Public Finance Review - CiteSeerX

Public Finance Review http://pfr.sagepub.com

Do Limited Liability Companies Explain Declining State Corporate Tax Revenues? William F. Fox and LeAnn Luna Public Finance Review 2005; 33; 690 DOI: 10.1177/1091142105279333 The online version of this article can be found at: http://pfr.sagepub.com/cgi/content/abstract/33/6/690

Published by: http://www.sagepublications.com

Additional services and information for Public Finance Review can be found at: Email Alerts: http://pfr.sagepub.com/cgi/alerts Subscriptions: http://pfr.sagepub.com/subscriptions Reprints: http://www.sagepub.com/journalsReprints.nav Permissions: http://www.sagepub.com/journalsPermissions.nav Citations http://pfr.sagepub.com/cgi/content/refs/33/6/690

Downloaded from http://pfr.sagepub.com by Roberto Hernandez Sampieri on October 17, 2008

PUBLIC 10.1177/1091142105279333 Fox, FINANCE Luna / DO REVIEW LLCS EXPLAIN DECLINING CORPORATE TAX REVENUES?

DO LIMITED LIABILITY COMPANIES EXPLAIN DECLINING STATE CORPORATE TAX REVENUES? WILLIAM F. FOX LEANN LUNA University of Tennessee

The effective state corporate tax rate fell significantly during the past fifteen years despite the very robust growth in corporate book profits. This article examines the causes of the decline with a focus on the effects of the relatively new option of forming limited liability companies (LLCs), state tax policy, and changes in the federal base. The effects are estimated using a simultaneous equation model and a twelve-year panel for U.S. states. The results confirm that the advent and growth of LLCs have been important causes of the decline in corporate tax revenues. In addition, changes in the federal corporate tax base, the propensity of states to grant tax incentives, and the failure of states to require combined reporting have been significant factors in falling corporate tax revenues. Keywords: state income tax; corporate; tax planning; organizational form

1. INTRODUCTION

State corporate income taxes have declined dramatically as a component of state tax systems during the past fifteen years, despite the generally strong economy through 2000. In 2002, corporate income taxes raised only $31.2 billion, the lowest total since 1994. State corporate taxes as a percentage of corporate profits fell from 7.1 percent in 1989 to 4.6 percent in 2000 (see Figure 1). Corporate taxes as a percentage of total taxes also declined, falling from 7.1 percent in 1989 to 5.9 percent in 2002. The 2002 corporate tax share was the lowest since 1960. Although corporate taxes represent a relatively small share of total collections, the declines in nominal collections and in the percentage of total collections, even with generally strong corporate profAUTHORS’ NOTE: The authors wish to thank Donald Bruce, Mathew Murray, and an anonymous referee for very helpful comments on an earlier draft of this article. PUBLIC FINANCE REVIEW, Vol. 33 No. 6, November 2005 690-720 DOI: 10.1177/1091142105279333 © 2005 Sage Publications

690


Fox, Luna / DO LLCS EXPLAIN DECLINING CORPORATE TAX REVENUES?

691

10 9

Percent

8 7 6 5 4 3 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 00 Year

Figure 1: State Corporate Profits Tax as a Percentage of Corporate Profits, 19852000

its, have attracted the attention of states desperate for revenue (Cornia et al. 2005). Recent corporate tax reforms in states such as Tennessee, New Jersey, Kentucky, and Maryland evidence the seriousness with which states are approaching reforms in corporate taxation. Ten states raised their nominal corporate income tax rates and ten decreased their rates during this window, suggesting that erosion of the corporate income tax base relative to measured profits must explain most of the decline in corporate taxes. If the corporate income tax is to remain a viable revenue source, policy makers must address the continued erosion of corporate income taxes and make changes necessary either to slow or reverse the erosion of the tax base or perhaps to eliminate the corporate income tax and replace it with another type of tax. Four general factors appear to offer explanations for state corporate tax base erosion (Fox and Luna 2002). First, declines in corporate profits may have been a recent factor, beginning around 2000. Second, changes in the federal tax base affect state income taxes because all but two states begin their taxable income calculation with federal taxable income or a variant. Both changes in federal policy and more aggressive corporate tax planning potentially offer reasons for a shrinking federal tax base. Corporate inversions, liberalization of depreciation rules, and treatment of stock options are examples of fac-


