidence for that proposition wherever they look. Breath-takingly obtuse.
I don’t understand why the UAE point should be included in the fit at all. If they have no corporate income tax, then it’s an anomaly, and doesn’t give you any information about the hypothesized correlation.
And if you take out the UAE point at the origin, the supposed upward trend is no longer obvious.
Seeing the right hand curve go through an outlying point should alert any observer that the line is false. But I don’t disagree with the general notion that as tax rates go up, revenue may plateau at some point. As we all know, both corporations and people sometimes cheat. So I don’t think the flat line on the left is correct either. Think of it this way, what’s your incentive to operate a business if the tax rate is 100%? That alone should tell you that the line is not straight. What I really want is a graph of tax evasion as tax rates increase.
i think “laugher curve” is misspelled on the graph on the right.
Laffer seems somehow appropriate, doesn’t it?
For those interested in fitting a decent curve: it looks to me like the data are heteroscedastic, so a weighted regression might be called for. Does anyone know if the original data are available? MaxSpeak has some data, but it’s not the same (I don’t know if it’s only some of the data, or whether there are differences in the sources)
Bob
Both look like pretty horrid fits, I bet the correlation coefficients (Pearson’s r, Spearman’s, takeyerpick) are all pretty awful.
Has anyone considered the possibility that the curve isn’t meant to link up the data points, but to show where a theoretically-derived boundary is? The points would only be there to show where individual countries are, relative to that boundary.
Re: #9
I agree that it is obvious that if you keep raising tax revenue eventually tax inflows will decrease. But we can’t really know what happens at extreme values because the data in the graph happens to pretty much only exist between tax rates of 20% and 40%.
The thing about the Laffer curve, in my understanding, is that it is just a cocktail party theory and that there has never been any good work to show that the U.S. is on the part of the curve where a decrease in tax rates would lead to more taxes brought in. Conservatives just assume that we must be because that fits the agenda they already want to pursue for other reasons.
It always astonished me that conservatives, who generally want to reduce government in most ways, use an argument for cutting taxes which assumes that increased government revenue is a good thing. The duplicity of the argument is rather worthy of someone from the discovery institute.
I, two, think both graphs are misrepresented. But, data is data – you’d think plotting data / numbers would be simple and straightforward. (see below)
Laffer actually attributes his ‘Laffer Curve’ to several sources, most notably Keynes.
Winston Churchhill was quoted as saying: “If you put two economists in a room, you get two opinions, unless one of them is Lord Keynes, in which case you get three opinions.”
I reconstructed the data by eyeballing it; it looks pretty close to me. It’s over at pastebin, if you want to have a look. I reconstructed it with gnuplot; I can paste the source in here, if anyone thinks it’s nonobvious.
Here’s the graph with outliers included; best fit linear, quadratic and cubic.
Here’s the graph with outliers (Norway and UAE) excluded; best fit linear, quadratic and cubic.
Weirdly enough, if they’d done a quadratic fit, the data would have actually looked like a more plausible version of what they had. Can’t imagine why they fudged it.
Oh, man, that’s embarrassing. Just swap the captions on the X and Y axes; I put this together in a hurry, and didn’t double-check that. Kick me.
Fixed with-outliers plot.
Fixed no-outliers plot.
That’s what I thought too — for about two seconds, until I read the axis labels and recalled the definition of the Laffer curve.
Also, one of Kevin Drum’s observations is worth repeating:
I used o do IT for a newspaper. If they had a bunch of raw numeric data they’d come to me to get it all nice and graphed.
I’d graph it up, and take the graph to (who else?) the editorial graphic artist, who would pretty it up.
Any best-fit line or curve I’d include would end up in some fairly random spot on the final graphic. It would be roughly shaped the same, but inevitably hanging on a few data points. The copy editors would always approve the graphic, no matter what protests I threw their way. They believed the curve had to hang on a few points or it didn’t look relevant.
So, this is more evidence supporting my theory that journalists of all stripes understand exactly two things:
1) Diddly
2) Squat
If, for some bizarre reason, that’s what was actually being attempted, then the graph would seem to indicate that most countries are weirdly inefficient.
