Taxes and Private Sector Investment – Evidence from the Real World
by Mike Kimel
Cross-posted on the Presimetrics blog.
Taxes and Private Sector Investment – Evidence from the Real World
Last week I had a post (which appeared both here and at Angry Bear). The post included the following graph:
The graph looks at every eight year period since 1929 (the first year for which National Accounts data is available from the Bureau of Economic Analysis) that can be thought of as a complete “administration.” It notes that there is a very strong negative correlation between the tax burden in the first two years of an administration and the economic growth that follows in the remaining six years of the administration. In plain English – the more the tax burden was reduced during the first two years of an administration, the slower the economic growth in the following six years. Conversely, the more the tax burden was raised during the first two years of each administration, the faster economic growth was during the following six years.
At this point I note… this is not my opinion, it is what the data shows. And there is no cherry picking – I went back as far as there was data and included every eight year stretch for which a single President occupied the Oval Office or in which a VP took over from a President in the middle of a term. And these real world results contradict just about everything that standard economic theory (Classical, Austrian, you name it) tells you.
So I tried providing an explanation:
Michael Kanell and I advanced several theories in
Presimetrics but the one I think makes the most sense is that changes in the tax burden are a sign of the degree to which an administration enforces laws and regulations.
The logic is simple – (1) collectively, Americans cheat on their taxes and (2) whether the tax burden, the percentage of GDP that the government collects in taxes, rises or falls seems to have nothing whatsoever to do with whether marginal income tax rates rise or fall. Thus, one way for tax burdens to go up is increased enforcement, and one way for tax burdens to fall is decreased enforcement.
Now, to me that’s self-evident. But I’m starting to realize not everyone sees it this way, so let’s run a simple test. If a regime tolerates corruption or encourages companies to game the system rather than to be productive, we should expect growth in the private sector to be minimal at best. All else being equal, we should expect faster growth in the private sector the less rot there is in the system. I assume this is not remotely controversial. And it implies that if tax collections are indeed an indicator of an administration’s intolerance for shenanigans, then growing tax burdens should be followed by rapidly increasing private sector activity and falling tax burdens should be followed by relatively slow growth in private sector activity.
Crazy, right? Lower taxes leading to less private sector activity! Insanity! It defies economic theory. And common sense. But how does it fit with what happened in the real world? Extremely well, actually.
The graph below shows the change in the tax burden in the first two years of each 8 year administration on the horizontal axis, and the annualized change in real private investment per capita in the remaining six years along the vertical axis.
Notice… administrations that cut the tax burden early saw mediocre increases in private investment later. On the other hand, administrations that started out by increasing the tax burden enjoyed big increases in private investment in the remainder of their term. This is yet another instance where real world results contradict just about everything that standard economic theory teaches, particularly the Chicago School, Austrian, and Libertarian variety. And sadly, that theory has so permeated our collective thought processes that it has come to be referred to as “common sense.” Just as it was common sense at one point that the earth was flat, and the center of the universe.
It’s worth pointing out, by the way, that the relationship between the tax burden and real private consumption is similar; administrations that raised the tax burden saw greater increases in real private consumption per capita than administrations that reduced the tax burden. The relationship, albeit a strong one, is slightly weaker than the relationship between tax burdens and investment. By contrast, the relationship between changes in the tax burden in years 1 and 2 and changes in real Federal Government spending per capita are much, much weaker.
So let me revisit once more the explanation that Michael Kanell and I put forward in
If you have a better explanation, let me know.
—
Data sources and comments.
The definition of the tax burden used in this post is Federal government current receipts from line 1 of NIPA Table 3.2divided by GDP from NIPA Table 1.1.5, line 1. Real economic growth was measured as the change in real GDP per capita, which was obtained from NIPA Table 7.1, line 10. Population came from the last row in the same table.
Real private investment came from line 7 of NIPA Table 1.1.6.
As always, the change in any series over the length of an administration is measured from the year before the administration took office (the “baseline” from which it starts) to its last year in office.
—
I intend to look at the relationships described in this post in a bit more detail going forward. However, expect the next post to cover another issue which seems to come up a lot – whether the results I’ve been posting are statistically valid or not.
Note also… if it’s not obvious, this post deals with the tax burden, the share of GDP going to the Federal government, and not marginal tax rates. Please do not insist on commenting on a topic unrelated to this post.
Mike
sounds like the lottery effect. people know a few percent raise in their wages won’t change their lives, so they spend a few percent of their wages on the very remote chance of winning the lottery.
looks like they vote for the tax cutters because they know the rise in GDP that would come with a raise in taxes will not change their lives, but they see the winners of the “free enterprise lottery” and imagine themselves rich. and that’s almost as much fun as actually being rich.
actually it’s more fun. it’s less work, and most of the fun of actually being rich is in the imagination anyway.
The choice to vote for tax cutters is not hard to understand. We live and work through a limited number of business cycles, and the influences on those cycles are fairly numerous. Identifying the actual effect of tax changes on economic performance takes a data exercise. There is a lot of ego involved in claims for “common sense” (naive) understanding of the economy, but ego ends up a victim to selective attention.
So thanks for the data exercise.
I guess I should have gotten around to the point I meant to make. Voting for tax cutters makes sense because it is unlikely we will be able to know the real impact of the tax cutter’s agenda. A promise of a tax cut looks like free money, even though the actual cost of the cut can be high.
Mike, I certainly think the rule of law is a critical factor in growth, but I continue to belive that something more is going on and I will continue to argue that tax burden increases lead to less risk taking–the payoff is not as great; more savings–particularly if there is a tax advantage in deferring consumption such as 401K’s or leaving profits in a C corporation; and generally a more future oriented view of economic activity–deferred gratification, if you will. All of these factors, if not overdone, should be more likely to produce sustained growth over time. Of course, the “common sense” folks always argue that any taxes are overdone leading to less innovation, underconsumption and an unwillingness to “seize the moment”. Even if it is difficult to explain the why, I appreciate the effort in setting the record straight.