692

PUBLIC FINANCE REVIEW

tors that affect state tax bases indirectly by reducing the federal base (Desai 2002). Third, states have made policy decisions that either purposely or inadvertently eroded their tax bases. These policy actions include negotiated tax concessions; tax credits for employment, investment, and other purposes; and changes in the apportionment formula. Finally, corporations have become much more aggressive in their attempts to minimize state income taxes through tax planning strategies. Firms have formed multiple corporations to separate different functions and have planned their siting of these separate entities across states to take advantage of differences in state tax laws. The combined effects of these factors may have been dramatic.1 The introduction and proliferation of limited liability companies (LLCs), which occurred during this same time period, provided multistate businesses with a new form of pass-through entity that facilitates some tax avoidance opportunities and may be a significant contributor to tax base erosion. States were initially slow to react to the rapid growth in LLCs and did not anticipate the ease with which businesses were able to shift operations from C Corporations fully taxed by a state to LLCs that sometimes avoided state income taxes on the entity or its members. This study should assist legislators and policy makers to determine whether various LLC provisions and other state policies have resulted in a permanent reduction in state tax revenues and, if so, should assist in the design of tax policies that will offset any unintended consequences. Recent research on causes of corporate tax declines is sparse, and few studies have sought to estimate the state revenue losses directly attributable to the proliferation of LLCs. The revenue estimates that do exist appear to understate the effects since they were performed before the explosion in popularity of LLCs.2 Also, the prior analyses are not placed into an econometric structure where the full extent of revenue changes can be measured. This study uses panel data to examine the extent to which LLCs and other state tax attributes have affected state corporate tax revenues. Because a broader corporate tax base may also encourage a business to form an LLC, we structure the model to account for the simultaneity that exists between the extent of corporate taxation and the decision to form an LLC. The results show that the



693

advent of LLCs, changes in the federal base, and other state policy actions have all contributed to the decline in corporate tax revenues. The article is divided into five sections following this introduction. The next section is a description of the growth of LLCs and provides examples of tax planning strategies in several states that help explain why this organizational form has been chosen. The next two sections describe the model specification and hypotheses and data sources. The final sections report the results and conclusions and limitations of the study.

2. GROWTH OF LLCS

LLCs are created pursuant to state statutes and provide businesses with a relatively new set of organizational options. Most state statutes became effective during the 1990s, and as of July 1, 1996, all fifty states and the District of Columbia had enacted LLC statutes.3 The data we use in this study indicate that the number of LLCs increased from a negligible amount in 1990 to about 2 percent of businesses in the mid-1990s and to more than 5 percent of all businesses in 2000 (see Figure 2). LLCs represented up to 10 percent of businesses in some states by 2000. The large growth in LLCs occurred primarily because the entity combines the attractive features of both partnerships and corporations. LLC members (i.e., owners) receive limited liability that was once only afforded to corporate shareholders or for nonmanager limited partners. Members may also gain the flexibility and tax advantages of a partnership (e.g., lack of a double tax on distributed earnings and pass through of losses) by making the election to be taxed as a conduit entity. The LLC is generally superior to an S Corporation because it is more flexible and there are no limits on the type or number of shareholders. LLCs can therefore be entirely owned by corporations and be members of a consolidated group. 4 The effects of LLCs on state tax revenues could depend on both the characteristics of LLCs that are prescribed by state law and the state’s overall tax structure. Two examples are identified here to demonstrate


694


6

5

Percent

4

3

2

1

0 88

89

90

91

92

93

94

95

96

97

98

99

00

Figure 2: Percentage of Businesses Selecting the Limited Liability Company (LLC) Form, 1988-2000

how multistate firms use LLCs to shift income between states to reduce tax liabilities.5 One opportunity for tax planning exists because many states do not tax the LLC entity but only tax the members on their distributive shares. There is at least some doubt regarding whether the state can tax the member’s distributive share if the member does not otherwise have nexus in the state where the LLC is organized and, even if nexus exists, whether the tax revenues will be collected (Fay and Amitay 2001; Haas 2003; Lowy and Vasquez 2003).6 In these cases, the LLC may allow the potential for multistate businesses to shift income. For example, in Tennessee, prior to imposition of the corporate taxes on LLCs, a common tax planning practice among multistate businesses was to form a Tennessee LLC that included all Tennessee operations. A Tennessee corporation would own 1 percent of the LLC, and the remaining 99 percent would be owned by a Delaware corporation. Delaware does not tax the ownership of intangibles, and the LLC ownership interest is an intangible asset under Delaware law. Because



