taxation of capital and labor: the diverse landscape by entity · pdf file ·...

Taxation of Capital and Labor: The Diverse Landscape by Entity Type

397

National Tax JournalVol. LXI, No. 3September 2008

Abstract - This paper derives standardized measures of labor and capital income that account for the different reporting requirements and incentives faced by different forms of organization. This facili-tates estimating how much capital income is subject to employment taxes and how much labor income escapes employment taxes (specifi -cally, those levied under the Federal Insurance Contribution Act (FICA) and Self–Employment Contributions Act (SECA)). The paper also shows an additional reason why economic inci-dence of taxes is different from statutory incidence, namely that different entity types are taxed on their labor and capital income in different ways, have different reporting requirements and, therefore, have differing incentives for taxpayers to mischaracterize their income with the intention of reducing liability. Ultimately, tax fi ling requirements cause some capital income of sole proprietors and general partners to be subject to employment taxes as if it were labor income. Furthermore, taxpayer behavior causes some labor income of limited partners and shareholders of S and privately held C corporations to escape employment taxes.

INTRODUCTION

The purpose of this paper is to derive standardized mea-sures of labor and capital income that account for the

different reporting requirements and levels of compliance faced by different forms of organization. Such an exercise facilitates estimates of how much capital income is subject to employment taxes, specifi cally, those levied under the Federal Insurance Contribution Act (FICA) and Self–Employment Contributions Act (SECA). Similarly, it facilitates estimates of how much labor income escapes FICA and SECA taxes.

Economists have long argued that economic incidence is different from the statutory incidence of a tax—in this paper, we show an additional reason why that is true. According to the standard view of incidence, if, in long–run equilibrium, total labor compensation equals the marginal revenue prod-uct of labor, then an increase in employer–paid employment taxes cannot increase total compensation. Therefore, the tax must necessarily reduce before–tax compensation, making it fully incident on labor. This paper emphasizes that there is a second reason why economic incidence can differ from statu-tory incidence, namely that different entity types are taxed on their labor and capital income in different ways, have differ-


Nicholas BullJoint Committee on Taxation, Washington, D.C. 20515

Paul BurnhamCongressional Budget Offi ce, Washington, D.C. 20515

NATIONAL TAX JOURNAL

398

ent reporting requirements and, therefore, have differing incentives for taxpayers to mischaracterize their income with the intention of reducing liability. Ultimately, tax fi ling requirements may cause capital income to be subject to the employment tax as if it were labor income. Further-more, taxpayer behavior may cause labor income to be taxed (or not taxed) as if it were capital income, and vice versa.

In this paper, we concern ourselves with four basic types of entity:

• C corporations that fi le Form 1120, 1120 A, or an 1120 consolidated return,1

• S corporations (Form 1120S),• Partnerships and LLCs (Form 1065),

and• Sole proprietorships (Schedule C or

F of Form 1040).

For each entity type, we discuss how the tax law treats capital and labor income, and what incentives are created for the taxpayer.

It helps to distinguish between “reported” labor and capital income and “true” labor and capital income. We defi ne reported labor income as any income cur-rently included in either the FICA or SECA tax base (ignoring the cap on taxable earn-

ings). True labor income, in contrast, is the amount that would be subject to FICA or SECA taxes if the owner were an employee being compensated fairly for his or her services.2 In each case, capital income is the difference between net operating business income (recalculated as if the business were a sole proprietorship) and labor income. Rental income and portfolio income, such as interest, dividends, and capital gains, are omitted from the analysis.3

We take two approaches to character-izing capital and labor income. In one approach, we attempt to determine the extent to which the partnership rules cre-ate a mismeasurement of capital and labor income when taxpayers accurately report their income. We do so by assuming that medium to large corporations have strong incentives to accurately report capital and labor income. We then examine how their currently reported capital and labor income would be treated if they fi led as a partnership. In this approach, we ignore taxpayers’ incentives to change the pat-tern of reported income.

In the second approach, we attempt to impute to taxpayers the true capital and labor income that is implied by their structural characteristics. To do this, we perform econometric analyses of privately held C and S corporations.4 Most corpo-

1 We exclude corporations that follow industry–specifi c tax rules, such as Regulated Investment Companies (1120–RIC), Life Insurance companies (1120–L), Property Casualty insurance companies (1120–PC), and so forth.

2 This defi nition means that some sources of income, such as “carried interest” income, will be treated as capital income, even though there is an active dispute as to their proper treatment. See Joint Committee on Taxation (2007) for a discussion of present law. Some argue that carried interests should properly be taxed as labor income (Fleischer, 2008) and others argue that they should properly be taxed as capital income (Weisbach, 2008).

3 Tax experts have identifi ed ways in which labor income has been recharacterized as either interest or rental income. For example, a taxpayer might attempt to make a loan to a partnership in which he holds an interest, with the resulting interest payments covering both the market interest and the partner’s labor contribution, thereby minimizing exposure to employment taxes. A taxpayer might also establish separate entities, with a service partnership leasing property from a property–owning partnership. To the extent the rent exceeded an arms–length rent, the result would be to mischaracterize labor income as rental income. But identifying the magnitude of such recharacterizations is beyond the scope of this paper.

4 In the second approach, we focus on privately held corporations, because large, publicly traded corporations appear unlikely to be able to collude to misrepresent the allocation of capital and labor income. The data do not directly distinguish privately held corporations from publicly traded corporations. They do, however, identify the number of shareholders up to 75. Presumably, all corporations with so few shareholders are privately held, so we limit ourselves to that subset.


399

rations have an incentive to misreport labor income, but doing so requires them to collude—an activity whose success becomes less likely as the number of own-ers increases. To account for the ability to collude, we incorporate information about numbers of shareholders into the econo-metric analysis. We use the econometric results to impute true capital and labor income to C and S corporations, and apply the results from S corporations to partner-ships and sole proprietorships. By com-paring results from the second approach with results from the fi rst approach, we also get a sense of the extent to which taxpayer behavioral response may affect reported labor and capital income.

RELATED LITERATURE

To our knowledge, no other paper has attempted to take into account the dif-fering fi ling requirements and incentives applicable to different entity types in providing an accurate measurement of true, underlying capital and labor factor incomes. Two streams of the literature are related to this paper. One stream focuses on measuring effective tax rates or payments and includes such notable contributions as those by Gravelle (1994), Mackie (2002), Gordon, Kalambokidis and Slemrod (2004), and U.S. Congressional Budget Office (2005). However, those papers generally start with the assump-tion that the tax return data accurately refl ect sources of income, and focus on developing measures of effective capital income tax rates or payments. This paper is motivated in part by that literature, and by the observation that capital income is measured inconsistently across entity types because of differing tax reporting requirements and incentives. Measure-ment of effective tax rates on capital (or labor) can only be as useful as the accuracy of the measurement of the underlying

income sources subject to those tax rates. Potentially, results of this paper might be useful in re–examining effective tax rates.

A second stream of related literature focuses not on effective tax rates, but rather on incidence. It is well known that a tax whose statutory incidence is on capital might be partially shifted to labor, in the sense that it causes reduced labor income. Examples in the literature abound and are summarized by Kotlikoff and Summers (1987). Again, mismeasurement of labor and capital (either before or after a tax–law change that provides signals about inci-dence effects) leads to the potential for mismeasurement of incidence; the present paper may be useful in this regard.