The primary factor is probably the underlying economy.
There are more recessions during Republican administrations (greater likelihood of lower rates), so the underlying poor economy results in reductions in private investment, regardless of tax rates.
Conversely, there is greater economic growth during Democratic administrations, so the higher personal income results in higher private investments (they’ve gotta do something with the extra money), regardless of tax rates.
The only tax rate related effect is probably related to when income is recognized. Investment income is only recognized when the investment is disposed, so a higher tax rate may induce people to remain invested rather than selling — the cumulative effect is greater investment.
Except for extreme situations (tax rates well over 50%), tax cuts merely shift the tax burden/government spending benefits between different individuals but don’t provide much benefit to the general economy (i.e., the size of people’s pie slices change but the pie itself stays the same size). There may even be a negative impact because the obsession about personal tax burdens reduces the willingness to subordinate personal interests for the common good.
kharris
i think its even more obvious than that. a person can see a tax cut. he can’t see how the taxes lead to an increase in the economy and a net increase in his own income. of course any increase in his own income is due to his own “hard work and enterprise” so you can see the taxes just punish hard work and enterprise.
i hope its obvious that the last sentence is “in the mind of the beholder.”
which of course is what you were saying, but it doesn’t hurt to put it another way.
terry
i can’t believe that a marginal increase in taxes leads to less “risk taking.” of course i have never been a high end investor, so maybe i don’t understand the psychology. but as a low end person i never notice taxes one way or the other except when i am actually filling out the damn 1040, and when i see the schools building another parking lot.
i am fairly sure the arguments for lower taxes are all based on self serving fairy tales. please note that is the “arguments” not the instinct, which we all have against paying taxes.
H Bob
I think there is another way “the underlying economy” affects the “tax burden” (percent of GDP going to taxes) but I am hoping for Mike to confirm that the arithmetic makes sense before I reveal it to an amazed world. (joke. it’s actually pretty obvious.)
Ah, but your formulation takes in the spin offered by politicians, and makes it part of the voter decision process. It may be, but Ocham doesn’t need it to be. The difficulty in discerning the result is enough for Ocham.
I think there is a simpler explanation that can be seen if you look at china.
Over in china they have a 50% savings rate. This is because they know the government cannot provide for them.
Likewise tax cuts are correlated with similar feelings and americans save more.
More savings less growth.
oh for sure.
the political “spin” offered by the right amounts to “what you always thought, is true.”
terra
interesting theory. and you might agree with keynes, at least some of the time. but definitely not with the conventional wisdom which is that lack of savings kills growth. of course the only time the worry about lack of savings is when they can call Social Security “not savings.”
on the other hand, china’s growth rate is considerably higher than ours. so i tend to think their savings rate is first because they can’t afford to buy what they produce, and they still live in fear of hard times.
Coberly, In my business I see people engaging in extremely risky, very short term decisionmaking all the time. It is my understanding that much of the financial crises was brought on by this type of behavior. Even the people who should have known better went along because otherwise they could not match the returns apparantly being earned by the gamblers. To the extent that the tax burden is increased on that profit–even marginally–other incentives rise like reducing risk, planning for the long term, being moral, obeying the law etc. sure it is only a marginal change, but that is all we talk about in economics anyway.
This post and the comments that follow seem to suggest that tax cuts, when they occur, do so on an equally distributed manner and have an equal impact of the various percentile ranks of income earners. That’s certainly not true and that probablt accounts for the fairy tale thinking of the middle class conservative that his taxes are going down under conservative administrations.
Another assumption that keeps reoccuring is that if taxes are cut the money saved will go to reinvestment. If one means reinvestment in a new ski lodge or Porsche they may be right, but it seems dubioous to suggest that investment behavior is based upon the tax regulations in place at the beginning of an investment plan. Given the likelihood of changes in the tax codes this would certainly be poor business planning. Investments occur because the investor sees a chance for asset appreciation and possible income. Taxes are going to be apploed one way or another. Business investors are simply always looking for tax cuts and tax advantages, not because of the potential effect on the investment, but because of the real effect on all income formerly taxed. Lowering taxes is a constant goal of all income earners and the very welathiest earners have the greatest likelihood of brokering (otherwise called lobbying) for tax codes to their personal best advantage. Guess where that leaves the rest of the population? Someone has to pay the bills. If not those that have much, then it will be those that have less.
You’ve hit it exactly, coberly. While their middle class is growing it still doesn’t really represent the overall population. Also, with the one child policy it’s unlikely that they can count on help from their children in old age and there is basically no social safety net.
coberly,
In this instance, I disagree… people genuinely do believe that cutting taxes leads to faster economic growth. Even most Democrats believe that. They are merely willing to trade off what they feel will be slower growth in exchange for other social goals.
But the fact that a belief is erroneous doesn’t mean people don’t act on it.
terry,
I’m going to have to give some thought to what yo wrote. I agree that at some point higher tax burdens will hinder growth. I’m just not sure we’ve seen anything at that range since 1929.
Rich people spend a lower percentage of their income than poor people. The government spends every penny it can get its hands on. If you raise taxes, particularly on high income people, it reduces their income which has a small negative impact on the economy, but it increases government income which goes right back into the economy as wasteful government spending. I think they called this a higher velocity of money back in the 1930s.
As a bonus, a lot of the people the government gives money to aren’t extremely rich, though all too many are, so they spend a lot of their income. That means there is a higher multiplier. If they saved the money, like richer people would, the economy wouldn’t benefit. Even more important is that all these government beneficiaries, both direct and indirect, buy stuff from the rich people which encourages investment and leads to an even brisker economy.