Of course, the problem that the x-axis is corporate taxation makes drawing conclusions from the data difficult. How much of each country’s GDP is composed of corporate earnings?
Ray S: Nobody disputes that 0% taxation will give no revenue, and that 100% taxation will give no revenue. The problem with the Laffer curve (and what we may generously term the theory behind it) is that the trajectory in between the end points is a cleaned-up cocktail-napkin drawing of what a few people thought should be the common-sense behavior, blissfully unanchored by any actual data to back it up.
Caledonian raises a valid point, in which case the WSJ could simply be pointing out that practically all the nations of the world diverge from some theoretically well-defined standard. In that case, however, some vigorous defense of that supposed theoretical standard would be necessary. (One tends to doubt models that fit none of the data.) The curve in the WSJ graph is spun out of thin air. Apart from the totally obvious conclusion that both a 0% tax rate and a 100% tax rate will produce no revenue (the former by definition, the latter by essentially total evasion and economic collapse), the Laffer curve has no significant predictive value. Who says it has to be a smooth curve with a unique maximum? (Or even well-defined multiple maxima?) Who can model what is likely to be chaotic? Martin Gardner poked fun at the Laffer curve years ago with his rendering of a curve that looked like a plate of spaghetti — but it matched the improbably smooth Laffer curve at the two endpoints. [Link]
According to the comments on Sean’s site (and as Caledonian alluded to), apparently the curve is meant to indicate a boundary of some “ideal” area for taxation, rather than a fit to the data. So everybody under the curve is “ideal” and over it is not ideal.
There is not, according to the folks there who read the article, any justification whatsoever presented for that claim.
And, hey, look! The US is on the wrong side of the curve! We’d better cut taxes ASAP!
Aren’t the dependent and independent variables switched in those graphs? Isn’t the independent variable (tax rate here) conventionally put on the horizontal axis?
Also, doesn’t this graph predict two tax revenue values per given taxation rate? Of what value is such a function? That’s the complaint I had with grendelkan’s graphs (#16) – they ‘doubled back’ on themselves. Then I noticed the flipped axes. To be useful for selecting a tax rate, wouldn’t you want a monotonous (I think that’s the word) function?
grendelkhan – thanks for the data. My eyeballing impression was that adding the quadratic doesn’t improve the fit, and the stats bear that out (the R2 goes up by 0.16%: it’s about 30%). The weighted fit is pretty similar, and it depends a bit on how you deal with the UAE (do you give it infinite weight, or what?).
Bob
I’m going to confess here that it’s been a while since I took statistics, and it’s not my day job.
I should also add that there was a very high standard
Stupid “Post” button!
As I was saying, gnuplot gave me a large standard error for some of the coefficients (up to 900% of the actual coefficient value); I think that may mean that they’re not very reliable. The weights were all equal; I didn’t do anything other than use gnuplot’s built-in curve fitting. If you do so, you’ll get a ‘fit.log’ file explaining what the coefficients are and how reliable they are.
What’s the r squared on those plots? Because, frankly, neither looks all that reliable to me.
Let’s keep in mind that the Laffer curve presupposes that the government spends the funds it takes through taxation effectively. If it squanders those resources, or even sets them against the systems it is supposed to be nuturing, it can turn into a Laffer flatline.
Dianne: my ignorance of statistics is showing. Do you mean “variance of residuals (reduced chisquare) = WSSR/ndf”? It was 2.90033 for the cubic fit, 2.91742 for the quadratic, and 3.06483 for the linear. Again, you can run the data through gnuplot yourself, and see fit.log for the results.
Obviously to replace one curve with another you need to have some sort of goodness-of-fit measure. Of course, as an ecologist, I am inclined to look for something that levels off. If I had to eyeball a curve to this I would expect the best-fit curve to peak around Canada and then either level off or decline a bit. And quite frankly, if they had done that, they could have made a much more convincing argument that corporate tax rates are too high. (I’m not in a position to say if that’s true or not, but it would be a far more convincing argument than using this graphic, which is simply an insult to the intelligence of their readers.)
