Analysis of effects of earnings shifting under Trump2017.json reform #1464
Conversation
Codecov Report
@@ Coverage Diff @@
## master #1464 +/- ##
=======================================
Coverage 99.71% 99.71%
=======================================
Files 40 40
Lines 2795 2795
=======================================
Hits 2787 2787
Misses 8 8Continue to review full report at Codecov.
|
|
Here is a summary of the implications of some arbitrary shifting assumptions taken from the The first row is the static/no-shifting analysis of the reform in the Notice that, in the @MattHJensen @feenberg @Amy-Xu @andersonfrailey @hdoupe @olsonjs |
|
Still new to GitHub, Python, and the tax calculator, so bear with me - You said if the parameters are met, the individual will "shift all of their earnings to pass-through income." Wouldn't they only shift the earnings that fit within the bracket(s) where the parameters are met, and not all of their earnings? i.e. say the regular income brackets are |
|
@olsonjs said:
Thanks for your question. All of this is very hypothetical, so there is no real-world experience to draw on. Irwin's NYT article speculates on what would actually have to happen between a salaried employee and his/her employer to accomplish the earnings shift. My reading of that article is that the shift from salaried employee to LLC contractor is a binary decision for the employee and employer. There is no practical way I can see that the same person could be both a salaried employee and a LLC contractor. There are too many pension and health insurance complications to make that arrangement viable. But maybe Irwin's speculation and my imagination are too limited. How would it work in the real world for somebody to be both an employee and a contractor? |
|
Good point. After reading around a bit, it is technically possible to be both an employee and a contractor in the same job, but not commonplace, and the IRS regulates the distinction a good amount. It would only be worthwhile to become both simultaneously if the marginal cost of doing so is less than the marginal benefit, but I would not know a non-arbitrary way of determining what the marginal cost is. Furthermore, let's assume that an individual can only be a contractor OR an employee. If a person has income from multiple sources, then they could certainly have a combination of income sources go through a pass-through and others not, depending on the interaction of marginal tax rates and cutting the income different ways, but I believe we don't have access to data with the necessary amount of precision, right? I imagine this would be most common among high-earners. |
|
@olsonjs said:
Not sure what other types of income you have in mind here, but yes it is unlikely that we have the data. So, the earnings-shifting estimates presented here are lower-bound estimates of the amount of income that might be shifted to a personal LLC, and therefore, lower-bound estimates of the aggregate amount of tax revenue loss from income shifting. |
|
On Mon, 10 Jul 2017, Martin Holmer wrote:
Here is a summary of the implications of some arbitrary shifting assumptions
taken from the earnings_shifting.sum summary results file, which is
generated by the earnings_shifting.sh script (which in turn calls the main
earnings_shifting.py script).
The first row is the static/no-shifting analysis of the reform in the
Trump2017.json file.
The other rows in the table below show the total number of taxpayers and
spouses who shift, the amount of earnings shifted, and the change in
combined tax revenue (relative to current law), under the set of shifting
assumptions on that row. In order to shift, it is assumed that an individual
must have earnings no less than MIN_EARN and tax savings no less than
MIN_SAVE. All the assumption sets below assume that if an individual exceeds
both these thresholds, then the individual will shift all of their earnings
to pass-through income. SPROB values of less than 1.0 would cause a
proportional reduction in the shifting response.
Notice that, in the earnings_shifting.py script, the shifting decision is
made, as it should be, for each individual (not for each filing unit as
apparently was done by TPC according my reading of #1461).