CURRENT LAW EMPLOYMENT TAXATION BY ENTITY TYPE5

The most straightforward tax rules are for sole proprietors—all net profi ts are subject to SECA taxes (and individual income taxes). Net losses are allowed to offset other self–employment income, but cannot result in negative income subject to SECA tax. In practice, however, hardly any losses are actually used to reduce SECA taxes, whether because there is no other self–employment income to offset or because the taxpayer is unaware that the offset is allowed.

Partnerships are required to report to their partners the amount of income subject to SECA tax separately from the amount passed through for income tax purposes. How that amount is deter-mined, however, depends on the type of partner.

• General partners. The distributive share of all operating business income received by general part-ners who are individuals is subject to SECA tax. This includes losses,

5 For a thorough discussion of differences in taxation by entity type, see Winchester (2006).


400

which, like sole proprietorship losses, can be used (but generally are not) by individual partners to reduce their SECA tax liability.

• Limited partners. The distributive share of operating business income allocated to limited partners is not subject to the SECA tax—limited partners who are individuals must pay SECA tax only on amounts they receive as “guaranteed payments” for services rendered.

• LLC members. Because LLC members are not classified as “general” or “limited,” there is ambiguity in the law concerning how they should be treated for SECA tax purposes.6 Actual practice does not seem at all uniform, with some following general partnership rules and oth-ers following limited partnership rules.

C corporations and S corporations are required to estimate a “reasonable com-pensation” for the services provided by their owners and deduct that from ordi-nary business income as “compensation of offi cers.” Amounts reported as offi cers’ compensation are subject to FICA taxes, just like employee wages. In the case of C corporations, ordinary business income is subject to the corporate income tax. In the case of S corporations, it is passed through to the owners to be included in their individual income tax calculations. Thus, they generally face opposite incen-tives to mischaracterize capital and labor income. These incentives are discussed in the next section.

INCENTIVES AND OPPORTUNITIES TO RECHARACTERIZE INCOME

Sole proprietors and general partners have no incentive to recharacterize labor

income as capital income or vice versa, because it will be taxed identically in either case. The incentive to minimize labor income facing S corporation owners and limited partners, however, is unam-biguous. The incentives facing C corpora-tion owners are more complicated.

S Corporation Owners and Limited Partners

Instead of being taxed at the business–entity level, S corporation and partnership income characterized as capital is passed through and taxed at the individual level. Income characterized as labor (whether as compensation of offi cers or guaranteed payments) is taxed at the individual level at the same rate as income from capital, so the individual income tax provides no incentive to recharacterize. But unlike capital income, labor income is also sub-ject to the FICA tax (for compensation of S corporation offi cers) or SECA tax (for guaranteed payments to limited part-ners). To avoid those employment taxes, S corporation owners and limited partners have a strong incentive to characterize as much labor income as possible as capital income.

Despite the relatively uniform incen-tives they face, S corporation shareholders and limited partners face different legal constraints on their abilities to rechar-acterize income. Limited partnerships have more flexibility to recharacterize income than do S corporations because they are allowed to (1) disconnect the distribution of capital income from the ownership of the capital assets, and (2) make guaranteed payments that do not refl ect reasonable compensation. S corpo-rations, in contrast, must distribute capital income in proportion to share ownership and comply with the reasonable com-pensation standard. Only the ambiguity

6 For a more detailed discussion of employment tax issues in S Corporations, partnerships, and LLCs, see Joint Committee on Taxation (2005, 95–104).


401

surrounding the defi nition of reasonable compensation leaves S corporations with any ability to recharacterize income with-out running afoul of the Internal Revenue Service (IRS).

In addition to legal constraints, partner-ships and multi–owner S corporations also face structural constraints on the ability to recharacterize income. The distribution of capital and labor among owners or part-ners has to be similar in order for recharac-terization to make all owners and partners better off. Single–person S corporations obviously face no mismatch between the distribution of capital and labor among owners, so there is no structural constraint on their ability to recharacterize income. As soon as a second owner is introduced, however, potential complications arise.

As the number of owners increases, it becomes more and more diffi cult to fi nd recharacterization opportunities that do not make somebody worse off. Unfor-tunately, available data do not support a comparison between the distributions of labor and capital income for most part-nerships and S corporations. Instead, we use the number of owners/partners as a proxy for the structural constraints to recharacterizing income faced by each business.

C Corporations

Unlike S corporations, the profi ts of C corporations are not passed through to the owners, but are taxed at the business–entity level. Furthermore, distributions of those profi ts are taxed at the individual

level, either as dividends or, when shares are sold, as capital gains. That makes the calculation of the overall tax rate on cur-rent profi ts rather complicated.

Let capital income (YK) be defi ned by equation [1] as follows:

[1] YK = R – X – YL – vYL,

where R is gross receipts, YL is the labor income of shareholders, X is all other deductible expenses (including com-pensation of non–shareholding employ-ees), and v is the average FICA tax rate (employer share only, after accounting for the portion of YL that is above the Old–Age, Survivors and Disability Insur-ance base and, therefore, subject only to Health Insurance taxes).

The optimization problem faced by the corporation then is to minimize, in the following equation, the overall tax burden (T) on shareholders:

[2] T = YK[u + (1 – u)tc] + YL(tw + pw),

where u is the marginal corporate income tax rate, tc is the marginal individual income tax rate on corporate income,7 tw is the marginal individual income tax rate on wages, and pw is the marginal FICA tax rate on wages (employee share only).

Substituting the right–hand side of equation [1] for YK and rearranging yields the following:

[3] T = (R – X)[u + (1 – u)tc]

+ YL{tw + pw – (1 + v)[u – (1 – u)tc]}.

7 Computing the effective individual tax rate on corporate income is nontrivial, and requires a number of assumptions. In particular, corporations can choose either to retain earnings or to distribute them. Retained earnings result in increased accrued capital gains, but, of course, the present value of the effective tax rate on those gains depends on the timing of their realization and the applicable capital gains rate at that time. Earnings can be distributed either by dividend payments or by share repurchases. In the former case, the applicable tax rate is determined by statute. In the latter case, the effective tax rate depends on the tax basis of the repurchased shares—for which we have little direct evidence. In addition, some shareholders may be nontaxable. Based on prior work with Compustat and other tax data for the 1990s, we assume that about 15 percent of corporate income is retained, 47 percent is distributed as dividends, and 38 percent is distributed as share repurchases.


402

Note that the fi rst term, representing gross receipts minus expenses not related to YL, is impervious to income recharac-terization and, therefore, irrelevant to determining the incentive to recharacter-ize. Thus, the incentive to recharacterize income can be determined by the sign of the coeffi cient on YL. If the sign is posi-tive, T is minimized by underreporting labor income; if the sign is negative, T is minimized by recharacterizing capital income as compensation for services rendered.