Grendelkhan @ #17: Thanks. Now I have answers to all the questions/objections I raised. I couldn’t read the captions on the original pictures, and assumed from your graphs that they had the axes backwards.
And sorry for mis-spelling your name.
The U.A.E. distributes the money it makes off oil to its people and it takes in no taxes from the people that live there. In essence the tax rate on the people and businesses that operate out of the nation would technically be negative (they give money to the people not take it). It’s a fairly progressive region as far as middle eastern nations go as well (I say region because technically it is a group of allied nations that for a single nation like thing).
1. The WSJ article is ambiguous in its wording as to whether this Laffer curve is supposed to be a fit of the data points shown or a theoretical prediction independent of those data. As Zeno pointed out, if the latter, well, that’s some pretty disconfirming evidence right there. Cut the pie anyway you want, it’s a horrible graphic, truly disappointing.
2. Everybody keeps saying that it is impossible to dispute the Laffer Curve in principle, since 0% corporate taxation and 100% corporate taxation have to both return 0 revenue; therefore there will be a maximum somewhere between and the curve will be an inverted U shape. Though 0% taxation has to give no revenue, I have trouble with the concept of 100% taxation…
For example, Assume a corp makes $100 profit. At 50% taxation, the government’s revenue is $50. Now let’s say they increase to 100% taxation. The assumption is that the corporation will leave, fold, or cheat, such that the gov will take no money from them. But what about partial cheating? If that corporation cheats and hides 50% of their profits from the government, they will appear to have a profit of $50, and the government will take it all, but it would be the same amount of revenue as if they had set it to 50% taxation. So it could be possible for the Laffer Curve to actually just have a plateau. 100% taxation needn’t produce 0 revenue.
PZ (or anyone), can you post an excerpt of the WSJ editorial in question? I don’t subscribe, so I have no idea what anyone’s talking about. All I can see is that graph, not their interpretation of it.
Epistaxis:
The original Wall Street Journal article
Reading the editorial, there is absolutely no attempt to represent the curve as a fit to the data.
It’s often been pointed out that corporations don’t pay taxes, they collect them. Moreover, if you raise corporate taxes, international companies will redistribute the allocation of their profits to pay taxes in the state that levies the lowest corporate tax. This principle was the basis of a significant part of Ireland’s ‘economic miracle’. For example, for a while Coca-Cola Ireland was the only Coca Cola subsidiary within the EU to be making a profit (and it was very profitable). Why? Coca Cola exported their syrup first to their Irish subsidiary, which then sold it at an exorbitant price to the other European subsidiaries, who all managed to lose money. The Irish subsidiary made a windfall, and so did the Irish government, which, although it was levying tax at only 12.5%, was doing so on a massive profit base.
The editorial makes a good point. Note the two per capita richest countries in the EU are on the top left: lowest corporate tax rates, and highest percentage of tax revenue from corporate tax.
Shorter WSJ:
The U.S. imposes neoliberal economic policies on the whole world with the “hidden fist” (tm Thomas Friedman) of its military force, and clearly these policies are the ideal ones–after all, everyone is choosing to implement them.
Which part of
did you miss?
Unless such a practice would result in fines and jail time for tax evasion.
While there is certainly a degree of wiggle room in which a company can reasonably choose to assign profits, that wiggle room is equally certainly not infinite. At some point it becomes sufficiently obvious that what’s going on is simply tax cheating and, if your government is reasonably competent and non-corrupt, at that point the corporation in question (and preferably its shareholders as well) are going to feel like a planet fell on them.
A trick that stopped working once the other countries in the Union wised up to it and started aggressively hunting tax cheats. To continue your CocaCola example, the Danish government simply decided that the absurd prices for syrup were fake, estimated a real price (and probably hacked a perceptible percentage off of it as a slap on the wrist) and classified any payment in excess of this as a gift to the mother-firm. A non-tax-deductible gift, that is.