YEAR MIN_EARN MIN_SAVE SPROB : NS(#m) ES($b) TAXDIFF($b)
2017 0.00 0.00 0.000 : 0.000 0.000 -420.5
2017 200000.00 10000.00 1.000 : 3.631 1587.918 -621.3
2017 300000.00 10000.00 1.000 : 1.586 1097.008 -579.8
2017 400000.00 10000.00 1.000 : 0.889 858.575 -556.8
2017 200000.00 20000.00 1.000 : 2.335 1279.844 -599.4
2017 300000.00 20000.00 1.000 : 1.548 1080.905 -579.1
2017 400000.00 20000.00 1.000 : 0.883 852.803 -556.7
2017 200000.00 30000.00 1.000 : 1.348 1001.195 -574.3
2017 300000.00 30000.00 1.000 : 1.196 961.149 -569.1
2017 400000.00 30000.00 1.000 : 0.872 845.985 -556.3
2017 500000.00 30000.00 1.000 : 0.603 726.395 -542.7
2017 500000.00 40000.00 1.000 : 0.597 722.002 -542.5
2017 500000.00 50000.00 1.000 : 0.586 715.094 -541.9
This way of parameterizing the problem is not only arbitrary, it is
unlikely to align with empirical evidence when that becomes available.
Don't we think the eventual evidence from Kansas will be in the form of an
elasticity? Someday there is likely to be a result from a regression like:
log(pass_thru_income) = a + b*diff + c*log(wage+pass_thru_income)
where diff is the difference between the wage and pass_thru rates. At the
moment a, b and c are just as arbitrary, but by relating the change to the
tax differential there is better potential to settle on reasonable
coefficients than if the change is unrelated to relevant economic
variables.
I tried to find some statistics for how much wage income shifted to
pass_through entities in Kansas during the period it was profitable but
wasn't able to find anything. Perhaps a research assistant could find
something. That was law for such a short time it wouldn't be a good
long-term estimate though. In the long term most income would become
pass_through if the law was stable. Of course as time went by legislation
to restrict the formation of LLCs for this purpose would would be enacted
so the long term effect is really unknowable. We already have regulations
on the point while the tax saving is very small (just Medicare).
Having gotten to this point, perhaps the wisest course is not to model
a separate pass-through rate at all, since any revenue estimate that is
not extreme is very likely to be very wrong.
dan
… @MattHJensen @feenberg @Amy-Xu @andersonfrailey @hdoupe @olsonjs
—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub, or mute the
thread.[AHvQVYyTHAsK0ZlqXX8cmv8WGsVDb0Uuks5sMlhRgaJpZM4OTGTD.gif]
|
|
On Mon, 10 Jul 2017, Martin Holmer wrote:
@olsonjs said:
You said if [both] parameters are met, the individual will
"shift all of their earnings to pass-through income." Wouldn't
they only shift the earnings that fit within the bracket(s)
where the parameters are met, and not all of their earnings?
i.e. say the regular income brackets are
10% for income < $100k
25% for income > $100k
and the rate on all pass-through income is 15%.
If an individual earns $300k, they will only shift income above
$100k, not all of their income.
Thanks for your question. All of this is very hypothetical, so there is no
real-world experience to draw on. Irwin's NYT article speculates on what
would actually have to happen between a salaried employee and his/her
employer to accomplish the earnings shift. My reading of that article is
that the shift from salaried employee to LLC contractor is a binary decision
for the employee and employer. There is no practical way I can see that the
same person could be both a salaried employee and a LLC contractor. There
are too many pension and health insurance complications to make that
arrangement viable. But maybe Irwin's speculation and my imagination are too
limited. How would it work in the real world for somebody to be both an
employee and a contractor?
The LLC could pay wages to the owner which would be taxable to the
employee at the wage rate. thus allowing the income to be split between
wage and non-wage income without inconvenience to the ultimate employer.