Plugging in actual values for the vari-ous tax rates yields a few easily identifi ed rules.8 First, corporations with losses can disregard the corporate income tax; for them, only labor income generates any tax liability. Thus, they have an incentive to minimize labor income regardless of the individual income and FICA tax rates. In contrast, at marginal tax rates of 35 percent or higher the corporate income tax is more onerous than the employment tax, so corporations have an incentive to maximize labor income regardless of individual income and FICA tax rates. For corporations in the 25 and 15 percent brackets, individual income and FICA tax rates determine the incentive. In the case

of the 25 percent bracket, the incentive is usually to overreport labor income, but it is much weaker than in the 35 percent bracket. Similarly, the incentive in the 15 percent bracket is to underreport labor income, but it is not as strong as the incentive faced by corporations with no taxable income.

Table 1 shows the coeffi cients on YL for each corporate tax bracket in each year between 2000 and 2004, relative to the estimated coeffi cient for corporations with zero taxable income—the group with the strongest incentive to underreport labor income. For instance, the 2004 value of 37 in the 15 percent bracket means that the incentive to underreport labor income in that bracket is 37 percent as strong as the corresponding incentive in the zero taxable income bracket. The value of –47 in the 35 percent bracket means that the incentive to overreport labor income in that bracket is 47 percent as strong as the incentive to underreport labor income in the zero taxable income bracket. The values of tc, tw, and pw have been calcu-lated as an average over all individual and employment tax brackets weighted by dividends plus capital gains. The differences by year refl ect primarily the

TABLE 1AVERAGE STRENGTH OF INCENTIVE OF C CORPORATIONS TO UNDERREPORT LABOR INCOME

BY YEAR AND MARGINAL CORPORATE TAX RATE

Year

20002001200220032004

No taxable income

100100100100100

15 percent bracket

3229263737

25 percent bracket

–13–18–24 –5 –5

35 percent bracket

–61–67–73–47–47

Source: Authors’ calculations based on Statistics of Income tax return samples.

Note: Values represent the coeffi cients on YL from equation [3], relative to the case of no taxable income.

8 We make the simplifying assumption that corporations only look at this year’s income. In fact, corporations might look at the expected pattern of their income over a period of years, to avoid altering their behavior year by year, in a manner that might appear to be purely tax–motivated, and that might, therefore, be thought to be a potential cause of concern either to shareholders or to the Internal Revenue Service. Whether this simplify-ing assumption biases results in a signifi cant way is a potential topic of future research. One indication that it may not result in signifi cant bias is the fairly high serial correlation across years in fi rms incentives to under or overreport labor income.


403

changes in the tax law that were enacted during the period. The values in this table are used for the “incentive” variable discussed in the econometric analysis below.

As for structural constraints on rechar-acterization, they are basically the same for privately held C corporations as for S corporations. Publicly traded C corporations have additional structural constraints in that the vast majority of shareholders contribute no labor income and should have little patience with recharacterizing capital income as compensation for services—hence, their exclusion from the analysis.

MEASURING REPORTED LABOR AND CAPITAL INCOME

Standardizing Operating Business Income

Before identifying labor and capital income, we must first standardize the measure of business income that is to be divided between them. Each form of organization uses a different tax form with different elements to calculate net operating business income. Those differ-ent elements make the measures of net operating business income reported on the forms inconsistent with one another. The three types of entities whose income is passed through to their owners’ tax returns all have a similar concept of oper-ating business income—one that excludes rental and portfolio income and expenses. Nevertheless, the amount reported as “net operating business income or loss” for partnerships and S corporations is not quite the same as the “net profi t or loss” reported on Schedule C by sole proprietorships. We use the latter as the standard measure of operating business income of owners and adjust the calcula-tions from the other forms to conform to that convention.

In the case of partnerships and LLCs (Form 1065), we normalize by starting with “ordinary income or loss from trade or business activities” and subtracting income from other partnerships, estates and trusts, gain or loss from Form 4797, and “other income” because those items do not appear on Schedule C or F. Amounts expensed under Section 179 are included on the depreciation lines of Schedules C and F but not of Form 1065. Therefore, we also subtract that amount (as reported on Schedule K of Form 1065). Finally, we add back any guaranteed payments to partners because no such payments to the owner can be deducted on either Schedule C or F.

In the case of S corporations (Form 1120S), we subtract from “ordinary income or loss from trade or business activities” the gain or loss from Form 4797, “other income or loss,” and Section 179 expenses. We add back the compensation of offi cers.

Normalizing the operating business income of C corporations (Form 1120) is more complicated, because rental and portfolio income and expenses are reflected in the calculation of taxable income. In this case, we normalize by starting with “total income” less “total deductions,” ignoring net operating loss carryovers and special deductions. Like Forms 1065 and 1120S, we subtract the gain or loss from Form 4797, “other income or loss,” and Section 179 expenses, and we add back compensation of offi cers. However, because “total income” includes portfolio and rental income, we also sub-tract dividends, interest, gross rents, gross royalties, and capital gains. Furthermore, “total deductions” includes amounts asso-ciated with portfolio income and gross rents, which must be added back. Unfor-tunately, those amounts are not identifi ed separately for C corporations. Therefore, we estimated percentages of deductions attributable to portfolio or rental activ-


404

ity for each C corporation based on the average share of such deductions for S corporations with a share of portfolio or rental income similar to the specifi c C corporation. Finally, we add back the deduction for charitable contributions, which is not included in the calculation of net operating business income for any other form of organization.

Table 2 summarizes the operating business income of each type of busi-ness. The fi gures represent the average over the period 2000 through 2004 and are expressed in 2002 dollars. Note that removing rental and portfolio income from the calculation left many C corpora-tions with negative net operating business income. In part, this may refl ect a rela-tively recent trend among corporations toward holding large amounts of assets that generate portfolio income (Warsh, 2006). The table also demonstrates how Form 1065 fi lers have sorted themselves by organizational form—fi rms with losses overwhelmingly chose the limited part-nership and LLC forms.

Identifying the Portions of Operating Business Income Taxed as Labor and Capital

Given the standardized defi nition of operating business income, we attempted to identify the reported labor portion for each entity type by following the

rules concerning FICA and SECA taxes described above. The remaining portion of operating business income was then deemed to be reported capital income.

In the case of sole proprietors, all net profi ts are subject to SECA taxes, so we count them as reported labor income. Because net losses are rarely used to reduce SECA taxes, however, we count them as reported capital income.

To approximate the treatment of sole proprietors, we count as reported labor income for partnerships and LLCs the greater of reported self–employment income or guaranteed payments. That leaves the following to be counted as reported capital income: (1) all operating business losses, (2) the distributive share of operating business income allocated to corporations, other partnerships, and exempt organizations, and (3) the dis-tributive share allocated to individual partners in excess of labor income.

We count compensation of offi cers as reported labor income for both S corpo-rations and C corporations. We count the residual amount of total operating business income as reported capital income.

Table 3 shows the amounts of labor income reported by each type of entity. The table highlights the low levels of labor income reported by limited partnerships and by firms with negative operating business income.

TABLE 2NET OPERATING BUSINESS INCOME1 BY TYPE OF ENTITY

Sole ProprietorsForm 1065 Filers General Partnerships Limited Partnerships Limited Liability Companies (LLCs) Limited Liability Partnerships (LLPs)S CorporationsC Corporations (<75 Shareholders)

Firms w/Positive Business Income

(millions)

$271,784$200,910 $45,952 $52,769 $72,363 $29,826$311,087$169,380

Firms w/Negative Business Income

(millions)

–$55,940 –$78,615 –$7,953 –$22,142 –$47,348 –$1,172 –$65,909–$106,420

All Firms (millions)

$215,844$122,295 $37,999 $30,627 $25,015 $28,654$245,178 $62,960

Source: Authors’ calculations based on Statistics of Income tax return samples.1Operating business income is the amount that would be reported as “net profi t or loss” had the entity fi led as a sole proprietor. Amounts are averages over the period 2000 through 2004.