695

Tennessee only taxed corporations, the technique effectively removed 99 percent of the Tennessee income from the tax rolls. Kentucky provides a second example of how LLCs can be used to lower corporate tax revenues. Kentucky businesses take advantage of the fact that the state requires LLCs to use a single factor sales apportionment formula.7 Since the relatively high payroll and property of a manufacturing operation are ignored for Kentucky LLCs, the effect is most often to reduce the state income taxes if the business has sales sitused in states with a three-factor formula. Taxes are additionally reduced because Kentucky levies a license tax (based on company value) on corporations but not on LLCs. Discussions with both Tennessee and Kentucky government officials indicate that large multistate corporations are using LLCs as one avoidance scheme. Furthermore, anecdotal evidence suggests that avoidance mechanisms of these types can have a dramatic effect on revenues, as evidenced by the pattern of Tennessee franchise and excise tax collections. During the economic boom years of 1997 to 1999, these revenues barely moved, actually decreasing from $893.2 to $891.2 million.8 Franchise and excise tax revenues rebounded to $1,197.8 million by 2004, as corporate tax revenues were declining in many other states, after Tennessee began taxing LLCs under the corporate tax structure.

3. RESEARCH DESIGN

We use a two-equation panel model to examine how LLCs and other state policies have affected state corporate tax receipts. The first equation accounts for the decision of whether to organize as an LLC. The second equation explains state corporate tax revenues as a function of the tax rate, economic factors, and the other causes of corporate tax erosion described above. While we are primarily interested in the second equation, we include the first equation because of the potential for simultaneous equation bias. The extent of corporate taxation may affect whether a firm chooses to form an LLC. The broader the corporate tax base, the greater the probability the firm will choose an LLC.


696


The Hausman specification test shows that simultaneity exists between corporate revenues and the share of businesses choosing the LLC, and failure to account for this simultaneity could result in biased and inconsistent parameter estimates in the corporate tax equation. We estimate the equations using two-stage least squares (2SLS), both with and without fixed effects. The next section provides a detailed description of the equation structure, starting with the share of LLCs equation and followed by the corporate tax revenue equation. We employ the following model: SHAREit = b0 + b1EFFECTIVEit + b2SMLLCit + b3FLEXIBLEit + b4TRANSFERit + b5RATE_DIFFit × EFFECTIVEit + b6∆REVENUEit + b7∆W/H_TAXit + b8BACHELORit + b9∆EMPLOYit + b10MANUFACit + b11SERVICEit + εit.

(1)

REVENUEit = δ0 + δ1CORP_RATEit + δ2SHAREit + δ3REQUIREDit + δ4VOLUNTARYit + δ5SALESit + δ6TAXINCENit + δ7NONTAXINCENit + δ8FEDBASEit + δ9∆PIit + νit.

(2)

3.1. SHARE OF LLCS EQUATION

A large body of literature discusses organizational form and the attributes and industry classifications that make a firm more likely to select one particular form of business over another.9 Scholes and Wolfson (1992) presented the guiding principal that business owners should seek to maximize after-tax rates of return in the organizational choice decision, but the decision is not based on tax costs alone (Omer, Plesko, and Shelley 2000). Business owners may consider many nontax factors such as the flexibility to operate efficiently, the transferability of interests, personal liability for debts, and other factors. Since this study relies on aggregate state-level data and is only focused on the relative levels of LLCs to other businesses, it does not require a comprehensive analysis of organizational form but instead focuses on the importance of tax and LLC characteristics. Furthermore, data on many of the nontax factors are only available at the firm level and cannot be obtained for states. We estimate the share of businesses selecting the LLC form based on a series of LLC, tax structure, economic, and demographic charac-