You could look this up in a history book, probably on the page after the sun rising in the east. Even the Romans understood this. The class warriors, generallly from the senatorial class which resented the emperors, bitched about “bread and circuses”, but they did quite well on the expanding empire.
terry
i yield to your superior knowledge of the players. the question that i would have, is how many of these people are there, and are they a significant factor in the economy. My guess is that gambling with stocks can crash the economy but doesn’t really have much to do with investing in it.
Mike
i think we might both be right. The folks could like the idea of a tax cut without understanding that it will not change their lives. But I think they like even more the fantasy that a low tax economy will allow them to get rich some day.
the dems, at least, ought to know better. at least i am pretty sure that what happens is that under a dem administration the poor do better. that means they have more income. that means they pay more taxes. and this would increase the “tax burden” without any increase in tax rates.
Lets say the bottom half of the population makes 100 “dollars” and pays no tax. the top half of the population makes 200 dollars and pays 20% or 40 dollars. tax burden is 20/300 or 7%.
now increase the income of the poor to 150 dollars and let them pay 10% on the extra 50, for 5 dollars. And the income of the rich goes up to 300 dollars, and they pay the same 20%, for 60 dollars. now the tax burden is 65/450, or 14%.
[note that just looking at the poor, you go from a tax burden of 0% to 5/150 or 3%. assuming that the rich get richer at even the same rate as the poor just makes the effect bigger.]
“…… but it increases government income which goes right back into the economy as wasteful government spending.” Kaleberg
Might you mean the wasteful spending on public education? Or did you have in mind the repair of the streets and other form of infrastructure? As I keep trying to point out to those of you who think like Kaleberg seems to think, things do need to get done and those things do need to be paid for. California is laying off teachers, as are many other states and localities. Mr. Bloomberg seems to have given up on providing any reasonable semblance of road work in NYC, All in the name of allowing those that earn most of the income to keep it for those all important expenditures. I wonder if their T-Bills are just so many IOUs. Certainly their hedge fund accounts seem to be doing quite well, but that money isn’t being invested in an entrepreneurial manner. More like a Las Vegas kind of thing. And they are doing well. Another Porsche, a pied-a-terre in Manhattan, etc. But what about the cost of maintaining basic services? Libraries, public education, roads, transportation to work, are they all just a big waste of money? Exactly what did you think was the point of an organized society? The 99% work their asses off to provide a better quality of life for the One Percenters?
Mike,
Don’t knock yourself out. Some times the proof is little more than discovering fire, all over again. Why bother? The masses are asses, as the lady once said, and that is the operative factor. The facts no longer matter, even here at AB. We have the ice caps melting at both ends yet CoRev and his ilk are certain it’s all bullshit. Social Security is well funded and a long term public program success, yet the Deficit Commission whores are fast tracking its demise and the press isn’t paying it any mind. Localities can’t provide public education nor fix the public roads, yet Ahhnold answers the problem by laying off and furloughing teachers and other state workers. Bloomberg would like to do the same, but NYers are more ornery about such crap.
The point is that…….. Oh, what’s the point? Listen to Kagan’s confirmation hearing for ten minutes, if you can stand it. One Senator after another sounding like complete egotists without a clue as to what they are supposed to be dooing. It’s a wonder we’re not in worse shape than we are. This is representative government? What’s the old saw about the lowest common denominator?
Mike Kimel,
Cutting taxes doesn´t lead to increased economic growth… People just don´t understand the positive effect that the government has on the economy… It´s all about MSB=MSC… check it out
http://en.wikipedia.org/wiki/Externality#External_benefits
You say: “Michael Kanell and I advanced several theories in Presimetrics but the one I think makes the most sense is that changes in the tax burden are a sign of the degree to which an administration enforces laws and regulations.”
Why look for a sign when there is actual data on the degree to which an administration enforces laws and regulations?
Granted, the (readily available) data may not go all the way back to 1929. But it’s not hard to get IRS operating budgets (which could be a reasonable proxy for enforcement effort) and even the number of IRS personnel going back a few decades.
Rather specific data going right at the heart of the matter (i.e. audit frequency) is available at least back to 1988. That gives you about 20 years of data to look at over three presidents (2 Rep. and 1 Dem.) Admittedly this is not a lot of data, but like I said one could find other proxies that might be almost as good that would go back further. I would check out such data before advancing such a theory.
But if you want to give this the back-of-the-envelope test, here’s a link to the frequency of audits of large corporations (assets of $250 million or more). I trust you would agree that audits of the nation’s largest corporations might be the single most important indicator to the movers and shakers of the business world of just what kind of shenanigans the administration will tolerate.
http://trac.syr.edu/tracirs/highlights/current/auditrate.html
They’ve got more data on the IRS too, in case you’re interested.
It looks pretty tough to tease out any significant effect on GDP growth. I tried. But it wasn’t a total waste of time. A glance at the data reveals what president of the last three (prior to Obama) presided over the largest decline in the frequency of audits for the largest corporations.
His last name wasn’t Bush.
But I won’t let George W. Bush off the hook completely. Both he and Clinton reduced the size of the IRS workforce, which one could interpret as signaling less importance on enforcing the laws. In the last 30 years, Reagan (!) presided over the largest increase in IRS personnel. Of course, technology made tax collection less labor intensive since 1990 too, so it is far from clear how much importance to put on the size of the IRS workforce. But in case you’re interested…
http://www.irs.gov/taxstats/article/0,,id=207706,00.html
In one sense, all of what I just said is more proof of absolutely nothing. There really, honestly, doesn’t appear to be any strong systematic relationship between these data and economic growth (contemporaneous or lagged), nor would I have expected there to be a priori.
Personally, I would advocate a stepping up of IRS enforcement. I think it would send a good signal, and potentially bring in a little revenue. But those personal feelings aside, your “rule of law” explanation with respect to growth doesn’t hold water. At least not for the last three presidents (20 years). And I’m not confident that going back further will help much.