IIRC, Coke sued. And lost.
From the graph I assume that you’re talking about Luxembourg and Norway. In that case, I would suggest that some remedial geography would be in order:
Luxembourg is a cartoon economy – it’s a city-state, for crissake. It is patently absurd to the point of ridicule to assume that the way a city-state can run its economy is scalable to the size of a real country.
Norway is an oil economy. Give me the Norwegian oil fields and I could easily turn any country of comparable size into the richest country in the Union, on a pr. capita basis. Regardless of the economic system.
– JS
I happen to own a microeconomics textbook which also discusses the Laffer Curve with a straight face. (See my books page.)
Zeno: And Daniel Dennett uses the same point to illustrate many mistakes concerning thinking about consciousness. (Which was, incidentally, where I first heard of it. I mean, I lived through Reaganomics, but as a Canadian kid … so it was still … laffable?)
bPer: I think you must mean “monotonic”, but even then, the world might not be obliging …
Thanks for the link, cm (#37).
WSJ says:
Even if we assume for the sake of argument that the curve in that graph fits the data, isn’t this conclusion still bass-ackwards, if we take it literally? The “wrong” side of the curve is under it: because the US data point is above where the curve would be if it were extrapolated, it means we’re taking in more revenue than we expect for our tax rate. Somehow, we’re beating the curve.
Regardless, we can still take the Laffer Curve somewhat seriously if we throw out the WSJ’s fudged fit and apply a plain old quadratic, like grendelkhan did (#16 & #17). Then, the data can be interpreted with a straight face as indicating the US could increase its revenue by decreasing tax rates.
I hope for Kevin Hassett’s sake that the WSJ’s curve came from some WSJ graphic artist’s imagination, and not his own. I also hope his quote about being on the “wrong side of the curve” wasn’t supposed to mean what it sounds like it means.
Which part of
The nearby chart shows the Laffer Curve effect from business taxation.
did you miss?
Which part of that sentence says the curve is a fit to the data points?
The editorial makes a good point. Note the two per capita richest countries in the EU are on the top left: lowest corporate tax rates, and highest percentage of tax revenue from corporate tax.
From the graph I assume that you’re talking about Luxembourg and Norway. In that case, I would suggest that some remedial geography would be in order
Yes, some remedial geography is indeed in order, unless Norway became a member of the EU surreptitiously in the last 60 minutes.
http://en.wikipedia.org/wiki/European_Union
It’s curious how often attitude is coupled with ignorance.
As a MA student in economics (focusing on econometrics), this kind of thing really aggravates me. I went back to the author’s source (OECD 2004 data) and reconstructed his analysis. Below is the R output from the analysis:
> data.laffer
corp.tax.rate corp.tax.rev
1 0.00 0.0
2 8.50 2.5
3 12.50 3.6
4 16.00 2.2
5 18.00 1.3
6 19.00 2.0
7 19.00 2.5
8 22.10 3.4
9 22.88 5.8
10 25.00 2.9
11 26.38 1.6
12 27.00 3.5
13 28.00 3.2
14 28.00 4.8
15 28.00 5.5
16 29.00 3.6
17 30.00 2.9
18 30.00 3.2
19 30.00 3.8
20 30.00 5.7
21 33.00 2.3
22 33.00 2.8
23 33.00 10.0
24 34.00 2.3
25 34.00 3.6
26 34.50 3.1
27 35.00 2.2
28 35.00 3.3
29 35.00 3.4
30 35.43 2.8
> summary(lm.laffer)
Call:
lm(formula = corp.tax.rev ~ corp.tax.rate)
Residuals:
Min 1Q Median 3Q Max
-1.7578 -0.9639 -0.3466 0.3418 6.1886
Coefficients:
Estimate Std. Error t value Pr(>|t|)
(Intercept) 1.39629 0.98875 1.412 0.1689
corp.tax.rate 0.07319 0.03564 2.053 0.0495 *
—
Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ‘ 1
Residual standard error: 1.677 on 28 degrees of freedom
Multiple R-Squared: 0.1309, Adjusted R-squared: 0.09982
F-statistic: 4.216 on 1 and 28 DF, p-value: 0.0495
> summary(lm.laffer.quad)
Call:
lm(formula = corp.tax.rev ~ corp.tax.rate + corp.tax.rate.2)