405

Theoretical Shortcomings of the Taxation of Reported Labor Income

In some cases, counting all business income of sole proprietors as labor income is theoretically correct. To the extent that sole proprietors make any capital invest-ments, however, correct measurement of labor income would allow for some return on that investment. Hence, reported labor income in the aggregate overstates true labor income. Unfortunately, neither Schedules C nor F include balance sheet information that would permit a direct estimate of capital investment on which a return should be expected. Instead, we use data from S corporations to estimate reasonable compensation for services provided by sole proprietors, as described below.

Similarly, the reported labor income of general partners overstates true labor income because it ignores the return on capital investment. In contrast, the labor income of limited partners is understated. The guaranteed payments that are sub-ject to the SECA tax are not required to represent reasonable compensation for the limited partners’ services. In fact, as explained above, there are powerful incentives for limited partners to accept less than reasonable compensation for

their services. As for LLC members, it is not clear from theory how well their reported labor income should conform to true labor income because of the legal ambiguities. To estimate the magnitude of the errors in the aggregate, we estimate reasonable compensation of partners and LLC members using the same technique as for sole proprietors.

Because corporations are supposed to pay reasonable compensation to their own-ers for services rendered, their reported labor income is, in principle, correct. However, the incentives and opportunities to recharacterize labor income as capital income (and vice versa) described above render the reported amounts unreliable. There is also outright noncompliance, as evidenced by the 42 percent of S corpora-tions and 34 percent of C corporations with less than 75 shareholders that report zero for compensation of offi cers.

Having discussed issues in measuring reported labor and capital income, we turn next to quantifying the effects of the two reasons why reported income differs from true income: filing requirements and the incentives that they create. We fi rst attempt to quantify how much of the difference owes to reporting requirements in the absence of behavioral response. Next, we use an econometric approach

TABLE 3REPORTED LABOR INCOME BY TYPE OF ENTITY

Sole Proprietors1

Form 1065 Filers2

General Partnerships Limited Partnerships Limited Liability Companies (LLCs) Limited Liability Partnerships (LLPs)S Corporations3

C Corporations (<75 Shareholders)3

Firms w/Positive Business Income

(millions)

$271,784 $92,660 $25,434 $3,870 $31,977 $31,379$139,911$129,620

Firms w/Negative Business Income

(millions)

$0 $1,641 $206 $189 $1,206 $40$11,453$13,658

All Firms (millions)

$271,784 $94,301 $25,640 $4,059 $33,183 $31,419$151,364$143,278

Source: Authors’ calculations based on Statistics of Income tax return samples.1Reported labor income=positive operating business income.2Reported labor income=greater of profi ts distributed to partners who are individuals and guaranteed pay-ments.3Reported labor income=compensation of offi cers.

Note: Amounts are averages over the period 2000 through 2004.


406

to quantify the full economic effects of both the reporting requirements and the incentives that they create.

THE EFFECT OF FILING REQUIREMENTS WITH NO BEHAVIORAL RESPONSE

To quantify the effect of differing fi ling requirements on reported labor and capi-tal income, we look at how the reported income of C corporations would change if they fi led following partnership rules. For the purpose of this exercise, we do not need to examine the effect of fi ling requirements if the corporation were fi l-ing as sole proprietorship (because that is essentially the same as a one–person partnership), nor as an S corporation (for which fi ling requirements themselves do not recharacterize capital income as labor income).

To examine this question, we formed a panel of C corporations that fi led tax returns in every year from 2000 to 2004, converting all monetary values to real dollars (2002 base year). Then for each corporation in the sample we computed relevant average incomes across the period. By constructing this panel, we hoped to minimize the effects of tran-sitional issues such as fi rm startups or

failures, and somewhat to ameliorate business–cycle effects: the intention is to get closer to an “average” C corporation. We then made the adjustments discussed above to normalize C corporation income to a passthrough income concept. In particular, we subtracted net passive income, which includes dividends, inter-est, rents, royalties, and capital gains, all net of (imputed) deductions relat-ing to these income sources. As Table 4 demonstrates, the average corporation in each size category has positive net income subject to the corporate income tax, but only because their passive income sources are suffi ciently large to offset their negative net operating business income. Notably (and foreshadowing the results presented in the next section) the share of gross business operating income that is allocated to labor factor payments is signifi cantly lower for the smallest fi rms (13.4 percent for C corporations with only one or two shareholders) than for the larg-est fi rms (approximately 19.5 percent); middle–sized fi rms are closer to the larg-est.9 The table also shows total tax liability generated both at the corporate level and, for actual distributions of dividends, at the individual level.

Next, in order to demonstrate how fi ling requirements affect reported labor

TABLE 4SUMMARY OF CHARACTERISTICS OF C CORPORATIONS IN PANEL BY NUMBER OF SHAREHOLDERS

Average Amount per Firm (2002 dollars) Total Net Income Subject to Corporate Income Tax Net Operating Income Net Passive Income Labor Factor Payments TaxesLabor Payments/Gross Operating Income (percent)

$18,853–$45,623$64,476

$314,367$10,574

13.4

$76,470–$214,382$290,852

$1,566,296$51,505

17.7

$1,431,733–$3,404,977$4,836,710$6,515,595

$603,65219.5

1–2 Shareholders

3–75 Shareholders

More than 75 Shareholders

Source: Authors’ calculations based on a panel of C corporation tax returns from the 2000 to 2004 Statistics of Income samples.

Notes: Labor Factor Payments = Wages and salaries, compensation of offi cers, and employee benefi ts.

9 Note that the concept of “labor factor payments” is not the same as that of “labor income” presented elsewhere in the paper. The former refers to total compensation, including that of employees. The latter refers only to the compensation of business owners.


407

factor payments, we consider the effect of allowing C corporations to fi le as if they were partnerships. To do so, we need to determine whether a shareholder would classify as a general partner or as a lim-ited partner. This determination of who is a general partner would depend, of course, on any specifi c enabling legisla-tion. But it seems reasonable to assume that it would include corporate directors and offi cers, as well as owners with a high share of control such as the founder, family members, and key investors. Using data from Amit and Villalonga (2007), we assume that general partners would account for approximately one–sixth of the ownership of publicly traded fi rms. For the smallest C corporations (i.e., those with fi ve or fewer shareholders) we assume that all of the owners would be treated as general partners. Furthermore, we assume the share of income accru-ing to general partners would decline steadily with the number of sharehold-ers. Finally, we take account of the likely

share of the limited partners that are non-taxable.

Table 5 shows how measured labor fac-tor payments subject to SECA taxes would change if C corporations fi led as partner-ships. The size (and sign) of the change would depend on whether the general partners have other self–employment income that could be offset by net operating losses (NOL). If not, then income subject to employment taxes would rise because some capital income would be swept up as labor income by the partnership fi ling rules. If, on the other hand, NOLs could offset other self–employment income, then income subject to employment taxes could actually decrease. In that case, the sign of the overall change in employment taxes would depend on the relative magnitudes of the positive capital income of general partners on the one hand, and the other self–employment income that would be offset by NOLs on the other.