Indeed. In fact, isn’t it widely acknowledged that the calls to balance the budget during the Great Depression were a mistake? The dramatic growth in the latter years of FDR are a reflection of how low things were when he took over.
Now, if you take FDR out because that is an obvious outlier for so many reasons, the correlation is much weaker. And in fact, several of the remaining points have very little change in taxes at all.
Ask me this: Could a small increase in taxes coincide with a large increase in economic activity? Sure! Why not? There are so many other things going on that could push growth up that would swamp any effect from a small increase in taxes. Does that mean that you could run through a big tax increase and expect an even bigger increase in growth?
Sadly, no.
It’s like the minimum wage. Suppose the minimum wage goes up a quarter. Will unemployment rise or fall? While Econ 101 says unemployment should rise, ceteris paribus, I would not necessarily predict it in this given case. Ceteris is not paribus. And a quarter is not much of an increase. So the other factors would probably have more of an effect on unemployment than the tiny increase in the minimum wage. Unemployment might fall–confounding those who only heard the simple-minded Econ 101 explanation.
Does that mean that you could increase the minimum wage by $5 and get an even bigger drop in unemployment?
Sadly, no.
That pretty well sums up my thoughts on the subject.
Edward Lambert,
The graph has a little box showing the correlation for all administrations and for the four single President eight year administrations since 1952 (Ike, Reagan, Clinton, and GW). That leaves out some outliers plus JFK.LBJ. You may have noticed that in the first graph, the correlations are about the same with the bigger and smaller groups. In the second graph the correlations are actually much stronger (its just about linear) for the second group.
What I think you are noticing is that the line becomes a bit less steep without FDR… but the sign of the line is unchanged. (I originally was planning to put up four graphs in this post – the other two being without the smaller group only – but decided that it was overkill and a shorter post communicated more. But I did the graphs… and I figured the box with the correlations could communicate the same message while taking up less stpace.)
William Polley,
FDR is not the outlier you think… as noted above, he changes the slope, but not the sign.
I agree with you that raising tax burdens is not the reason for faster growth. I started out a long time ago by trying to test for whether cutting tax burdens had a positive effect on growth. I ended up finding that cutting tax burdens is correlated with slower growth. Now, cutting tax burdens cannot in and of itself be a cause of faster growth, so it was necessary to figure what it was that rising and falling tax burdens were signs of.
Now… if tax burdens get “too high” they will discourage growth. But it seems clear to me that they haven’t been in that range for a while. Having worked for one of the Big Accounting Firms, I now believe a few things as a result:
1. At the time I wasn’t the kind of person who could function well in a Big Accounting firm.
2. The purpose of a Big Accounting firm is to ensure that clients pay much less in taxes than they would if they went with anyone other than a Big Accounting firm. Even if you haven’t worked for a Big Accounting firm you have to recognize that the clients of Big Accounting firms are paying a very premium price for a reason, and there isn’t much evidence that these clients are stupider than the average person
Grumble.. I just lost a big response. Let me try again quickly.
1. Michael Kanell and I discuss that with respect to prosecution of public officials in the book.
2. The specific issue of IRS audits and how much or little they indicate “enforcement” was covered by a post I wrote a while back (can’t find it now) and I believe a couple of posts by the since deceased OldVet, a former auditor for the IRS.
The problem is this… some admins increased audits but had smaller settlements. Some had fewer audits but increased settlements. Which constitutes more enforcement? Now, add one more wrinkle… if the economy is adding companies quickly, increasing audits is not necessarilly a sign of more enforcement unless you keep up with firm growth. Similarly, if you increase the size of settlements, it could be because firms are cheating more precisely because they know that few get caught so why not get away with it? (Think of it this way – if the costs of a speeding ticket fall to something trivial, cops may find themselves catching a lot more speeders.) Also, settlements may increase just because firms are more profitable, all else being equal.
Thus… in the book we discuss prosecutions and convictions of public officials, but we point out that we aren’t prepared to use them as a measure of public corruption or enforcement. Similarly, here I am not prepared to use either audits or settlements as signs of increased enforcement because the same outcome has several very good (and contradictory) explanations and I have not been able to come up a way to determine which one is valid.
Edward Lambert,
Both graphs show the correlation for the full sample and for the four administrations made up of one single President serving eight years after 1952 (Ike, Reagan, Clinton, and GW). That’s a similar group to your no outlier group. Notice that in the first grpah, the correlations are about the same for both groups (big and small), and the correlations are much stronger for the smaller no outlier group in the second group.
Leaving out the outliers can reduce the slope in the second graph, but doesn’t change the sign. (BTW… the original version of this post when I wrote it had all four graphs, but I decided the change was small enough that it wasn’t worth including them all and showing the correlations got across the same info.)
Bill Polley,
I agree that at some level, increasing tax burdens will have a negative effect. What I think the evidence seems to show is that we aren’t at that point and probably haven’t been between 1929 and the present.
Hi. I am traveling right now and for some reason haven’t been able to post. So Dan is kind enough to post a few responses for me…
1. Edward Lambert. Both graphs show the correlations for the full sample and for a smaller sample that includes only the four Presidents that served full eight year terms since 1952. That second sample is analogous to your no outlier sample plus JFK/LBJ. Leaving out the outliers changes slopes but not signs. In the first graph, the correlations are about the same for both samples, in the second the correlations are stronger for the smaller group. (Note… my first draft of this post included graphs of the smaller sample too but I decided the added info wasn’t worth the added graphs and a smaller post was better.)
2. William Polley. A couple things. First, I agree that at some point, higher tax burdens hurt growth. I think the evidence shows we aren’t there and haven’t been there at least since 1929. Second… IRS audit figures… Note that I had a post on this a long time ago, and I think that OldVet (since deceased) also had a post on this subject (or perhaps it was just something he left in comments). It is also analogous to a discussion Michael Kanell have in the book where we discuss figures for prosecutions and convictions of public figures and why it doesn’t work as a measure of corruption or even enforcement.