Residuals:
Min 1Q Median 3Q Max
-2.01737 -0.77298 -0.37013 0.08982 6.35579
Coefficients:
Estimate Std. Error t value Pr(>|t|)
(Intercept) 0.081878 1.466175 0.056 0.956
corp.tax.rate 0.237916 0.141082 1.686 0.103
corp.tax.rate.2 -0.003938 0.003265 -1.206 0.238
Residual standard error: 1.664 on 27 degrees of freedom
Multiple R-Squared: 0.1753, Adjusted R-squared: 0.1142
F-statistic: 2.87 on 2 and 27 DF, p-value: 0.07414
I have posted the relevant plots to flickr:
http://www.flickr.com/photos/90829310@N00/812709857/
http://www.flickr.com/photos/90829310@N00/812709881/
On the heteroskedasticity comment: this issue would affect the bias of std. errors, but the OLS regression estimated coefficients remain BLUE.
Also, it should be noted that the UAE (with a 0% tax rate) is not in the OECD data; the author added it in to his dataset.
I think the neo-Laffer curve gives a better fit.
#1: By my understanding, the Laffer curve (which is actually about *total* tax revenues, not as a percentage of GDP – part of the theory is that high taxes reduce revenues because they discourage people from working) is supposed to hit zero at tax rates of 0% and 100%, so the WSJ’s curve (which looks to plunge to zero by about 33%) would be even worse as a representation of theory than it is as a curve-fit to the available data.
If I understand correctly, it’s quite consistent with theory for tax revenues as a percentage of GDP to remain steady, or even rise, as the tax rate climbs to 100% – it’s just that the GDP drops to zero, taking the total revenues with it.
#2: Don’t spend too much time trying to decide whether linear best-squares, or quadratic, or what-have you, are the best ways to fit this sort of data. The sad truth is that *none* of them are going to provide much insight; all of them require assumptions that are grossly appropriate here (equal weighting, normally-distributed errors, etc etc).
For instance, outliers like Norway make it obvious that tax rate is not the only significant influence on their performance; if we start at the USA and move left, we’re not looking at “the USA as it would be if it had a slightly lower tax rate”. We’re looking at “the USA as it would be if it was in a different part of the world, with a different population, different geography, and different history”.
In the end, the linear fit is probably the least bad approach, but only because it’s least likely to generate false confidence in a meaningless ‘fit’.
The UAE is not a country. It is a confederation of independent states. There is no taxation at the confederation level.
#44:
How about we don’t draw any conclusions from these data at all?
@Geoffrey & Epistaxis:
I agree with you. This kind of analysis is definitely a not the way to go if you’re trying to determine the nature of the economic relationship. Doing it the right way would really take panel data methods (perhaps with structural restrictions). In my mind, the point of what I posted is more how badly WSJ’s work is fudged. I would never use any of these regression results to make statements about the effects of policy.
As it has been pointed out Norway is an oil state. It has been raking in dough on this account for many years, in fact more money that it can spend, so the money is stockpiled, for future use.
I am not an expert on how they get the money from the oil, but an educated guess would be that it is collected through some kind of tax, and thus accounts for the high percentage of the GDP.
That is a really good point. Of course this sort of weird double-think regarding government power isn’t limited to the political right. I know plenty of people who are upset with the Bush administrations positions on medical issues like abortion, sex education, euthanasia, stem-cell research, etc. Yet strangely many of these people are also strong supporters of a national health care system, a system which would have greatly increased Bush’s ability to interfere in what should be private medical decisions.