If NOLs could completely offset other employment income, then labor factor

TABLE 5AVERAGE CHANGE IN LABOR PAYMENTS AND TAXES IF C CORPORATIONS

IN PANEL WERE TAXED AS PARTNERSHIPS

1–2 Shareholders

3–75 Shareholders

More than 75 Shareholders

Change in Labor Payments Subject to Employment Taxes (2002 dollars) With NOL offset Without NOL offsetChange in Labor Payments Subject to Employment Taxes/Gross Operating Income (percent) With NOL offset Without NOL offset

Taxes (2002 dollars) Individual Income Taxes SECA Taxes With NOL offset Without NOL offsetChange in Taxes Due to Partnership Filing Requirements (2002 dollars) With NOL offset Without NOL offset

–$45,623$19,709

–7.51.4

$7,460

–$1,129$488

–$4,243–$2,626

–$168,061$70,555

–3.41.2

$40,925

–$4,160$1,746

–$14,739–$8,833

–$567,496$97,090

–1.70.3

$547,450

–$14,046$2,403

–$70,247–$53,798

Source: Authors’ calculations based on a panel of C corporation tax returns from the 2000 to 2004 Statistics of Income samples.

Notes: SECA = Self–Employment Contributions Act. With NOL offset = entire net operating loss offsets other self–employment income of general partners. Without NOL offset = general partners have no other self–employment income to be offset by net operating loss.


408

payments subject to employment taxes would be lower as a share of gross operat-ing income when fi ling as a partnership. The decrease would be between 1.7 per-centage points and 7.5 percentage points. If NOLs could not be used to offset other self–employment income, then labor fac-tor payments subject to employment taxes would be higher as a share of gross operat-ing income when fi ling as a partnership. The increase would be between 1.2 and 1.4 percentage points for the fi rms with fewer than 75 shareholders that are the focus of the next section.

The effect on total tax liability of allow-ing a C corporation to fi le as a partnership would be negative, refl ecting both the elimination of entity–level taxes and the preferential treatment of capital gains. The size of the reduction would depend on whether or not the general partners could offset other self–employment income with NOLs. If not, then employment taxes would rise; if NOLs could completely offset other self–employment income, then employment taxes would fall. The table shows, under both assumptions, the net change in total taxes per firm resulting from fi ling requirements alone, and in the absence of any behavioral response.10

This analysis demonstrates that the average corporation would be better off from a tax standpoint filing as a partnership. For a variety of legal and organizational reasons, however, they do not do so. The analysis also hints that corporations with one or two shareholders might be more successful at mitigating the tax disadvantages of corporate fi ling by minimizing their labor factor pay-ments subject to employment taxes. In the next section, we attempt to quantify the extent to which corporations have

successfully minimized their employment tax exposure by recharacterizing the labor income of shareholders as income from capital. We also attempt to quantify the fl ip side of the issue, namely how much the partnership (or sole proprietorship) fi ling requirements place an additional burden on capital income.

TECHNIQUES FOR MEASURING TRUE LABOR AND CAPITAL INCOME

In general, we use the characteristics of corporations reporting offi cers’ compen-sation to impute amounts to corporations reporting none. We also develop a tech-nique that produces estimates of under-reporting by firms reporting nonzero amounts. For Form 1065 fi lers and sole proprietors, we use parameters derived in our analysis of S corporations to impute true labor and capital income. We used S corporations because under the simulated conditions (namely, where the reasonable compensation standard applies to Form 1065 fi lers and sole proprietors), those entity types would face the same incen-tives as S corporations.

How to Estimate Labor Income When None Is Reported

We use regression techniques to relate reported offi cers’ compensation to explan-atory variables including industry code, gross receipts, wages and salaries, and some direct or indirect measure of capi-tal. We expect offi cers’ compensation to increase with gross receipts (because more revenue requires more management) and to decline with wages (because work done by employees does not have to be done by owners). We generally expect negative coeffi cients on the capital–related vari-

10 Determining the overall effect on federal revenues would require a more careful assessment of the impact on individual taxpayers, as well as an assessment of the likely behavioral responses, and is, therefore, out-side the scope of this paper. In addition, note that the panel data from which these effects are derived is not cross–sectionally representative and may not be representative of currently active fi rms.


409

ables on the theory that capital substitutes for labor.

To impute an amount of officers’ compensation to returns reporting none, we simply apply the coeffi cients of the regression to the explanatory variables on those returns.11 Identifying returns reporting no offi cers’ compensation to which to impute amounts, however, is not as straightforward as it sounds. Some fi rms might have erroneously reported their offi cers’ compensation on the line for employee wages and salaries. In such cases, the zeros would refl ect not a shifting of income between labor and capital, but merely a shifting of labor income between offi cers and employees. Such a shift has no effect on tax liabilities and is, therefore, benign. Approximately half of the compa-nies reporting no offi cers’ compensation also reported no wages and salaries. Of those reporting wages and salaries, the data do not support a rigorous attempt to identify fi rms reporting offi cers’ compen-sation as regular wages. In the sensitivity analysis below, we compare estimates of nonreporting under the following assumptions as upper and lower bounds: (1) only zero values for fi rms reporting wages and salaries less than the expected value of offi cers’ compensation represent shifting to capital income (the lower bound), and (2) all zero values represent shifting to capital income (the upper bound).

How to Estimate Misreporting of Nonzero Values

There are at least two related techniques that can be used to detect, in the aggregate, misreporting of offi cers’ compensation on corporate income tax returns. The more complex technique (Method A) exploits all of the information on corporate returns that report nonzero amounts of offi cers’

compensation. It depends on the assump-tion that misreporting is systematically related to the number of shareholders and, perhaps, to other variables such as gross receipts. In this technique, we construct a series of variables that rep-resent the obstacles to recharacterizing income. Because those obstacles increase with the number of shareholders, we cre-ate a dummy variable for all fi rms with two or more shareholders (pt2), another for fi rms with three or more sharehold-ers (pt3), and so on up to nine or more shareholders (pt9). We then interact each of those variables with the incentive vari-able whose values are shown in Table 1 (producing what are referred to hereafter as “temptation” variables) and estimate the coefficients on each. One might expect corporations with a high level of receipts to misreport more egregiously than those with a lower level of receipts. Therefore, we also interacted the tempta-tion variables with gross receipts and esti-mated the coeffi cients on those terms as well.

When the incentive variable is positive (that is, the incentive is to underreport offi cers’ compensation), the interpretation of the coeffi cient on “incentive*ptx” is the amount by which offi cers’ compensation should be higher for a corporation with x or more shareholders than one with x – 1 shareholders. When the incentive vari-able is negative (that is, the incentive is to minimize income subject to the corporate income tax), the interpretation of the same coeffi cient is the absolute value of the amount by which offi cers’ compensation should be lower for a corporation with x or more shareholders than one with x – 1 shareholders. Regardless of the sign on the incentive variable, the coeffi cient on the temptation variable should always be positive, so we eliminated tempta-tion variables with negative coeffi cients

11 Note that at this stage, the imputed amounts still refl ect a level of underreporting similar to that of returns reporting nonzero amounts.