Consider… the number of corporate audits may increase and still not keep up with incresing number of companies. Also, as I recall, for most administrations, there is a trade-off between the number of audits and the size of settlements.
Settlements are also not a good measure… after all, settlements can increase because of a growing economy. Or they can (perversely) increase because its enforcement falls which reduces the expected cost of cheating. You could have more auditing and smaller penalties or less auditing and bigger penalties. Which is less corruption and more real enforcement?
BTW… I worked for a year at a Big Accounting firm. The one thing I learned… people pay the big buks that the Big Accounting firms charge in the expectation that it will lead to lower tax burdens.
jack
do not despair. i despair enough in one day for both of us. and i agree with your analysis of the public picture. but i think we are not allowed to give up. or if we do we never get to find out if we would have won after all.
jack
i think kaleberg was speaking ironically about “wasteful govrnment spending.” the rest of his comment seems to agree with you and me.
William Polley
yes, but why spoil a perfectly good and useful insight by a reductio ad absurdam type argument. Kimel shows that small tax increases do not sink the economy as some politicians keep telling us, and nearly all of them believe. no one is proposing huge increases in taxes.
i would also object to an excessive fascination with “statistics.” again, Mike is not predicting from a sample to a population. he is just describing what happened. and i’ll take the contrast between FDR and W as a pretty good guide to policy. especially since the rest of the data hardly seems to contradict it.
Admittedly I am not noted for my optimism. I am a “student” of human behavior, at one time professionally, but more so generally. Read Linda’s double post reviewing the two recent CBPP reports. Listen to the Senate hearings regarding Kagan’s nomination. Reflect on the fact that Kagan is the best that we can hope for. Reflect on the asinine attitudes of the Senators doing the interviewing. Reflect on the ever decreasing quality of major party nominations to significant congressional positions. Does the name Sharron Angle ring any bells? “Second amendment remedies,” and she’s not immediately arrested on charges of sedition? Remember what the response to Eldridge Cleaver’s and Bobby Seal’s similar similar remarks had been. Imagine if even Chomsky or Nader had made such a suggestion. And then there’s Bobby Jindal and Jeff Sessions. Maybe Angle is actually an improvement. She needs to move to the deeper south. Nevada’s probably not crazy enough for her.
So tell me Coberly, what exactly is there to feel optimistic about? Maybe Angle has the right idea. I do admire the effectiveness of M. Robespierre in his time, though he and his cronies did get a bit carried away, literally as well as figuratively.
Sorry Kaleberg if Coberly is right. I’m a bit testy lately and may have glossed over the subtlety of your comment.
Mike, I have yet to hear anyone mention the impacts of annual tax law changes, and the shifts in their targets. Bottom line, the IRS has an immense advantage as the giant in the adversarial tax war. Accordingly, confusion caused by the annual IRS/Tax law changes and their magnitude and targets, and the defensive position required by all levels of tax payers due to the strength of IRS advantage(s) may actually have a bigger impact than enforcement. A better proxy might be the growth of the tax preparation/tax accounting businesses.
Mike, I would say that both large firms and the small tax payer may be motivated by the same things. I would not make a claim that big firms paying for tax help is much different than the average citizen paying for the same help. that should should also include thepurchase of tax software. Most of these efforts for big and small tax payers help reduce tax burden, AND REDUCE RISK OF AUDITS.
Like the militarism, and similar corporate welfare.
jack
as i said, i have enough despair for both of us. i didn’t say you needed to feel optimistic. i said you needed to hang in there.
CoRev
I don’t know what effect it has on GDP, but I agree with your assessment of ‘tax burden’ as having more to do with figuring out what the damn law requires than actually paying the tax.
With my miniscule income I had to hire an accountant to figure out what i was supposed to pay. and i ended up paying him exactly what he “saved” me from paying in taxes. i was going to write IRS and suggest we make a deal. but i have heard they have no sense of humor.
Coberly – your math is off.
“Lets say the bottom half of the population makes 100 “dollars” and pays no tax. the top half of the population makes 200 dollars and pays 20% or 40 dollars. tax burden is 20/300 or 7%.
Now increase the income of the poor to 150 dollars and let them pay 10% on the extra 50, for 5 dollars. And the income of the rich goes up to 300 dollars, and they pay the same 20%, for 60 dollars. now the tax burden is 65/450, or 14%.”
No, the tax burden goes from 40/300 (13.3%) to 65/450 (14.4%). Doesn’t invalidate your argument about rising incomes raising tax burdens, however. I’m just being picky today.
You say: “Thus… in the book we discuss prosecutions and convictions of public officials, but we point out that we aren’t prepared to use them as a measure of public corruption or enforcement. Similarly, here I am not prepared to use either audits or settlements as signs of increased enforcement because the same outcome has several very good (and contradictory) explanations and I have not been able to come up a way to determine which one is valid.”
First, on audits vs. settlements vs. convictions as the best measure of a president’s commitment to the “rule of law,” that is an open question that I chose not to addresss in the comment due to the fact that it was already getting long. The data on all of them is available. I will say that a president probably has the least control over convictions since that’s also a function of the court system. Audits vs. settlements? As you say, sometimes it goes one way, sometimes the other. But that just compounds the problem of trying to link either of them to growth. You’re just not going to find any statistically or economically significant link any way you look at it–your words here suggest that you understand that.
I just find it extremely puzzling that you would say that you looked at this but aren’t prepared to use them as a measure of corruption or enforcement and then turn right around and posit that the reason tax burdens are related to growth is because it indicates the level of corruption– without any evidence other than the somewhat contradictiory evidence that you dismiss.