697

teristics.10 We include the year in which the LLC statute becomes effective (EFFECTIVE), with a value of 0 in all prior years and 1 for the adoption year and all subsequent years. We also use three specific LLC characteristics that can influence the propensity to select this organizational form in the equation. These are whether single-member LLCs are permitted (SMLLC), whether flexible rather than bulletproof statutes were legislated (FLEXIBLE), and whether unanimous consent is required to transfer a member’s LLC interest to others (TRANSFER). Many states revised their LLC statutes to allow for single-member LLCs (SMLLCs) following the issuance of the federal check-the-box regulations. Only eleven state statutes expressly allowed SMLLCs prior to 1997, but Massachusetts is the only state that does not permit SMLLCs today. The SMLLC option allows corporations a tax planning mechanism because it permits individual companies to create and actively operate LLCs. A positive relationship is anticipated between SMLLC and the share of businesses operating in the LLC form. A dummy variable is used to account for whether a state permits the use of a SMLLC (where 1 equals SMLLC permitted and 0 equals SMLLC not permitted). Some states initially enacted so-called “bulletproof” LLC statutes to ensure that an LLC following state guidelines qualified for both pass-through treatment and limited liability for its owners. The federal check-the-box regulations eliminated the need for the bulletproof statutes because entities no longer had to worry about qualifying for federal pass-through tax treatment. Even if an entity possessed three or four of the corporate characteristics (which previously would have mandated taxation as a corporation), businesses could now achieve pass-through tax treatment with a simple election. States responded to the new federal regulations by adopting flexible LLC statutes that greatly increased the organizational options available to business owners. We use a dummy variable (FLEXIBLE) to reflect the type of state statute in place during each year (where 1 equals a flexible statute and 0 equals a bulletproof statute). The passage of flexible statutes is expected to increase the probability of becoming an LLC. Free transferability of interest exists if a member may substitute another person for himself or herself in the organization without the con-


698


sent of other members. Under many state statutes, assignment of a member’s interest requires the unanimous consent of the other members; some other states require only a majority. The take up rate of LLCs is expected to decline when greater restrictions are placed on transferring interests. We use a dummy variable (TRANSFER) where 1 requires unanimous consent and 0 requires majority or other type of consent. In addition, studies have shown that the choice in organizational form is influenced by the difference between individual and corporate tax rates (Guenther 1992; Terando and Omer 1993) and that a change in corporate marginal tax rates relative to individual rates would motivate changes from C status to S status (Watkinson 1989). Federal corporate tax rates have remained relatively steady through the study period, but state-level tax rates have changed and there is cross-sectional variation across the states. The influence that individual versus corporate tax rates has on selection of the LLC form is investigated (RATE_DIFF), using a variable equal to the maximum state corporate tax rate minus the maximum state individual income tax rate. Higher corporate tax rates relative to individual tax rates should encourage LLC formation. The RATE_DIFF variable is interacted with EFFECTIVE so that the difference in tax rates does not influence the “take up rate” until the LLC became a legal organizational form in the state. We include two other fiscal variables in the SHARE equation. First, the dependent variable in the second equation (REVENUE), or corporate tax revenue as a percentage of personal income, is included as a measure of a state’s overall tax burden on business. This variable is best thought of as a proxy for the breadth of the tax base since the relative corporate rate is held constant in the equation. A high tax structure is expected to cause firms to select the LLC form and choose to be taxed as unincorporated businesses to avoid the corporate tax, suggesting a positive coefficient. Second, whether the state imposes a withholding tax on LLC income transferred to out-of-state members is included. The withholding tax (W/H_TAX) imposes an entity-level tax on the out-of-state owners and makes formation of LLCs less attractive. Of course, members who believe they do not have nexus can file a return and seek a refund. We include fixed effects to account for other differences in states that influence the LLC take up rate.



699

Finally, the equation includes measures of the state economic and demographic environments. The education level (BACHELOR), as measured by percentage of the adult population with a bachelor’s degree, is included (Goolsbee 2002). The change in state employment growth (∆EMPLOY) is used because the capacity of LLCs to rise as a share of total firms depends on the rate at which new firms are being created. The influence of industry structure on the formation of LLCs is accounted for using the levels of employment in the manufacturing (MANUFAC) and the service (SERVICE) industries. 3.2. CORPORATE TAX REVENUE EQUATION

Arithmetically, corporate tax revenues equal the tax base times the corporate tax rate. Our approach combines this arithmetic relationship with the behavioral determinants of the tax base. Thus, the revenue equation includes the corporate tax rate and the key factors that influence the corporate tax base within a state. State corporate tax revenues as a percentage of personal income for every state i for every year t from 1988 through 2000 are used as the dependent variable in the corporate tax revenue equation. Specifying revenue as a percentage of personal income accounts for the differential scale of state economies and should reduce problems of heteroskedasticity.11 States with higher tax rates (CORP_RATE) are expected to generate higher tax revenues, meaning a positive relationship is expected between the corporate tax rate and revenues. We use the maximum corporate tax rate for each state for each year. We include variables to proxy for the factors that should explain erosion of the corporate tax base. First, corporations have become adept at minimizing corporate income taxes by forming multiple entities, both within and across state lines. LLCs are one means for corporations to engage in tax avoidance. We use the proportion of businesses choosing the LLC structure to measure the presence of LLCs in the state (SHARE), where the value is 0 for each state until a statute was passed authorizing the formation of LLCs.12 As described above, we expect that the formation of LLCs reduces state corporate tax revenues because of the way LLCs are treated within state tax


