In a recent post, Patrick Honohan raised the issue of what a sustainable tax system would look like, and in a follow up to that post, discussed whether a goal of keeping low income workers out of the tax net implied, with the current tax revenue requirement, tax rates on other earners that were so high as to have serious disincentive effects. In the ensuing discussion, John McHale suggested that I was being too sanguine about the incentive effects at the top of the distribution and helpfully pointed me towards a literature that I wasn’t familiar with, on the tax rate elasticity of taxable income, and particularly to a paper by Gruber and Saez (J.Pub.Econ., 2002), which finds an average elasticity of 0.4, with higher elasticities for high earners.
There are two reasons why we should be worried if income elasticities for this group are so high. First, a pragmatic one: it suggests that revenue will rise relatively little if we increase tax rates on this group. Second, a more worrying one: this group contains the job creators; if they’re discouraged from taking the risks and reduce their labour market effort, then there are far bigger knock-on effects in jobs that would have been created with lower tax rates, but now won’t be. The latter concern dominates much of the discussion on this matter – see, for example, Greg Connor’s comment here.
And so, an elasticity of 0.4 would indeed have to cause a rethink on my part. So I went off to read the paper.
The paper is fascinating. It does indeed find an elasticity of taxable income to marginal tax rates of 0.4, with an even higher elasticity of 0.57 for high earners. (Note to explain the counter-intuitive sign: this is actually an elasticity wrt the net-of-tax rate, i.e. if the marginal rate goes up by 1%, so that the net-of-tax rate goes down by 1%, this causes a 40% decrease in income). But the elasticity of ‘broad’ income – income before tax exemptions are taken out – is much lower; it is 0.12 on average, and 0.17 for high earners. The bulk of the difference between these two elasticities is due to changes in what the authors call ‘itemization behaviour’ – in other words, tax avoidance. This point is reinforced by several other analyses in the paper.
One of the two policy conclusions drawn is that
“[t]he large elasticities that we observe are driven by ‘holes’ in the tax base that allow taxpayers, particularly at higher income levels, to reduce their tax burdens. With a broader tax base we would distort behavior less and could therefore raise revenues more efficiently.”
[The second is that concern about the distorting impact of high implicit tax rates in the $10k-$50k income range due to changes in effort (hours) “…may be overblown”, and that attention should instead be paid to incentives that reward participation rather than marginal increments to hours worked.]
So the paper’s message is (i) that the effect on (potentially job-creating) effort by high fliers of increasing tax rates is not zero, but is not high and (ii) that getting rid of tax write-offs should be a priority, particularly if marginal rates on high earners are to be raised.
9 replies on “Incentive Effects of Taxing High Earners”
Doesn’t the empirical literature tend to indicate that high marginal rates have relatively little effect on the labour supply of the primary earner in households, but a greater effect on second earners (i.e. predominantly women)?
With regard to the fear that “if [high earners are] discouraged from taking the risks and reduce their labour market effort, then there are far bigger knock-on effects in jobs that would have been created with lower tax rates, but now won’t be”: the extapolation of this fear would be, would it not, that countries with more compressed income distributions would have inferior growth and employment records. But, again, the empirical evidence doesn’t back up any negative relationship between these things, so I don’t think this is a huge worry.
(See, eg, Peter Lindert’s work on the welfare state as a “free lunch”: http://www.econ.ucdavis.edu/faculty/fzlinder/)
This is one research application where the US evidence is not sufficient for policy decisions. Ireland is an extremely open economy where high-earners come and go in response to preferences and opportunities, often bringing or taking capital investment with them. On the other hand, from the perspective of high-earners the US as tax home is almost a closed economy. Anyone who is a US citizen must pay the higher of US taxes or the taxes where they are a foreign resident (uniquely, the US chases its citizens around the world to pay taxes). The only migration incentive in the US case would be on foreign workers entering the US tax net — that is a small proportion of US high earners. Ireland on the other hand is very reliant on attracting high-earner DFI managers, foreign entrepreneurs, and keeping Irish high-earners here to create jobs.
A migration-based example (with admittedly different circumstances) would be US city income taxes. Philadelphia introduced a city income tax about 20 years ago and thereby did enormous long-term damage both to itself and to the transportation structure of the Philadelphia region. The business community moved out to widely dispersed business parks in the surrounding suburbs, leaving an impoverished city centre. Obviously moving a business from Philly to the suburbs is less drastic than moving from Dublin to say Warwick England, but this at least gives a more appropriate case study. I am much more worried about labour and associated capital mobility versus labour activity levels.
We need Irish-based evidence on marginal tax effects (including migration effects) to make good policies on this. Otherwise it would be sensible to be circumspect about going overboard on increasing top rates. I agree with the broad consensus that the top rate should be raised in current circumstances, but worry about the magnitude of the increase and its long-term damage.
I can’t say I’ve been following the discussion in the literature. But I have been following the discussion out here in the real world more closely: especially among the (diminishing) number of high earners (defined typically as those earning somewhere north of the sum earned by the person commenting on the issue).