On the contrary, I am pointing out that (as you apparently know) there is a variety of data out there on enforcement and it doesn’t consistently show any significant link to economic growth. Yet (for reasons I don’t quite understand) you want to link tax burden to growth through this enforcement connection (tolerance for corruption).
You’re saying that presidents who are tough on enforcement will see an increase in the measured tax burden, and the resulting increase in growth is actually due to less tolerance for corruption (an indirect link). Except that direct evidence of a link between enforcement (audits, settlements, convictions, etc.) and economic growth is anything but solid. That doesn’t necessarily mean that it’s not there. There might be a connection. But the effects might be so small (economically) so as to be swamped in the sea of other changes.
The only conclusion I can draw is that the “tolerance for corruption” argument is not the reason that you’re observing a relationship between tax burden and growth. You’re trying to get it in the back door, but it still doesn’t fit.
Again, I say this as someone who would like to see more auditing and enforcement. I think it would be good for a variety of reasons. But this isn’t how I would go about making my argument.
It has always seemed obvious to me that tax increases spur people to earn more, by working longer hours, increasing their productivity so as to gain promotions, starting their own businesses, getting other family members to enter the workforce, etc, thus compensating for income lost by the tax increase..
It has also always seemed obvious to me that Republican economic ideology is credible only to those who have shit for brains.
I was being ironic, but I understand just how tiring the usually automatic responses are with parroted phrases like “wasteful government spending”, “repressive taxation”, “job killing regulation” and so on. Sometimes I just want to scream. In truth, though, I am an old fashioned New Dealer. If find that their economic theory made perfect sense, and it worked in practice. I’ve used it to guide my investments, and it has paid off comfortably.
Minor quibble: Kimel shows that small tax increases do not necessarily reduce growth. Those who have been skeptical of his project all along (including myself) have consistently said that these two-variable relationships leave out a lot of things. Truth be told, FDR and W are going to be diametrically opposed on many dimensions. They also saw different growth rates. Now, are any of those two variable relationships necessarily useful indictators in isolation? No. Do I think it is enlightening to list them one by one as if they are? No.
Is he literally predicting from a sample to a population? Not literally, no. He is, however, presenting these individual vignettes in a series, that gives the appearance that he is building up an argument that Democratic policies were better. If this mountain of evidence is solid, then he or his readers could justify saying that therefore we should adopt more Democratic policies. Economists (or anyone else for that matter) can rarely ever “just describe.” Mike is engaging us in a conversation. He is trying to persuade. Nothing wrong with that–that’s what all intellectual discourse is about. It’s why I have engaged him on it and presumably why he responds. If it was just about describing the past, why would either of us bother?
Yes, there is a big difference between FDR and W. There are many reasons for this–some within and some beyond the scope of Mike’s project. And you are entirely correct that the rest of the data “hardly seems to contradict” the differences we see between FDR and W. However, the rest of data also does not necessarily confirm all of those hypotheses either, or sometimes the confirmation is weak. You may call that an excessive fascination with statistics. But Mike and I have been trained in statistics. It is our language. Mike knows way more statistics than he is using in this project. I’ve said all along that maybe it would be a good idea to use just a bit more, because when you do the results are not always as clear. But more econometrics might bore the reader, and he has the right to write the book the way he wants.
Mike admits that some of his thoughts on corruption come from his experience in a big accounting firm. Fair enough. Some of my thoughts on how an argument like this can be abused comes from my experience as a professor where I see students and others (who should know better) extrapolate barely significant linear relationships way too far. We all have our professional sensitivities.
“FDR is not the outlier you think… as noted above, he changes the slope, but not the sign.”
I don’t think an outlier has to change the sign of a regression to be considered an outlier.
“I agree with you that raising tax burdens is not the reason for faster growth.”
Ok.
“…so it was necessary to figure what it was that rising and falling tax burdens were signs of.”
And that’s why we do multi-variate regression, build models, etc.
“Now… if tax burdens get ‘too high’ they will discourage growth. But it seems clear to me that they haven’t been in that range for a while.”
Given your findings as well as other data and theory, it seems quite reasonable to me that expectations of future taxes, the implementation of the tax changes, and other factors matter as much or more than the level of the tax rates (or burdens) in determining short-term (i.e. business cycle frequency) fluctuations in GDP. The trend growth rate has been remarkably stable through a number of different tax regimes.
So I would modify your statement to say that if tax burdens are subject to more uncertainty or rise too quickly (not just “too high”), that would be bad for growth.
But your analysis doesn’t tell me anything useful about that. You say you agree that higher taxes don’t cause faster growth, but what you present is a univariate regression that shows a correlation and some just-so stories to try to explain it.
In fact it seems that omitted variables and random shocks are just as convincing a story to explain your charts… and that the real interesting questions are not even addressed by your charts at all because they can’t be boiled down to a univariate growth regression. That’s too bad.
William Polley,
Actually, somewhere along the way I tried to figure out the difference in policy. While the book focuses on the parties, the parties are shorthand for policy which on economic issues tend to fall down party lines. But not completely… because the policy that is under the President’s control (in the book we make a distinction between “earned” growth and “unearned” growth) I have found seems to have more explanatory power than any other is tax policy. And there, Truman, and until now, Obama, do not follow typical Democratic policies… they resemble GW more than they resemble FDR.
Now, you may feel I’ve pushed a simple relationship too far. My response is simple. At one point I was told that no way would admins that increased the tax burden produce faster growth. Then I was told that the Fed or some other outside agency had to be responsible. Then I was told to check lags. And so forth. AB readers may skew left of center, but they range the gamut, and most of these posts have grown out of figuring out how to test one or another objection that was raised by folks who were not in agreement.
But every single one of the objections that were going to knock my simple story flat have themselves fallen flat. I realize that most simple ideas are wrong, particularly if they contradict common sense. But what happens when one of these stupid, simple ideas is still standing after every objection that has been raised has also been tested?