700


structures and because of tax avoidance opportunities available through the LLC form. Also, new businesses may form as an LLC simply to ensure that early losses are passed through to owners rather than being bottled up in a corporation as net operating losses, and this could lower corporate taxes further. Separate reporting allows easy avoidance opportunities through transfer pricing and passive investment companies (Luna 2004). Only fourteen states require related corporations engaged in a unitary business to file a combined report. Combined reporting, therefore, is one potential means for reducing the effectiveness of separate corporate structures as an avoidance mechanism (Pomp 1998). We include a dummy variable for states that require combined reporting (REQUIRED). However, states can only require combination of those corporations that are unitary and, in some states, only ones that individually have nexus in the state. Therefore, combined reporting may not eliminate all avoidance possibilities available through the use of multistate income shifting (Lowy and Vasquez 2003). On the other hand, voluntary combined reporting allows firms to select the group of companies with which to combine for tax purposes. We expect tax revenues to be lower in the fifteen states that allow voluntary combined reporting (VOLUNTARY), though in some cases the provisions allowing for voluntary reporting may be too restrictive to have a significant effect on revenue. State policy decisions have also contributed to the decline in corporate revenues. Examples include the continued granting of tax concessions and changes in the apportionment rules to favor instate production. Until relatively recently, most states followed the Uniform Distribution of Income for Tax Purposes Act (UDITPA) guidelines and apportioned multistate income using equally weighted sales, property, and payroll factors (Edmiston 2002). During the past several decades, many states have increased the sales factors’ weighting as one form of tax concession to encourage in-state production. Indeed, several states, including Iowa, Nebraska, and Texas, have moved to a single factor sales apportionment formula for at least some industries. Firm-specific effects of increased weighting of the sales factor depend on the relationship between the sales factor and the property and payroll factors (Gordon and Wilson 1986; Pomp 1998; Goolsbee and



701

Maydew 2000). In principle, states that are net importers of corporate tax outputs have incentives to increase weighting on the sales factor since their revenues should rise, and states that are net exporters have revenue incentives to decrease the sales factor to increase their tax revenues (Anand and Sansing 2000). Of course, states are considering other goals, such as employment expansion, when decisions on the corporate tax structure and weighting of the factors are made. As a general rule, the political expectation is that aggregate state tax revenues would fall to some extent with greater emphasis on the sales factor, since changes in the weighting were pushed by business as a means to lower their tax burden (Klassen and Shackelford 1998). SALES is a continuous variable, ranging from 0 to 1, and represents the sales factor weighting. A higher weighting is expected to have a negative relation with REVENUE.13 States offer tax and nontax incentives to attract new business. These incentives range from ones targeted on only a small number (perhaps one) of large corporations to ones that are in the tax code and are applicable to many corporations. The dollar amounts of the concession packages in some cases have grown to exceed hundreds of millions of dollars in recent years but in other cases can be offered at relatively low cost. States offer both tax and nontax incentives to help entice firms to locate or expand in their state. Tax concessions include investment tax credits, property tax abatements, and employment tax credits. Financial or nontax incentives include state loans, grants, and bonds. We include in the model the number of tax and nontax incentives offered by each state to control for state policies that may influence corporate tax revenues. On average, states gave approximately nine tax and/or nontax incentives each year. In 2000, Connecticut and New York offered the largest number of incentives, granting thirty tax incentives and forty nontax incentives, respectively. Data are not available on the negotiated concessions granted to individual firms. Tax incentives have two main effects on corporate tax revenues. First, they are expected to have a small but positive effect on economic activity and therefore should have a positive effect on corporate tax revenues (Wasylenko 1997). Second, tax concessions will have a direct negative effect on corporate tax revenues since by design they are