410

until only those with positive coeffi cients remained.12

We estimate underreporting of non-zero values by applying the regression coefficients to all returns after setting all values of ptx equal to one; in other words, we assume conditions in which we could fi nd no evidence of recharac-terizing income. The amount by which that estimate exceeds either the actual or imputed amount represents actual or potential underreporting. Ultimately, only two or three temptation variables with positive coeffi cients emerged from each regression. In most regressions, the only underreporting that could be detected statistically was by corporations with one or two shareholders.

An alternative technique (Method B) is to arbitrarily set a threshold number of shareholders above which report-ing is assumed to be accurate, estimate the regressions on that population, and apply the coefficients to corporations with a number of shareholders below the thresholds. This technique picks up any underreporting by corporations below the selected threshold, even if it was not statistically detectable using Method A. However, this technique ignores any relationship between underreporting and number of shareholders among corpora-tions below the threshold. Furthermore, the selection of a threshold is arbitrary and still risks excluding misreporting above that threshold.

Selecting a Functional Form

We tested two different functional forms. In the fi rst, a logistic regression, the

dependent variable was labor’s share of the operating business income of owners: YL/(YL + YK). Although usually associated with binary dependent variables, a logistic regression can accommodate dependent variables with continuous values such as this one. Unfortunately, it cannot accommodate dependent variables with values greater than one (which occurs here when YK is negative but smaller in absolute value than YL) or less than zero (which occurs here when YK is negative but larger in absolute value than YL). Therefore, we were limited to using the logistic functional form for records with positive values for both YL and YK. The key explanatory variables were the log of gross receipts and wages/long–term capital account.13

In the second functional form, esti-mated using Ordinary Least Squares, the dependent variable was the log of YL, and the key explanatory variables were the logs of gross receipts, wages, and four variables that depend on the amount of capital in service: depreciation, repairs, interest paid, and rent paid. This form allowed us to include fi rms with nega-tive values of YK, but presented its own disadvantages. First, it does not relate labor income directly to income from capital as the logistic form does. More important, however, is that although the form generates within–sample predicted values of log(YL) that are correct in the aggregate, that does not translate into the correct aggregate values of YL—the predicted aggregate values of YL were persistently low in every regression. That has signifi cant implications for applying the coeffi cients out of sample. Without

12 This practice compromises the ability to interpret all of the coeffi cients properly, but is necessary to avoid implausible simulation results.

13 The long–term capital account is equal to depreciable, depletable, and intangible assets (net of accumulated depreciation, depletion, and amortization) plus other assets (net of other liabilities) that are necessarily linked to a business rather than its owner (for example, mortgages payable and accounts receivable net of accounts payable). This defi nition was thought to be more tractable for application to sole proprietorships than one that included assets that could be comingled with an owner’s personal accounts, such as cash and short–term securities.


411

adjustment, they would produce imputa-tions that are too small. For corporations, we adjusted aggregate within–sample predicted values by gross receipts class to hit actual aggregate values. We then applied the same adjustment factors to amounts imputed as true labor income.

Regression Results and Sensitivity Tests

The main results presented in this paper refl ect the upper–bound estimates from the log–log functional form and imputation method A. That combination is referred to hereafter as the base specifi -cation. We also ran a series of sensitivity tests on both S and C corporations with

positive capital income to see how the simulated labor share of operating busi-ness income varied according to func-tional form (logistic or adjusted log–log) and imputation method (A or B). Within imputation method B, we tested two dif-ferent thresholds above which reported labor income was assumed to be accurate (three and ten). Finally, we generated both upper– and lower–bound estimates. In these tests, we omitted all industry dummies.

For S corporations, the base specifi ca-tion produced coeffi cients that generally conformed to expectations (see Table 6). The only exception was depreciation, which had a positive (albeit insignifi cant)

TABLE 6COMPARISON OF REGRESSION COEFFICIENTS FOR S CORPORATIONS

Adjusted Log–Log Form

Method A

Method B

3+ Owners 10+ Owners Logistic Form

Constant

Log(gross receipts)

Log(wages)

Log(depreciation)

Log(repairs)

Log(rent paid)

Log (interest paid)

Wages/capital account

pt2*incentive

pt2*incentive*log(gross receipts)

pt3*incentive

2.8614*(0.0514)

0.6177*(0.0042)

–0.0033*(0.0007)

0.0012 (0.0009)

–0.0256*(0.0008)

–0.0045*(0.0008)

–0.0188*(0.0008)

—

—

0.0108*(0.00046)

0.2258*(0.00995)

3.1590*(0.0785)

0.6039*(0.0060)

0.0148*(0.0014)

–0.0025(0.0017)

–0.0374*(0.0012)

0.0062*(0.0012)

–0.0075*(0.0011)

—

—

—

—

2.0094*(0.1709)

0.6257*(0.0017)

0.0398*(0.0035)

0.0210*(0.0045)

–0.0029(0.0022)

–0.0088*(0.0020)

–0.0026(0.0019)

–––

–––

–––

–––

2.7477*(0.000027)

–0.2417*(0.0000028)

—

—

—

—

—

0.6507*(0.0000057)

0.2074*(0.0000046)

—

0.1751*(0.0000053)


Note: * indicates that the coeffi cient is signifi cantly different from zero at the 5 percent level.


412

coeffi cient. The only two temptation vari-ables that were positive and signifi cant were those associated with one– and two–owner fi rms, with the former also varying by gross receipts. The coeffi cient on the log of gross receipts was similar for both methods A and B. In each of the two smaller samples used in method B, wages and one of the capital–related expense variables (but not the same one) had a positive coefficient. The logistic functional form also produced coeffi cients that conformed to expectations. The nega-tive sign on the log of gross receipts means that capital income becomes more impor-

tant relative to labor income as fi rm size increases. Under this specifi cation, only the temptation variables associated with one– and two–owner fi rms were positive and signifi cant, and neither varied with gross receipts.

For C corporations, the base specifi -cation also produced coefficients that generally conformed to expectations, although the coeffi cient on one or two of the expense variables was positive in each equation using the log–log functional form (see Table 7). In the log–log func-tional form, the same two temptation vari-ables were positive and signifi cant as for

TABLE 7COMPARISON OF REGRESSION COEFFICIENTS FOR C CORPORATIONS

Adjusted Log–Log Form

Method A

Method B

3+ Owners 10+ Owners Logistic Form

Constant

Log(gross receipts)

Log(wages)

Log(depreciation)

Log(repairs)

Log(rent paid)

Log (interest paid)

Wages/capital account

pt2*incentive

pt3*incentive



2.7227*(0.0694)

0.6464*(0.0054)

–0.0019*(0.0009)

0.0167* (0.0012)

–0.0181*(0.0010)

0.0029*(0.0010)

–0.0323*(0.0009)

—

0.2424*(0.0122)

0.0029*(0.0006)

—

3.1039*(0.1173)

0.6071*(0.0086)

0.0504*(0.0020)

–0.0079*(0.0022)

–0.0122*(0.0015)

0.0105*(0.0015)

–0.0360*(0.0014)

—

—

—

—

2.3768*(0.2132)

0.6753*(0.0162)

0.0577*(0.0053)

–0.0433*(0.0047)

0.0179*(0.0030)

0.0086*(0.0029)

–0.0472(0.0031)

—

—

—

—

6.2803*(0.000056)

–0.3862*(0.0000034)

—

—

—

—

—

1.0060*(0.000011)

0.4590*(0.0000060)

—

—

0.0039*(0.0000005)


Note: * indicates that the coeffi cient is signifi cantly different from zero at the 5 percent level.