As to why I’m using simple tools – I’ve stated this before. We’re dealing with a politicized issue. And on a politicized issue sometimes people push a bit too hard. Now, the tools you and I use elsewhere can be used to prove many things. But these tools are based on assumptions which can drive the answers. And if we’re trying to guess the likelihood that a missile will jink left, or that some species of frog is more likely to survive certain conditions than another frog, it is probable that we would agree on the starting assumptions, and reach about the same outcomes regardless of the fact that you and I would probably choose slightly different tools.
But… if we’re trying to determine the effect of tax changes, our prior beliefs might affect our choices enough to lead to radically different outcomes. Even the really brilliant guys – certainly better at this than I, often disagree with each other about the basic assumptions, and while some might be outright dishonest, I suspect most of them on each side are making their choices as honestly as their biases allow.
So in the end, they end up arguing esoteric issues, and underneath those esoteric issues are simply basic assumptions. But nobody who doesn’t have a certain set of tools in their toolbox can even understand the first part of the argument that is going on.
I figured I’d try to do something basic to get what basic assumptions should be. And I think, if nothing else, I’ve managed to show one thing – at the ranges we’ve seen since 1929, for a culture like that of the American society we’ve seen since 1929, if tax cuts would otherwise have a beneficial effect, that effect is completely, overshelmingly and absolutely swamped by other things that push in the exact opposite direction of the beneficial effect that is expected.
“if tax cuts would otherwise have a beneficial effect, that effect is completely, overshelmingly and absolutely swamped by other things that push in the exact opposite direction of the beneficial effect that is expected.”
I have not read your book. But show me where you say this in these words (or something simlarly clear and direct) in your book and I’ll cut you some slack. Note: I’m not looking for hypotheses without evidence as those “other things.” I’ve been waiting for that statement that the effect is “completely, overwhelmingly, and absolutely swamped by other things.”
The thing is, once you accept that, then unless you can show convincing evidence of what those other things are, then I’m not sure what the point of those univariate regressions really is. And that’s been my whole point all along.
Stunney
I agree about the fecal infected grey matter however there is no doubt there are some people who decide to forgo work because they dont want to pay the extra taxes ( I work with some idiots like that). What they wrongly assume is that somehow this is bad for the economy, that somehow the work isnt getting done. Bullshit, someone else does the work and makes the money that they decided wasnt worth their effort.
In aggregate their is no loss of effort from a tax increase, it just gets redistributed. Someone will always accept the pay that you turn down.
Mr.Polley:
While I admire your work, the issue is not econ 101 as much as statistical. Prove how he uses the data correctly or incorrectly or otherwise you really do not have much to say on the matter whether you be a prof or not. Is the data distribution normal or non-normal? Anything else, as you have offered, is merely conjecture on your part. You should know better. Providing the evidence to disprove is your task and Kimel has provided all of his data for you to use.
William Polley,
The book comes out in August.
While the book is simple in its approach, you might be pleasantly surprised. We have an entire chapter where we look at graphs comparing the changes in the tax burden or the average tax burden itself over the length of an administration to things like changes in growth, debt, employment, etc. In each case we point out that the theory just didn’t match reality. (It didn’t occur to us to do what was done in this post, which is have first two years of tax vs. remaining years of the other variable but the fact is, I’m moving on past where we were in the book.) We point out in every case that even if the correlation is the exact opposite of what theory expects, we do not reach that conclusion at that point.
Then… at the end, in an appendix we do one regression (the only one in the book – we’re trying to keep it simple) that “explains” it all using multiple variables. Sadly, due to lack of observations, the regression is perfectly identified, but I’ve found dropping some of the variables doesn’t change the result. (Some variables cannot be dropped – off of memory (I’m operating on no sleep now) they are tax burdens and the effect of the Fed.) Besides, the main thing to watch are the signs of each of these variables, which in each case are what one would have expected from single variable analyses as described earlier in the book.
We do one thing beyond that… some of the variables in the model are beyond the control of the sitting Pres. Either they are controlled by the Fed or inherited from their predecessor. So we use the simple multivariate regression in the back of the book to pull out the piece of the growth that is “controlled by someone other than the current President” leaving behind what we call “earned growth.” (Once again, I’ve done something similar with fewer variables and it doesn’t change the results.) Earned growth, I suspect, will be a surprise to many.
William Polley,
Upthread I just put up a response to the multivariate regression issue.
Greg,
“What they wrongly assume is that somehow this is bad for the economy, that somehow the work isnt getting done. “
Bingo. With one mod… sometimes the work isn’t done, but something similar is done by someone else, and perhaps by someone who is less ruthless and therefore in ways less likely to impose massive externalities. The economy abhors a vacuum.
If Goldman’s people were discouraged from working, perhaps the economy would run faster.
“Sadly, due to lack of observations,…”
Well, choosing to group everything by presidential term will do that to you.
run75441,
I know you are a long-time reader and commenter of AB as I am, so I will give you the courtesy of a reply. For the benefit of all who might read my comments, I will say that when Mike first started on this project, he and I corresponded a bit by e-mail. He definitely understands my critiques, and it is my impression from all of my communication with him both publicly and privately that he would not say that my critiques are way off base or out of bounds.
I certainly have made conjectures in the comments from time to time. Given the nature of the conversation in comments, that is often necessary. However, Mike’s own statements reveal that he too is not above making some conjectures. For example, he admits that he will not use audit or settlement data as an indicator of enforcement because he claims there are different and contradictory explanations. So instead he links tax burden to growth by positing (without using the evidence I linked to and that he knows about) that tolerance for corruption is the link. If that’s not conjecture, I don’t know what is.
