PSLmodels
Components of interest deduction
In functions.py in Tax-Calculator, I believe the only interest deduction we use is e19200, defined as Sch. A: total interest deduction.
Realistically, there are different components, and it could be useful to differentiate between them. These include mortgage interest on homes, interest on various types of business debt, and investment interest paid (only deductible from investment income).
Is it possible to separate these different inputs to the total interest deduction amount we use now? If so, this could be useful in two ways:
- We could separately estimate the effective marginal subsidies on different types of debt.
- Potential reforms may eliminate some deductions but not others. For example, the Ryan-Brady plan from 2016 would eliminate the net interest deduction for pass-through entities (which could include interest on business debt) but it would not eliminate the mortgage interest deduction. It would be useful to be able to separate these different sources of interest income if possible.
I recently mentioned this to @andersonfrailey. I'd love to know if this request is possible to implement in taxdata.
@MattHJensen @martinholmer @Amy-Xu
@codykallen said:
In
functions.pyin Tax-Calculator, I believe the only interest deduction we use is e19200, defined as Sch. A: total interest deduction.Realistically, there are different components, and it could be useful to differentiate between them. These include mortgage interest on homes, interest on various types of business debt, and investment interest paid (only deductible from investment income).
Is it possible to separate these different inputs to the total interest deduction amount we use now?
You can answer that question by looking at the variables available in the 2009 IRS-SOI PUF. The documentation for that dataset is here.
If there are no components of e19200 in the PUF, then I don't see how we could decompose that variable.
@martinholmer mentioned the PUF documentation. The documentation includes
- e19200: Sch A: Total interest paid deduction (we use this)
- p25380: Sch E: Rent/Royalty interest expenses (not in records_variables.json)
- e58950: Form 4952: Total investment interest expense (not in records_variables.json)
Note that we do include e58990 (Investment income elected amount from Form 4952). However, due to specific problems with this variable (i.e. "elected amount"), we cannot use this to back out the actual interest expense.
Even if we cannot decompose e19200 into its components (e.g. personal mortgage interest, business mortgage interest, interest on other business debt), it would be useful to include p25380 and e58950 in the PUF.
@codykallen asked:
Even if we cannot decompose e19200 into its components (e.g. personal mortgage interest, business mortgage interest, interest on other business debt), it would be useful to include p25380 and e58950 in the PUF.
What is the relationship between p25380 and the other variables in records_variables.json?
What is the relationship between e58950 and the other variables in records_variables.json?
Are they subtotals of other variables we are using or what?
@martinholmer asked,
What is the relationship between p25380 and the other variables in records_variables.json?
p25380 is part of Sch E. The only two measures we currently use from Sch E are e02000 (Sch E net income or loss) and p25470 (royalty depletion or rental depreciation, which we use when calculating investment income for the EITC). In reality, p25380 has been subtracted from p25350 (gross rent/royalties received), and e02000 is net of this and other measures. But if the net interest deduction is eliminated for businesses, then the interest deduction on rent/royalties would presumably also be eliminated. For example, eliminating the interest deduction on rent or royalty income would be implemented by adding p25380 onto e02000.
@martinholmer also asked,
What is the relationship between e58950 and the other variables in records_variables.json?
On Form 4952, a taxpayer determines how much investment interest expense they can deduct. The taxpayer elects to offset the investment interest expense against some amount of investment income. The taxpayer chooses the actual amount of investment income to include in this, as the investment income listed on Form 4952 is not eligible for the reduced tax rates on long-term capital gains and dividends. The variable that we actually use (in the GainsTax function) is e58990, which is the net of the elected investment income and the investment interest expense as calculated on Form 4952. Eliminating the investment interest expense would increase e58990, increasing the amount of investment income taxed.
@codykallen provided:
... [much complicated information about
p25380ande58950] ...
It seems to me that if you want to make the case for adding these variables to the puf.csv file, you need to show exactly how they would be used in the Tax-Calculator source code. A quick read of the info you supplied left me in the dark about how these variables would be used in the code. How many PUF filing units have nonzero values for these variables? Also, how would we specify the values of these variables for the CPS records? And can you provide examples of reforms that would make use of these variables?