413

S corporations. In the logistic functional form, however, a different temptation variable was signifi cant. In that case, the statistical distinction between one– and two–owner fi rms found for S corporations was not present (although both could still be distinguished from those with three or more owners). However, additional underreporting among fi rms with six or fewer owners could be detected statisti-cally, and the level of underreporting among such fi rms was found to vary with gross receipts.

The sensitivity tests for S corporations demonstrate that the results are some-what sensitive to the functional form, the imputation method, and the assumption concerning the misreporting of offi cers’ compensation as wages (see Table 8). Using the base specifi cation, labor’s share of operating business income would be more than 4.3 percentage points lower under the assumption that any reported wages suffi cient to cover estimated offi -cers’ compensation represents misreport-ing. Method B produces estimates that are between 2.3 and 4.7 percentage points lower than does Method A. Finally, the logistic regression yields results between 2.6 and 4.3 percentage points lower than the adjusted log–log regression. None of those differences represents more than a ten percent deviation from the base specifi cation.

The sensitivity tests for C corporations, however, demonstrated much more uncer-tainty around the results. The difference between the upper– and lower–bound estimates using the base specifi cation was 6.6 percentage points instead of 4.3, either because C corporations are more likely to misreport offi cers’ compensation as wages or because C corporations with 75 or fewer owners have larger payrolls than S corpo-rations. The results for the two imputation methods also differ by more than they do for S corporations—from 7.7 percentage points lower to 14.7 percentage points higher. Finally, the logistic functional form produces estimates that are between 9.3 and 12.8 percentage points lower than the adjusted log–log functional form. This uncertainty surrounding the C cor-poration results reinforced our decision to use only the S corporation coeffi cients to estimate labor income for partnerships and sole proprietorships.

TRUE LABOR AND CAPITAL SHARES BY ENTITY TYPE

Table 9 presents reported labor income and our estimates of true labor income as percentages of operating business income by entity type for fi rms with positive oper-ating business income. The “reported” figures represent the shares currently subject to employment taxes; the “true”

TABLE 8SENSITIVITY OF SIMULATED LABOR SHARE OF OPERATING BUSINESS INCOME TO FUNCTIONAL FORM, IMPUTATION METHOD, AND MISREPORTING ASSUMPTION

(Percent)

ReportedAdjusted Log–Log Form Method A (use “temptation” variable) Method B (use fi rms with x or more owners) x = 3 x = 10 Logistic Form

56.1

52.751.453.5

60.4

58.157.156.1

88.3

82.6103.079.0

94.9

87.2108.182.1

S corporations C corporations

Lower bound

Lower bound

Upper bound

Upper bound

37.2 58.3


Note: Percentages are calculated using all returns between 2000 and 2004, expressed in 2002 dollars.


414

TABLE 9REPORTED AND TRUE LABOR INCOME AS A SHARE OF OPERATING BUSINESS INCOME:

FIRMS WITH POSITIVE BUSINESS INCOME(Percent)

Sole ProprietorsForm 1065 Filers General Partnerships Limited Partnerships Limited Liability Companies (LLCs) Limited Liability Partnerships (LLPs)S CorporationsC Corporations (<75 Shareholders) 35% tax rate or above 25% tax rate 15% tax rate No taxable income

Reported1

100.0 46.1 55.3 7.3 44.2105.2 45.0 76.5 41.7 81.4 90.7102.3

True2

78.5 35.1 35.2 23.7 47.9 24.0 69.0 87.6 44.7 91.5107.2118.3

Source: Authors’ calculations based on Statistics of Income tax return samples.1Reported labor income = amount included in FICA or SECA tax bases.2True = reasonable compensation for service rendered.

Note: Percentages are calculated using all returns between 2000 and 2004, expressed in 2002 dollars.

0

50000

100000

150000

200000

250000

300000

350000

CCor

pora

tions

(<75

Share

holde

rs) .

SCor

pora

tions

Form

106

5 File

rs

SolePro

priet

orsh

ips

Entity Type

$Millions

Reported Labor Income

Reported Capital Income

Simulated Labor Income

Simulated Capital Income

Figure 1. Capital and Labor Income of Firms with Positive Business Income by Entity Type


415

fi gures represent our estimates of reason-able compensation as a share of operating business income. (Note that these fi gures do not match those in Table 8 because that table was limited to returns with posi-tive capital income. This table includes returns with negative capital income as long as it does not totally offset labor income.) Figure 1 presents the results graphically for the four major entity types.

The results show that over 22 percent of sole proprietors’ income that is sub-ject to SECA tax is actually income from capital. In the case of general partners, capital income makes up 36 percent of

income subject to SECA tax (see Figure 2). Limited partners, in contrast, are subject to SECA tax on only 31 percent of their labor income. The reported and simulated shares of labor income appear to be quite similar for LLC members, but that might just be an artifact of offsetting errors, with those that have adopted general partner-ship rules paying tax on some capital income and those that have adopted limited partnership rules escaping tax on some labor income. If so, the results imply that more LLCs have adopted limited partnership rules for SECA tax purposes than have adopted general partnership rules.

-10000

0

10000

20000

30000

40000

50000

60000

70000

80000

$Millions

Gener

alPar

tner

ships

Limite

dPar

tner

ships

LLCs

LLPs





Entity Type

Figure 2. Capital and Labor Income of Form 1065 Filers with Positive Business Income by Entity Type


416

Owners of corporations are paying FICA taxes on signifi cantly less than their true labor income. In the case of S corpo-ration owners, they are failing to report approximately 35 percent of their labor income. In the case of owners of C corpo-rations with fewer than 75 shareholders, the underreporting is only 13 percent. The difference between S and C corpora-tions is best understood in terms of the different incentives they face. The table breaks down C corporations by marginal tax rate to illustrate how the results vary among them. In the 35 percent bracket and above, the incentive is to overreport labor income. We detected such overreporting ($2.1 billion out of $28.3 billion reported),

but it was more than offset by amounts imputed to returns reporting no offi cers’ compensation. Thus, labor ’s share of the total is similar in both scenarios (see Figure 3). Firms in the 15 percent bracket and below, in contrast, have an incentive to underreport labor and our simulation shows that their labor income is imputed to be almost 18 percent higher than they report.

Taking all entity types together, the simulated amount of true labor income ($643.4 billion) is only 2.2 percent more than the amount reported ($629.3 bil-lion). In other words, despite the large discrepancies by entity type, income from fi rms with positive operating business

-20000

-10000

0

10000

20000

30000

40000

50000

60000

70000

80000

$Millions

35%

brac

ket

25%

brac

ket

15%

brac

ket

Nota

xable

incom

e

Entity Type





Figure 3. Capital and Labor Income of C Corporations with Positive Business Income and Fewer than 75 Shareholders by Entity Type


417

income that is subject to the employment tax roughly equals the aggregate reason-able compensation of their owners. For these fi rms, therefore, the mismatch of the employment tax base and labor income can largely be characterized as one of effi ciency and horizontal equity, while the effect on revenue is a comparatively minor issue.