I’ve come to believe that the perception that conservatives want small, limited government and that liberals want large, powerful government isn’t just simplistic – it’s utter nonsense. Everyone likes a strong government when it is taking actions they agree with and everyone calls for a weaker government when it is behaving in ways that they dislike. When people fail to recognize this reality when they are in power, they end up paying the price when the political pendulum swings the other way. The arrogant belief that they would remain in charge forever has been at the root of so many of the republican party’s worst abuses of power during these past seven years.
Regarding Norway and the Union, Harbinson is, of course, completely correct. I suppose I am just so used to conflating Europe, the Union and ‘places I can go without bringing my passport’ that it slipped under the radar. I would be rather curious as to which countries he refers to then, however. It cannot be Ireland, because clearly that country is not significantly above average on the y-axis.
With regard to the sentence about showing the Laffer-effect, however, he is clearly demonstrating wishful thinking. If the curve in question is going to show anything at all, then it must have some cursory relationship to either the data or a theoretical computation. Otherwise it’s just gibberish and does not show anything.
One is always free, of course, to insert gibberish in a text or on a plot. I cannot, however, for the life of me think of any use for such a gibberish-curve that does not point to some serious breaches of journalistic as well as intellectual integrity.
If the curve is a best-fit, then their fitting algorithm is nonsense for all the reasons explained above. If the curve is a theoretical model then the model is nonsense, for reasons also explained above.
I further notice that the only argument for lowering corporate tax rates that has so far been made here is that it facilitates transfer pricing scams. Not exactly the quality you want from a tax law…
– JS
It was mentioned on another web site that one reason for the Norwegian outlier is that that might include royalties for North Sea crude oil extraction. That is one of the dangers of comparing statistics from different countries: the countries may account for them differently. If Alaska was to consider its oil extraction revenue a “tax” it would probably be heavily in surplus, but it would probably be about as high as Norway.
I further notice that the only argument for lowering corporate tax rates that has so far been made here is that it facilitates transfer pricing scams
What an odd way to put it. Differential corporate tax rates encourage a variety of behavior, transfer pricing just being one, that seek to minimize taxes by ensuring profits are made where prices are lowest. Ireland still has a GDP that, as the Economist has noted only recently, is artificially inflated by paper transactions connected to its low corporate tax rate. Recent entrants to the EU have been attempting to do the same thing.
To call such strategies ‘scams’ betrays a telling hostility to corporations, which have a fiduciary duty to maximize return to shareholders, and which therfore are obligated to adopt such strategies, as long as they do not break the law. And the sort of micromanagement of corporate behavior needed to prevent such strategies by law is probably going to be detrimental to the economy.
I have no fundamental problem with corporate taxation; it needs to be judged by its efficiency as a revenue collection measure, its fairness in distributing the tax burden, and its deleterious unintended consequences. One unintended consequence is that, given a reasonably free market, it actually reallocates the tax burden from states with low corporate taxes to those with high taxes. In the example I quoted, the Coca Cola drinkers of continental Europe, probably not rich people by and large, were paying for Irish government services. As a citizen of Ireland still, I applaud their generosity, while wondering at their governments’ policies. More pertinently to where I live now, look at the relative positions of Canada and the US, and remember we also are members of a free trade area.
Ireland, by the way, is certainly above the median on the vertical axis.
I don’t think it can be assumed that 100% taxation yields no revenue. There may be other reasons to operate a business than profit. But I have to take it as obvious that changes in the tax rate affect individual’s decisions as to what activities to perform and that such effect is non-linear. I doubt very much that the WSJ graph and the data used to generate it are meaningful. As already noted, some of those countries derive revenue from sources not available to the other countries; resources that are currently at all time high prices. Was there any attempt to account for that fact? This graph is too simplistic and carries far too many assumptions for my liking.
No one seems to be pointing out that the data is wrong. It’s true that the basic corporate tax rate for Norway is 28%, but their very lucrative North Sea oil is treated very differently – oil profits are taxed at 78%.
By the logic employed in this paragraph, it would not be immoral for a corporation to engage in pyramid scams if they were not actually illegal. In point of fact, you seem to be arguing that a pyramid scam would not – in fact – be a scam if it were legal. That seems like a rather odd definition of ‘scam.’