Are we sure the proposed law doesn't include some regulation defining what
an employee is? There is currently a small tax benefit to non-wage income
that the IRS polices with modest success. It would become a huge deal
under this proposal, but I expect the problem is that the drafters of the
law don't appreciate the difficulties of enforcement rather than that they
expect to allow anyone to take advantage.
dan
…
—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub, or mute the
thread.[AHvQVU97BIxzawAKtAgYd0dFcmHsFQrkks5sMmHJgaJpZM4OTGTD.gif]
|
|
@feenberg said:
While I agree that the values are arbitrary and more empirical research on income-shifting needs to be done, I don't see why we shouldn't attempt to model it at all. Whether or not a tax reform falls apart to any extent due to income-shifting is an incredibly important part of the debate. Having a data point that says "this model estimates X amount in lost revenue if X amount of people shift income under these parameters" would be useful for policymakers in deciding how to handle pass-though income, even if the listed parameters are based on hunches for now. |
|
@olsonjs, In pull request #1464 I've been looking into the issue you raised about the notion that a person forming a personal LLC could lower tax liability even more by converting most, but not all earnings, to pass-though income. Your original comment in the discussion was here. @feenberg has suggested that the personal LLC has the legal right to split compensation between salary and pass-through income. So, even though the decision to form the personal LLC is binary, the division of LLC payments to the person is flexible and has no effect on the former employer, who is now contracting with the personal LLC for labor services. This comment is a summary of the work I've done on this issue. First, I added to the Second, I specified many positive values for the new parameter (ranging from one to forty thousand dollars in 2017) using what I understand to be the TPC assumptions about who would form a personal LLC. In all cases, the positive amount of LLC salary makes the person's tax liability do up, not down. I checked these aggregate results generated by the The aggregate results are for the The analysis of the single filing unit does not rely on the Here is what I did to confirm that this filing unit experiences an increase in (both income and payroll) tax liability when they have their person LLC pay $10,000 in salary income and pay the other $490,000 as pass-through income. The If you are interested in using Tax-Calculator on your computer, you could start by seeing if you can replicate my results. Looking at the intermediate tax values is likely to suggest the reason why tax liability goes up (rather than down) when the LLC pays a small amount of salary income. But it could be that I've made a mistake in my work. In that case, your effort in cross-checking my work could lead to your first Tax-Calculator bug report. |
|
Just to double-check, but are you having pass-through income always be subject to payroll taxes? I'm still shoddy with the actual code so I have been trying to replicate the problem in excel, and I think the unexpected result might have to do with the interaction with payroll taxes. For example, if salaried income is subject to payroll taxes, but pass-through income is not, then splitting income will always result in higher taxes paid if regular+payroll rates are higher than pass-through rates, since part of the income goes through payroll taxes, whereas if all income was pass-through, none of it would go through payroll taxes. I'm attaching my excel sheet, if that helps. Apologies for the messiness of the logic. shifting and payroll.xlsx |
|
@olsonjs said:
Pass-through income has been subject to payroll taxes for many years.
Yes, the payroll/income tax interactions are important. |
|
@olsonjs said:
I'm sorry I didn't make it clear, but using the command-line interface to Tax-Calculator, as I suggested above, does not involve doing any Python programming (or programming in any language, for that matter). We are happy to help you learn how to use the open-source tools we are developing (and fix those tools when bugs are reported), but our mission does not include helping the public develop Excel spreadsheets to do tax analysis. |
|
I completely understand. I just recently started being more active with OSPC (i.e. making suggestions on TaxBrain) while I whiddle away time to start my graduate program, and I'm excited to learn more as time allows. |
|
@codykallen said in a series of comment on issue #1461:
The Here are some 2017 results from First, consider the static, no-earnings-shifting results on combined tax revenue: Tax-Calculator estimates a loss in revenue of $420.5 billion in 2017. Next, consider the TPC assumption of large-scale creating of personal LLCs: fifty percent of those earning at least $100,000 per annum are assumed to eventually create personal LLCs and there is no minimum wage fraction regulation. Slightly more than nine million people, who have over $1.7 trillion in pre-reform earnings, are assumed to create personal LLCs. When those LLCs pay out compensation exclusively as pass-through income, the 2017 loss in tax revenue grows from $420.5 to $588.4 billion. That is an increase in lost tax revenue of $167.9 billion, which is about a 40% increase in the loss. If there was a regulation that somehow required 70% of the compensation shifted to the personal LLCs to be paid out as wages or salaries (and the remaining 30% was allowed to be paid out as pass-through income), the results would look like this if we continue to use the TPC assumptions: So the increase in lost tax revenue would be smaller: only $67.1 billion, which is a 16% increase in the loss. Finally, we switch to assuming that employers will negotiate the salary-to-contractor change only with their very highly paid employees, and that such employees will initiate those negotiations only if they save each year a substantial amount a money from forming a personal LLC. One way to operationalize these assumptions to say earning-shifting happens only for an employee with at least $500,000 in earnings who could save at least $50,000 per annum in taxes from setting up a personal LLC. We assume all of such employees create a personal LLC. Here are the results for those assumptions: Under these assumptions, when there is no regulation of LLC payouts, the tax revenue loss is $541.9 billion. That is an increase in the loss of $121.4 billion, which is an increase of nearly 29% from the no-shifting tax loss of the reform. Only 0.6 million people form personal LLCs, but those people have pre-reform wage and salary income of about $0.7 trillion. However, if regulations can effectively limit the compensation pay-out of the new personal LLCs, the number of people who can save at least $50,000 per annum declines by half to 0.3 million. The resulting loss in tax revenue goes down to $451.2 billion, which is only $30.7 billion more than the no-shifting loss. So, the percentage increase in lost revenue drops to about 7% under these more restrictive assumptions about who would create a personal LLC and contract with their former employer. But the handful of people who are simulated to create personal LLCs reduce their tax liability by about $116,000, on average, by doing this. What would actually happen under this kind of reform? It is very difficult to know in advance. Much depends on what regulations are included in the reform and whether IRS would have the budget to enforce any regulations effectively. Also, much depends on the willingness of employers to shift higher-paid employees to contractor status, and on the administrative costs of running a personal LLC. The point of this analysis is not to predict what would happen under this reform. The analysis is meant to show the tax implications of any set of assumptions about what might happen. And it is meant to illustrate the kind of advanced behavioral analysis that can be performed using Tax-Calculator. @MattHJensen @feenberg @Amy-Xu @andersonfrailey @hdoupe @olsonjs |
|
@martinholmer, this is a good addition. However, I don't seen any use of e00900 or e00900p (sole proprietorship income). Sole proprietorship income would be included in pass-through income, and thus eligible for the reduced rates in the Trump plan or the Brady-Ryan plan. Although it should be fine to use partnership income as the "destination" for earnings shifting, regulations applied to limit the amount of earnings shifting would apply to pass-through income as a whole ( |
|
@codykallen said about pull request #1464:
I'm not simulating the regulations for the whole reform; the Feel free to prepare a Tax-Calculator pull request the adds such a regulatory parameter to the @MattHJensen @feenberg @Amy-Xu @andersonfrailey @hdoupe @olsonjs |
This pull request adds several scripts and results files in the
taxcalc/reformsdirectory.These new files show how to use Tax-Calculator to estimate, under a wide range of behavioral assumptions, the effects of earnings-shifting: the transformation of wage and salary income into pass-through self-employment earnings from a newly-created personal limited-liability company (LLC) as described in several articles, including Neil Irwin, "Under the Trump Tax Plan, We Might All Want to Become Corporations," New York Times, April 28, 2017, which is here. There is also a recent request for this kind of analysis in issue #1461.
This work, which stands alone outside of the Tax-Calculator source tree, does not change any tax-calculating logic or tax results or code coverage. It is just an example of the kind of advanced tax reform analysis that can be performed with Tax-Calculator.
Read the documentation at the top of the
earnings_shifting.pyfile for information about how to conduct your own earnings-shifting analysis on your computer.Comments and suggestions are welcome.
@MattHJensen @feenberg @Amy-Xu @andersonfrailey @hdoupe @olsonjs