702


intended to lower the tax revenues that will be paid from a given amount of corporate profit. The tax savings to firms that would have located in a state even without the concession (since states are unable to discern which firms are influenced by the concessions) will result in some loss of tax revenue. This study does not attempt to separate these opposing effects; therefore, the sign on TAXINCEN is not predicted in advance. Nontax incentives should have a positive, albeit small, effect on economic activity, and therefore a positive relation is expected between NONTAXINCEN and REVENUE. Federal tax policy is another potential source of state tax base erosion since the starting point for state taxable income is generally a variant of federal taxable income. FEDBASE, calculated as IRS corporate tax collections in year t in state i divided by the maximum federal corporate tax rate, is used to account for this effect. IRS collections by state are a residence-based measure, drawn from the filing state of each corporation, and are only a proxy for the apportionable income into a state. This variable is intended to account for other factors affecting tax revenues in the 1990s, such as accelerated tax depreciation, increased use of stock options, corporate inversions, transfer pricing schemes, and the ability to locate U.S. profits offshore, as well as other corporate tax sheltering. Any federal corporate tax planning will automatically reduce corporate state income tax revenue without state legislation. As a result, many states have recently decoupled from aspects of federal law, particularly as it relates to bonus depreciation and the Internal Revenue Code (IRC) Section 179 immediate expensing election. The final source of the deterioration in state corporate income tax revenues is cyclical declines in profits. We cannot adequately capture the slowing trend of corporate profits because no data are available on state specific corporate profits. The growth rate in state personal income is included to allow for the response of revenues to the state-specific position in the business cycle (∆PI).14 Estimates are also provided using state-fixed effects that account for differences in economic structure and other unmeasured state factors (such as policy idiosyncrasies) that influence corporate tax revenues.



703

4. DATA

Lexus/Nexus provided data on the number of LLCs and the number of total businesses for all fifty states for the period from 1988 to 2000. Furthermore, every state was contacted directly in an effort to replace missing data.15 In the end, 107 of the potential 650 observations for the SHARE variable were missing, and these observations were excluded from the analysis. Missing values arose either because the number of LLCs was not known, the number of total businesses was not known, or both. We obtained the LLC characteristics EFFECTIVE, SMLLC, FLEXIBLE, and TRANSFER from the Commerce Clearing House annual publication, “Comparison of LLC Acts.” We obtained tax revenue data from the Census of Governments and tax rate and tax structure data from a variety of different sources.16 Table 1 provides descriptive statistics. Table 2 summarizes the number of states that adopted LLCs and the various LLC characteristics. The data changed significantly over the thirteen-year time series of the data. Only two states had effective LLC legislation at the beginning of the 1990s, but all states had effective LLC legislation by 2000. The mean percentage of firms operating as a LLC across the entire time period was 1.6 percent, but the percentage was essentially 0 until 1992 and had grown to 5.1 percent in 2000 (see Figure 1). Also, in 2000, forty-nine states allowed single-member LLCs, fifty states had flexible statutes, and forty-one states required unanimous agreement for a member’s interest in an LLC to be transferred. Corporate tax revenues as a percentage of personal income averaged 0.43 percent over the thirteen-year time period. The maximum share excluding Alaska was 1.17 percent in Michigan. The mean corporate income tax rate remained relatively constant during the time period, averaging between 6.6 and 7.1 percent. The mean weight on the sales factor in the corporate income tax apportionment formula was 0.36 in 1988 and rose to 0.46 in 2000.

5. RESULTS

We estimate the baseline results using a balanced panel, meaning the same time series for each state (except for missing values), for the


704


TABLE 1:

Descriptive Statistics

Variable

Mean

BACHELOR ∆EMPLOY ∆PI CORP_RATE EFFECTIVE FEDBASE FLEXIBLE MANUFAC NONTAXINCEN RATE_DIFF REQUIRED REVENUE SALES SERVICE SHARE SMLLC TAXINCEN TRANSFER VOLUNTARY W/H_TAX

22.199 1.900 5.843 7.430 0.605 80.050 0.566 16.117 8.823 0.970 0.184 4.347 0.477 26.828 0.018 0.334 9.743 0.533 0.199 0.115

Standard Deviation 4.657 1.790 2.033 2.187 0.489 118.115 0.496 5.478 5.346 3.110 0.387 1.807 0.500 3.784 0.027 0.472 3.712 0.499 0.400 0.320

Minimum 11.100 –7.000 –7.000 2.000 0.000 1.092 0.000 3.000 1.000 –5.000 0.000 0.658 0.000 16.000 0.000 0.000 1.000 0.000 0.000 0.000

Maximum 38.700 7.000 11.000 12.000 1.000 1,838.440 1.000 29.000 41.000 12.000 1.000 11.708 1.000 37.000 0.222 1.000 30.000 1.000 1.000 1.000