All this information is essential to conducting a benefit-cost analysis of adding the variables.
@codykallen A couple of papers are listed in #33 on analyzing reforms to the mortgage interest deduction. The papers or their authors might have some ideas for imputing the decomposition of e19200.
@martinholmer said
It seems to me that if you want to make the case for adding these variables to the puf.csv file, you need to show exactly how they would be used in the Tax-Calculator source code.
The main purpose of including these is to differentiate between different kinds of interest paid, as the different types of interest would not necessarily be treated equally under potential tax reforms. For example, suppose that a potential reform proposed eliminating the mortgage interest deduction but retaining the investment interest deduction. Although politically difficult, this is reasonable from a tax standpoint, as investment interest is typically viewed as a cost to investment (much as unreimbursed work expenses are deductible). The current version of Tax-Calculator would not allow us to implement this reform. If we can impute the inputs to e19200, we would need to modify the ItemDed function. We would replace the current ceiling and haircut on interest paid (ID_InterestPaid_*) with ceilings and haircuts on the components of total deductible interest paid (ID_MortgageInterestPaid_*, ID_InvestmentInterestPaid_*, ID_BusinessInterestPaid_*, etc.). This would allow a user to separately apply haircuts or eliminate the mortgage interest deduction, the investment interest deduction, and the net interest deduction for noncorporate businesses.
Another potential use of e58950 (investment interest) is to replace the use of e58990. My direct interest in this measure is to allow interest income to be taxed at the same rates as qualified dividends. If the net interest deduction is eliminated as part of a tax reform proposal, then the justification for taxing interest income at ordinary tax rates (tax arbitrage) is no longer valid. For example, the Ryan-Brady plan switches the taxation of interest income from the ordinary income schedule to the same basis as qualified dividends.
One of the problems we have currently is the use of e58990 as an input variable instead of a calculated variable. In reality, e58990 is the elected investment income against which the investment interest expense is deductible, and this depends on the various types and amounts of investment income an individual has as well as the amount of investment interest to deduct. On Form 4952, the filer chooses how much investment income to put down, but any income included on that form is eligible for taxation at the lower rates for ltcg and qdivs. Suppose that a filer has investment interest D, investment income taxable on the ordinary income schedule O, and investment income taxable on the preferential rate schedule P. The individual chooses to include some portion of their total investment income E.
- If
D < O, thenD <= E <= O. - If
O < D < O+P, thenO < E < O+P -
e58990 = E - D
The idea is that a filer chooses to include some investment income on Form 4952, and that the filer prefers to use income taxed at ordinary rates rather than income taxed at reduced rates. If investment interest is less than investment income taxed at ordinary rates, then I believe we would apply investment income taxed at ordinary rates to Form 4952. If investment interest exceeds investment interest taxed at ordinary rates, then we would apply investment income taxed at ordinary rates, and an amount of investment income taxed at preferential rates to make up the difference, such that e58990 would be zero.
One of the problems of using e58990 as Tax-Calculator does currently is that it inhibits the ability to add reforms to switch interest income to the preferential tax schedule. As long as we continue to use e58990 as an input, there will be some filing units for which e58990 includes interest income. For these units, switching interest income to the preferential schedule would not be useful, as the GainsTax function would ensure that all income included in e58990 is taxed at ordinary income rates.
@martinholmer also asked
How many PUF filing units have nonzero values for these variables?
I don't know, as p25380 and e58950 are not included in the PUF for Tax-Calculator, and I don't have the raw version.
@martinholmer also asked
Also, how would we specify the values of these variables for the CPS records?
I'm not particularly familiar with the CPS file, so I can't answer that.
SOI table 2.1 does include the distribution of components of total interest paid (e19200) by AGI class. We might be able to impute with the data there. The components provided there are
- Home mortgage interest (paid to financial institution/individuals)
- Deductible points
- Qualified mortgage premium insurance
- investment interest expense deduction
The imputed micro-data wouldn't be super accurate, but the distribution by AGI would be approximately right. Not sure whether how that's related to #33 since it seems to me those paper are more about mortgage value imputation.