In principle, no micromanagement is needed: Simply tax total corporate revenue, rather than profits alone. While this is usually undesirable for a number of entirely valid reasons, it is certainly possible that transfer pricing and similar scams will force countries to engage in this practice. That would be sad – as noted there are valid reasons why this is a bad idea – but if the alternative is that corporations are effectively untaxable, then it will be the only reasonable way to go.
Here you are essentially saying that a consequence of the free movement of capital combined with national sovereignty over taxation systems will likely lead to some countries using an essentially parasitic taxation scheme.
Lowering taxes in order to take advantage of corporate transfer pricing is functionally, but not necessarily legally, equivalent to a company not paying sales tax on a service and splitting the savings with the customer: Both partners in the scam get richer and tax burden is relocated to those who don’t engage in such scams. We consider one immoral and illegal, and accept that countries protect their taxation systems from it, but consider the other perfectly legal and moral, and demand that countries not protect themselves from it. You have to admit that this asymmetry is rather striking.
But not significantly so. I have not run the numbers, but it is quite evident from simply looking at the graph that it is well within the ‘noise’ range of the median.
Finally, I will note that I am not hostile to corporations in general. Only to those rules and regulations that permit corporations to scam people, trash the environment, evade taxes by moving around funny-money, etc. As you yourself note, they are almost obliged to do so as long as we permit it. I simply argue that we should stop permitting it.
– JS
When a curve is plotted on the same graph as a set of data points, it is unversally understood to mean that the curve is related to the data points. So there are only two legitimate possibilities:
1) The curve is derived from a theoretical analysis and the intent of the article is to point out the data does not fit the curve and the Laffer curve theory is therefore full of crap. The article says nothing of the sort. Instead, it says, “The Laffer Curve analysis indicates that these corporate tax increases are likely to raise little if any additional revenue, because companies will have a new incentive to move even more of their operations out of the reach of the IRS.” So the author clearly expects the reader to conclude from the graph that the Laffer Curve analysis is valid.
2) The curve is fitted to the data and intended to summarize it. The strategy of overlaying a line or curve on uncorrelated data to decieve the eye into seeing a nonexistent trend is a classic example of dishonest graphing.
as Ray S points out,
For example, profits are generally calculated after you’ve subtracted operating expenses including payroll.
A business which knows it is going to have to hand over all profits to the government would probably be unusually generous in paying out benefits to its associates, as the primary motive for not doing so has been eliminated.
There would therefore be a highly motivated workforce working their butts off to make the company a success and maximize revenues, and voting themselves terrific payraises just before the revenooers showed up. Suddenly the idea of 100% corporate taxes doesn’t sound too bad! ;)
It is also worth noting that the data points aren’t entirely independent. Ih these days of globalisation corporations have ways of moving money around in order to avoid taxation. A country reducing its corporation tax rates will put pressure on others to do the same.
A little examination of how efficient governments are in collecting taxes – and preventing the huge volumes of money siphoned off to ‘tax havens’ (a massive piece of missing data from this graph) – would surely be pertinent?
Although it wouldn’t suit the purposes of the guys who wrote the article of course…
I’m neither a statistician nor an economist, but there is one problem I have with this whole issue – the comparison of different countries.
I know it is dangerous to make generalisations about societies, but since the whole Laffer Curve theory is about a society’s preparedness to pay taxes at increasing rates is there not a socio/psychological dimension to all this?
Take two examples: The US and a Scandanavian country.
The US might be classed as a more “tax averse” society on the whole, disinclined to pay higher taxes for social programmes. On the other hand, Scandanavian populations might be considered more egalitarian in nature and hence more inclined to pay more for social services etc.
Once again, I recognise that making generalisations is dangerous. But if this is the case then wouldn’t each society need its own curve? The most efficient tax rate for Americans (before they start resisting and therefore not paying) might well be different to that of Scandanavians? And if this is the case then comparing Society One with Society Two could be considered a pointless exercise except to determine a broad global average of something.