NOTE: Data exclude Alaska and states that do not impose a corporate income tax (Nevada, South Dakota, Texas, Washington, and Wyoming). BACHELOR = percentage of population with a bachelor degree in state i for time t. ∆EMPLOY = percentage change in employment levels in state i for time t. ∆PI = percentage change in personal income in state i for time t. CORP_RATE = corporate tax rate in state i for time t. EFFECTIVE = 1 if limited liability company (LLC) legislation effective in state i during time t, else 0. FEDBASE = federal corporate income tax collections in state i for time t divided by the highest federal tax rate for time t divided by average personal income. FLEXIBLE = 1 if flexible statute enacted in state i during time t, else 0. MANUFAC = level of employment in the manufacturing sector in state i for time t. NONTAXINCEN = number of non-tax incentives in state i for time t. RATE_DIFF = highest corporate tax rate minus the highest individual tax rate in state i during time t. REQUIRED = 1 if combined reporting required in state i for time t, else 0. REVENUE = corporate tax revenue as a percentage of personal income in state i for time t. SALES = sales factor weight for apportionment purposes in state i for time t. SERVICE = level of employment in the service sector in state i for time t. SHARE = number of LLCs as a percentage of total businesses in state in state i for time t. SMLLC = 1 if single member LLCs permitted in state i during time t, else 0. TAXINCEN = number of tax incentives in state i for time t. TRANSFER = 1 if unanimous consent required for transfer of ownership interest in state i for time t, else 0. VOLUNTARY = 1 if combined reporting available but not required in state i for time t, else 0. W/ H_TAX = 1 if state imposes withholding on nonresident member income, else 0.

years 1988 through 2000.17 This time series also includes years prior to when most states had LLC legislation. This approach made it possi-



TABLE 2:

Year 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000

705

Number of States with Various Limited Liability Company (LLC) Characteristics, 1988-2000

EFFECTIVE 2 4 4 8 15 31 43 46 49 50 50 50 50

FLEXIBLE 1 2 2 3 10 23 32 39 44 47 50 50 50

SMLLC

TRANSFER

0 0 0 1 1 6 9 10 15 31 48 49 49

4 8 8 8 14 29 38 42 45 43 42 42 41

NOTE: EFFECTIVE = 1 if LLC legislation effective in state i during time t, else 0. FLEXIBLE = 1 if flexible statute enacted in state i during time t, else 0. SMLLC = 1 if singlemember LLCs permitted in state i during time t, else 0. TRANSFER = 1 if unanimous consent required for transfer of ownership interest in state i during time t, else 0.

ble to include in the estimated coefficients the effects arising from the introduction of LLCs. In addition, the balanced panel allowed the data to include cross-sectional variation that exists when some states allow LLCs and some do not. The disadvantage of this approach is concern about the coefficients in the SHARE equation because the dependent variable (SHARE) and some of the right-hand-side variables are constrained to zero before introduction of LLCs in the state. Therefore, as a robustness check, we also run the model using an unbalanced panel, with each state’s time series only including the years during which LLCs were permitted.18 The results are similar to those obtained using the balanced panel. Results are presented without fixed effects, with state-fixed effects only, and with both state- and time-fixed19 effects.20 Again, the results are generally similar, but the coefficients are often smaller when fixed effects are included. We use the balanced panel with state- and time-fixed effects as the baseline in the following discussion, but differences with the other specifications are noted throughout.21


706


5.1. SHARE OF LLCS EQUATION

Table 3 presents the results of the SHARE equation used to estimate the LLC selection decision. The findings are generally consistent with expectations, though the significance levels are often low. The effective date when firms were permitted by state legislation to organize LLCs has a strong, statistically significant effect on the proportion of firms that chose to organize as an LLC but only without timefixed effects. As expected, limitations on the ability to transfer interests in LLCs (TRANSFER) have a negative effect on selection of this organizational form, and FLEXIBLE has the expected positive coefficient; neither coefficient is statistically significant. SMLLC is positive and highly significant without time-fixed effects. The ability to create single-member LLCs may be important to many businesses (particularly multistate firms) that want to structure their operations in a state as an LLC.22 The relationship between the corporate and individual income tax rates has an unexpected negative effect on choice of the LLC form, though it is insignificant without time-fixed effects.23 One explanation is that in many cases LLC members are corporations rather than individuals, so that the individual rate is unimportant in the decision. Also, the costs of forming an LLC may be very low relative to the corporate taxes imposed by all taxing states, so that the marginal incentives to engage in avoidance behavior are unaffected by the specific corporate tax rate used in a state. Another explanation is that the various avoidance mechanisms either involve exclusion of LLCs from non-incomebased taxes or the ability to shift income out of state, so the corporate income tax rate would not affect the decision. Furthermore, federal rather than state individual and corporate tax rates may have the greater tax effect on the LLC selection decision.24 As expected, the growth in the number of LLCs has a positive relationship with the percentage of the population with bachelor degrees and the change in employment levels, though the latter is insignificant. In addition, the results support the notion that firms in states with higher levels of employment in manufacturing (MANUFAC) are less likely to form as LLCs. The coefficient for REVENUE is negative but not statistically significant. The withholding tax also has the expected negative sign, but it is statistically insignificant. A withholding tax