That discussion points to a few interesting insights (but no elasticities), namely:
1. high earners typically have a large performance related element to their earnings either in the form of bonuses and/or high fixed salaries reflecting the profitablity of their employer – both are under pressure in every sector of the economy.
2. high earners are typically people who work long hours (without overtime), reflecting the nature of their sector, their clients or the time zones of their markets: they don’t usually like working long hours – but see the high salaries they earn as compensation.
3. high earners are usually married to other high earners (assortive pairing and all that) – and both tend to be subject to the same income/work trade offs in 1 and 2 above: if they have dependent children it’s an even less appealing arrangement.
4. high earners usually have more control over their working hours than lower earners: so those who haven’t been forced to reduce their hours/incomes due to the recession will nevertheless have the option of reducing their working hours if the marginal benefit falls substantially due to increased taxes. After years of busting a gut during the Celtic Tiger quite a few might be glad of a bit of down time.
In a nutshell, and take it from me this has been said to me by more than a few ‘high earners’, those taxpayers now contributing an unfair share of the income tax burden as it stands (5% contribute 47% of the total) will take a long, cold look at what they are gaining from the working arrangements and adjust their behaviour accordingly.
Maybe not a shrug, but certainly a shift.
The reintroduction of the remittance basis of taxation for certain foreign employees seconded to Ireland in the finance act – essentially for US employees of foreign MNCs – might help blunt any tendency for higher income taxes to induce capital flight.
@Greg, the Gruber and Saez paper actually might give some insight on the migration effect, as they estimate separate elasticities for federal and state taxes. For state taxes, the elasticity of broad income is 0.29, compared to 0.1 for federal taxes, which the authors attribute to the additional margin of response from residential mobility (neither of these is statistically significant, though). Assuming that moving state in the US is easier than moving into or out of Ireland, even if we are very open, we could regard 0.29 as an upper bound. I do take your point about an Irish analysis being vastly preferable, but the data are not available and highly unlikely to become available – it would require a large panel of individual-level tax returns.
Aedin, thank you for a very thoughtful response to the elasticity point.
You are absolutely right to point to the difference between the broad-income elasticity and the taxable-income elasticity — I should have done so also. I generally agree with your interpretation: the relatively low broad-income elasticity lessens the concern about reduced resident factor supplies (though, as Greg points out, there may be important effects along the residency dimension itself.)
But I continue to believe that the taxable-income elasticity is also relevant. Martin Feldstein’s initital theoretical work on this question showed the importance of this elasticity to the calculation of the direct welfare burden of the tax. This is not a point that will send people to the barricades, but the findings do tell us that the welfare losses suffered direclty by those on whom the tax is imposed are quite large — at least based on the US data. A standard measure of this burden is the “equivalent variation” of the tax; that is, the lump sum reduction in income that would produce the same welfare loss as the actual distorting tax. As I noted in the original comment, based on an elasticity of 0.5, the extra burden of raising an extra $1 in tax is approximately $1.50. Now we may well put relatively low welfare weights on high income earners in doing our welfare calculations, but that degree of excess burden/deadweight loss is hard to ignore.
It is worth pointing to one final finding from the paper that we haven’t mentioned. Gruber and Saez actually apply their elasticity findings to the optimal design of the income income tax (assuming that welfare weights decline with income). Putting aside the actual numbers, the implied optimal structure is quite interesting: “a tax system which is progressive on average but not on the margin, with a large demogrant that is rapidly taxed away at the bottom of the income distribution, but with marginal rates that are flat or falling with income.”
This is broadly consistent with a suggestion I made in response to Patrick’s post for that phasing out the tax credit for higher income individuals as an alternative to raising the higher tax rate. However, one concern with that idea is the resulting high marginal tax rate in the phase out range. Without pretending that a single article based on US data can be anything other than suggestive, it is interesting to quote Gruber and Saez one last time:
“[Low elasticities] suggest that the substantial concern currently expressed about the distorting impact of high implicit tax rates at the bottom of the income distribution may be overblown. Most of the concern is focused on the $10,000 – $50,000 income range that we examine where the EITC [Earned Income Tax Credit] is phased out. But we find no evidence that, at least for the explicit taxes that arise through the federal and state income tax system, taxpayers in this range are substantially changing either their real incomes or reported incomes in response to tax policy. This suggests that the distributional advantages of tightly income targeted tax subsidies may outweigh the efficiency costs of high implicit tax rates on the lower middle income taxpayers, as is illustrated by the high optimal rates in this bracket in our simulations.” (p. 29)
hear hear
the debate about the economic crisis should focus on how to get us out of here, rather than on fairness and whodunnit — the size of the cake should be our primary concern
besides the effects identified by Aedin, Gregory and Gerard, we should also consider
1. that high wages in Ireland partly compensated for low quality public services (health, schools, transport); and
2. that a well-informed person would expect that Ireland will have to maintain its “emergency taxes” for a longer period than other countries
I am not an economist but I wonder how this discussion is affected by the idea that in a context framed as a national crisis, one might hope that ‘high income’ individuals could be motivated in the short term to continue to work hard for the good of the country. In national emergencies generally, such as wars, natural disasters, etc., we take it for granted that people’s behaviour will be shaped not just by economic calculation but by a sense of duty. Why couldn’t that be tapped into here and now?