As for whether the data is normal or non-normal, well, the sample size here is so small that I think we all just throw up our hands and hope for the best on that one. That’s certainly not the heart of our disagreement.
What is the heart of our disagreement (and Mike, I think, knows this), is that a univariate regression of one macro variable on another needs to be taken with many, many grains of salt and a huge dose of humility. Perhaps the most concise statement of our disagreement is just how huge a dose of humility we require. He is willing to draw conclusions (or entertain the thought of drawing conclusions, or inviting his readers to draw conclusions) than I am. That is not particularly a matter of political ideology, but of both statistics and economics.
While I am probably a bit to the right of Mike and most of the people on this blog, I’m not one who could say that I’ve never met a tax cut I didn’t like. I also happen to believe that tax increases can raise or lower GDP depending on a number of specific factors. Because of that, and because I can allow for a lot of variation in those factors over time and across administrations, I find myself very skeptical of any conclusions one might draw from these univariate regressions. When I see that leaving out the outliers (FDR and W) leaves very little effect at all, it reinforces my prior belief that tax increases could have a positive or negative effect on GDP, depending on other things. I haven’t heard Mike say that enough in the posts–that’s the humility I’m talking about. But he said it in the comments, so it was nice to hear. It is economics AND statistics, not one or the other. Mike knows that. And I think over time he has come to understand my position, even if he doesn’t agree.
It’s not about disproving his result, because, quite frankly, I don’t see that he as proved anything. I’m not even sure he claims to have “proved” anything. I think at one time I was told (or it was suggested) that I was reading too much into his results. Ok, then what, pray tell, was the point? It has always been a bit elusive to pin down. The target keeps moving.
Ultimately, I enjoy these conversations with Mike. If I didn’t, I wouldn’t care about it, and would ignore it. His work is clever, thought-provoking, and kind of fun to talk about (and argue about). But at the end of the day, I’m not sure what I’m supposed to take away from it, and I’m suspicious that people might take more away from it than I’m prepared to. That’s our fundamental difference.
William Polley,
Well, the corruption angle is the best one I have for the facts at hand. I cannot provide any other reason why
a) cutting the tax burden in years 1 & 2 results in lower growth in real gdp per capita in years 3 through 8
b) cutting the tax burden in years 1 & 2 results in lower growth in real private investment per capita in years 3 through 8
(And that leaves out all the relationships I have brought up in other posts.) Obviously, higher tax burdens in and of themselves are not “creating” faster growth or more investment. So either this relationship that has held up through every objection many doubters have thrown at me (including, as you’ll see in the book, the multivariate analysis) is spurious or that tax burden has to be correlated with something. And frankly, it isn’t difficult for me to conclude that GW, Ike, Nixon and Truman led particularly corrupt administrations, or at least administrations where there was more money to be made playing the rules than playing by the rules.
GW is easy – the entire housing bubble is a perfect example. Nixon requires no explanation. Even Ike (out to golf) and Truman (plenty of people made a killing during the demilitarization).
William Polley,
If tree frogs matter for your hypothesis, the number of tree frogs available is your upper limit on observations. And frankly, it isn’t unreasonable to assume that a President has an effect on the economy. My bet is that both you and vote as if we believe that to be true.
Leave the tree frogs out of this… I like the tree frogs! 🙂
Sure, and you know what I meant. And I’m pretty sure that I believe that presidents have less of an effect on the economy than you believe they do. I wouldn’t say they have zero effect, but I’m pretty skeptical of claims that candidate X is going to save the economy (or tank it). In the end, their power is more limited than most of the general public would think.
If you want to argue corruption, and you don’t want to use the measures I suggested (because they don’t confirm your story), then why not construct an index of corruption for the presidents that seems plausible to you and see how that does?
You’re arguing that we see B and C move together which is not what conventional wisdom would predict. So you claim that it’s really A that affects C (and B is simply a manifestation of A). Ok, that’s possible. So then why not test the relationship between A and C?
And before anyone says that if I have a better idea that I should provide the evidence and refute yours, let me point out that you’re the one arguing that it’s corruption. And you’ve stated in so many words that you have chosen not to use the best extant direct evidence on audits, collections, settlements, convictions, etc. as I have suggested and instead rely on statements like “it isn’t difficult for me to conclude” and make comments about Ike’s golf habit. I realize you’re being facetious about the latter, but come on. This doesn’t prove anything.
In all honesty, Mike, I believe that it is entirely possible to observe higher tax burdens correlated with higher GDP in some instances depending on the many other factors (within and beyond a president’s control). Corruption may very well explain part of it. Actions of Middle Eastern dictators might explain part of it. Chance takes a hand as well. I can accept that a lot of things could contribute to what we see. But as you have found out, it is very hard to identify how much each of those contributes. Paucity of observations being a big limitation. So you’ve done the best you can–I get that. But ultimately there’s only so much you can do before you’re left with a bunch of just-so stories about the policies and politics of less than a dozen men.
And where does that leave us? If I infer conclusions or policy recommendations from your posts, you (or other commenters) maintain plausible deniability and say that you never meant it to be a policy recommendation and that you are just presenting the data. No economics, please; we’re statisticians. Except that you’re not just presenting the data. You’ve got your stories to explain it–some better than others. This was not one of your better ones. If I tell you you’re doing too much with too little, you can just say that you weren’t really trying to do that much.
Ok, I guess. At least I got you to say “…completely, overwhelmingly, and absolutely swamped by other things.” Now if you’ll just admit that with the data you have, you’ll almost certainly never determine any relative contributions of those other things with any statistical precision, that will be another step. Your statistics need a story to go along with them. But then, the story becomes bigger than the statistics because with such limited observations, the statistics just can’t do the heavy lifting. The statistics are a hook to get us to listen to your story. They don’t say much on their own.
“If Goldman’s people were discouraged from working, perhaps the economy would run faster.”
Aint THAT the truth!