But this is only part of the story. Corpo-rations with negative operating business income report very little labor income (see Table 3). Partnerships report even less. But unless one believes that the value of

an owner’s labor can be negative, some positive amount of labor income should be attributed to these fi rms.

By estimating regressions on C and S corporations with negative operating business income, we were able to apply the same simulation method used above to generate estimates of reasonable com-pensation for the owners of fi rms with negative operating business income (see Figure 4). Our estimate of labor income from such fi rms ($165.0 billion) is over six times the amount reported ($25.7 billion). Table 10 shows the differences

-200000

-150000

-100000

-50000

0

50000

100000

.

$Millions

Entity Type

SolePro

priet

orsh

ips

Form

106

5 File

rs

SCor

pora

tions

CCor

pora

tions

(<75

Share

holde

rs)





Figure 4. Capital and Labor Income of Firms with Negative Business Income by Entity Type


418

between reported and true labor income by entity type. Positive numbers represent capital income that is currently subject to employment taxes, and negative numbers represent labor income that is currently escaping employment taxes. Combining profitable and unprofitable firms, the total amount of labor income of busi-ness owners (not including C corpora-tions with more than 75 shareholders) included in the employment tax base is lower than true labor income by $153.4 billion.

CONCLUSION

We have shown, using standardized measures of labor and capital income across different forms of organization, that a modest amount of capital income is subject to employment taxes, a larger amount of labor income escapes employ-ment taxes, and a much smaller amount of labor income is subject to corporate income taxes. We have also shown an additional reason why economic inci-dence of taxes is different from statutory incidence, namely that different entity types are taxed on their labor and capital income in different ways, have different reporting requirements and, therefore,

have differing incentives for taxpayers to mischaracterize their income with the intention of reducing liability.

We think these results will be robust to future research, but we will be the fi rst to admit that there are serious limitations imposed by currently available data. Most serious is that we have no information on hours worked or the nature of services being provided by owners and partners. Those data are not collected on tax forms, and sources that contain such data do not link it to net operating income at the entity level.

Furthermore, other potentially useful data are either not collected on tax forms or are collected but not made available in electronic form—specifi cally data that would allow us to trace income flows between entities and individual taxpay-ers, and identify the taxpayers form of participation in entities. For instance, an individual who is both general partner of a partnership and also owner of an S corporation might have the S corporation own capital and rent it to the partnership. Thus, by having no capital in the partner-ship, the taxpayer would avoid having capital income taxed as labor income. Unfortunately, by redirecting capital income but not labor income to the S cor-

TABLE 10DIFFERENCE BETWEEN REPORTED AND TRUE LABOR INCOME BY ENTITY TYPE

(Millions of 2002 Dollars)

Sole ProprietorsForm 1065 Filers General Partnerships Limited Partnerships Limited Liability Companies (LLCs) Limited Liability Partnerships (LLPs)S CorporationsC Corporations (<75 Shareholders) 35% tax rate or above 25% tax rate 15% tax rate No taxable incomeTotal

$57,326$22,149$9,261

–$8,629–$2,712$24,229

–$74,742–$18,768–$1,855

–$917–$6,451–$9,545

–$14,034

–$44,366–$35,600–$2,755–$3,937

–$28,223–$684

–$25,161–$34,282–$1,493

–$76–$1,016

–$31,697–$139,410

$12,960–$13,451

$6,506–$12,566–$30.935$23,545

–$99,903–$53,050–$3,348

–$993–$7,467

–$41,242–$153,444

Firms w/ Positive Business Income

(millions)

Firms w/ Negative Business Income

(millions)All Firms (millions)



419

poration, such a transaction undermines one of the premises of our paper—namely that we can exploit information about S corporation behavior to make inferences about the labor income of partnerships. With the ability to identify the general partners of the partnership and link them to the specifi c S corporation to which it pays rent, we could have removed the S corporation from the sample we used to derive the imputation parameters, thereby avoiding the distortion introduced by the transaction. Such an effort, however, will have to await the day when the necessary data are available.

Another avenue of future research would be to investigate the marginal individual income tax and employment tax rates on the labor and capital income of shareholders, partners, and sole propri-etors. Such an effort, however, would be impeded by the unavailability of certain information collected on Schedules K–1, but not currently transcribed in electronic form—for example, whether one is a gen-eral or limited partner. We hope that, in the future, such data limitations will ease.

Acknowledgments

The views expressed in this paper are those of the authors and should not be interpreted as those of the Congressional Budget Offi ce, the Joint Committee on Taxation or its staff. Laurie Coady, Cecily Rock, Tom Barthold, and Bill Randolph provided insightful guidance and com-ments.

REFERENCES

Fleischer, Victor. “Two and Twenty: Taxing Partnership

Profi ts in Private Equity Funds.” New York University Law Review 83 No. 1 (2008).

Gordon, Roger H., Laura Kalambokidis, and Joel Slemrod. “Do We Now Collect Any Revenue from

Taxing Capital Income?” Journal of Public

Economics 88 No. 5 (April, 2004): 981–1009.

Gravelle, Jane. The Economic Effects of Taxing Capital Income.

Cambridge, MA and London: The MIT Press, 1994.

Joint Committee on Taxation. Options to Improve Tax Compliance and Reduce

Tax Expenditures. JCS–02–05, Washington, D.C., 2005.

Joint Committee on Taxation. Present Law and Analysis Relating to Tax

Treatment of Partnership Carried Interests. JCX–41–07, Washington, D.C., 2007.

Kotlikoff, Laurence J., and Lawrence H. Summers. “Tax Incidence.” In Handbook of Public Eco-

nomics, Volume 2, edited by Alan J. Auerbach and Martin Feldstein, 1043–88. New York: Elsevier Science Publishers BV, 1987.

Mackie, James B. III. “Unfi nished Business of the 1986 Tax Re-

form Act: An Effective Tax Rate Analysis of Current Issues in the Taxation of Capital Income.” National Tax Journal 55 No. 2 (June, 2002): 293–337.

U.S. Congressional Budget Offi ce. Taxing Capital Income: Effective Tax Rates and

Approaches to Reform. Washington, D.C., 2005.

Villalonga, Belen, and Raphael H. Amit. “How Are U.S. Family Firms Controlled?”

Review of Financial Studies (August, 2008). http://rfs.oxfordjournals.org/cgi/content/full/hhn080v#1top.

Warsh, Kevin. “Corporate Cash Balances and Economic

Activity.” Speech given at the American Enterprise Institute, July 18, 2006. www.federalreserve.gov/newsevents/speech/warsh20060718a.htm.

Weisbach, David. “The Taxation of Carried Interests in Private

Equity.” Virginia Law Review 94 No. 3 (May, 2008): 715–64.

Winchester, Richard. “Working For Free: It Ought To Be Against

the (Tax) Law.” Mississippi Law Journal 76 (Fall, 2006): 227–95.

taxation of capital and labor: the diverse landscape by entity · pdf file ·...

Documents