Thanks for all your hard work. The below article, written by me, appeared in today (Saturday’s) Irish Examiner newspaper on page 17. I would be happy for it or at least its ideas to be disseminated and discussed as widely as possible.
The Great ‘We Must Bail Out the Banks’ Myth and Why It’s Wrong
A dangerous myth stalking the streets and airwaves these days is that if
we continue to bail out the banks they will soon start lending again to
our hard pressed businesses and consumers. This is simply untrue.
After the destruction of what the Asian Development Bank described
recently as “US$50 trillion” since August 2007, there literally is no
longer enough money in the global financial system to satisfy the needs of
our banks and their creditors around the world.
Our banks undertook financial commitments at the height of the boom which
they still expect to be honoured (by tax payers) in the depths of the
bust. Our Government has no legal obligation to honour commitments
undertaken by non-state entities such as our banks.
Some have estimated the financial obligations of our banks to be in the
order of US$400b, if the Irish state agrees to guarantee all these
privately entered into debts (or even a quarter of them) we shall be in a
worse financial position than the world’s Least Developed Countries ever
were. Unless we accept this reality and allow our banks to renege on
financial commitments they made to banks around the globe over the past
ten years, we will face failing hospitals, schools and a stagnating
economy for a generation.
When we talk about ‘debt overhang’ and ‘non performing assets’ we are
really talking about financial commitments banks in Ireland like, say,
Anglo Irish undertook to hedge funds and other banking entities abroad at
the height of the boom.
It worked like this: Bank A abroad had X amount of money it wanted to get
a good return on. It lends X amount of money to, say, Anglo Irish Bank.
Anglo Irish Bank lends X to Developer One. The moment Developer One signed
on the dotted line, Anglo Irish sold the right to benefit ultimately from
the repayments on that loan to another, usually foreign, banking entity,
say, Bank B. This is called securitisation.
After it has sold this securitised debt to Bank B, Anglo Irish is only an
agent for Bank B and when Developer One made its quarterly payment to
Anglo, Anglo took a small cut and passed that payment on to Bank B.
Sometimes Bank A and Bank B are actually owned by the same people. In
most, if not all cases, Bank B will have sold on the right to benefit
ultimately from Developer One’s loan’s repayments to Bank C, and Bank C
has sold it on to Bank D and so on.
What all this, in effect, means is that when Ireland agrees to ‘bail out’
its banks it is agreeing to honour the commitments made by, say, Anglo
Irish to Bank A and to the Bank B’s and Bank C’s of the world. In effect,
instead of Anglo Irish being Bank A, B, and C’s agent, the Government is
undertaking that duty instead.
Ireland’s banks should be declared bankrupt. They are already Zombie banks
and bankrupt in all but name. Bank A, B and C are welcome to the property
and land around the country which underpins the securitised debts they
bought from our banks. The very same day we declare our banks bankrupt,
AIBv2.0, BoI v2.0 and so on could open up as ‘new’ banks, using the same
computer systems and premises, continue to take in and pay out our
savings. Sure, Ireland might be cut off Argentina style from the world
financial system. But would it really, especially when we do have well
performing loans which Bank A, B and C would still want to receive money
for?
Willem Buiter, ex-member of the Bank of England’s interest rate setting
committee, suggested why he fears our governments are not letting the
banks go bankrupt: ( http://blogs.ft.com/mavere con/2009/04/the-green-shoo
ts-are-weeds-growing-throu gh-the-rubble-in-the-ruins
-of-the-global-economy/ ).
“It is becoming increasingly hard to deny the possibility that the
extraordinary reluctance of our governments to force the unsecured
creditors (and any remaining non-government shareholders) of the zombie
banks to absorb the losses made by these banks, may be due to rather more
primal forms of state capture.”
It is some compensation after the recent budget that our government is not
alone in being captured by the very forces which caused the boom and the
crash. The US and the UK governments have similarly been captured by the
special pleadings and specious arguments of the Bank A, Bank B and Bank
C’s of the world and the snakeoil salesmen advisors who got us into the
mess they say only they know how to extricate us from. We must exercise
our sovereignty by letting our banks go to the wall and those creditors of
securitised loans originating in Ireland, where they can no longer be
honoured, should be left to sing for their money. The alternative is
decades of worse hospitals terrible schools and economic stagnation.
Stephen Douglas worked in the International Private Banking Department of Coutts & Co. London. He is Assistant Director of the Alliance Party of
Northern Ireland. He writes in a personal capacity (086 373 1476;
StephenDouglas@Cantab.net)