707


Two-Stage Least Squares (2SLS) Limited Liability Company (LLC) Share Regressions (standard errors in parentheses)

∆EMPLOY

RATE_DIFF

RATE*EFFECTIVE

TRANSFER

FLEXIBLE

SMLLC

EFFECTIVE

REVENUE

INTERCEPT

Variable

.001 (.001)

–.009 (.014) .002 (.002) .017** (.007) .020*** (.003) .006 (.006) –.003 (.004) –.001 (.001) NA

Without Fixed Effects

–.001 (.001)

.038 (.058) .005 (.006) .004 (.008) .011*** (.004) .008 (.007) –.004 (.005) –.001 (.001) NA

With State-Fixed Effects

Balanced Panel 1988-2000 (n = 425)

–.000 (.001)

.079 (.065) –.001 (.004) .004 (.008) –.003 (.004) .007 (.006) –.002 (.004) –.001* (.001) NA

With Stateand Time-Fixed Effects

–.001 (.001) .002 (.002)

.020*** (.004) .007 (.008) –.004 (.006) NA

–.001 (.028) .001 (.002) NA


–.003 (.003) –.000 (.002)

.009* (.005) .006 (.009) –.010 (.009) NA

–.126 (.097) .011 (.007) NA


Unbalanced Panel (n = 244)

(continued)

–.004 (.003) –.000 (.002)

–.004 (.006) .008 (.008) –.008 (.008) NA

–.015 (.102) .006 (.007) NA


SHAREit = b0 + b1EFFECTIVEit + b2SMLLCit + b3FLEXIBLEit + b4TRANSFERit + b5RATE_DIFFit × EFFECTIVEit + b6REVENUEit + b7W/H_TAXit + b8BACHELORit + b9∆EMPLOYit + b10MANUFACit + b11SERVICEit+ εit

TABLE 3:

708


.001*** (.000) –.001 (.004) –.000 (.000) –.001 (.001) .40

.002*** (.001) .003 (.005) –.003* (.001) .003** (.002) .52

With State-Fixed Effects .001** (.00l) .002 (.005) –.002 (.001) –.001 (.002) .58

With Stateand Time-Fixed Effects .001*** (.001) –.001 (.005) –.000 (.001) –.001 (.001) .13

Without Fixed Effects .002** (.001) .009 (.010) –.002 (.003) .006** (.003) .46



.002** (.001) .005 (.010) –.001 (.002) –.001 (.003) .51


NOTE: SHARE = number of LLCs divided by total business in each state during time t. EFFECTIVE = dummy variable taking a value of 1 for all years LLCs permitted in state i during time t; 0 otherwise. SMLLC = dummy variable taking a value of 1 if state i permits single-member LLCs during time t; 0 otherwise. FLEXIBLE = dummy variable taking a value of 1 if state i has a flexible statute during time t; 0 otherwise. TRANSFER = dummy variable taking a value of 1 if state i requires unanimous consent during time t; 0 otherwise. RATE_DIFF*EFFECTIVE = the difference between the highest individual and corporate tax rate in state i during time t multiplied by EFFECTIVE. REVENUE = corporate tax revenues in state i divided by personal income in state i during time t. W/H_TAX = dummy variable taking a value of 1 if state i imposes a withholding tax on nonresident members during time t; 0 otherwise. BACHELOR = percentage of adult population in state i with a bachelor’s degree during time t. ∆EMPLOY = change in state employment growth in state i during time t. MANUFAC = percentage of employment in the manufacturing industry in state i during time t. SERVICE = percentage of employment in the service industry in state i during time t. *Significant at p < .1. **Significant at p < .05. ***Significant at p < .01.

R2

SERVICES

MANUFAC

W/H TAX

BACHELOR

Variable



SHAREit = b0 + b1EFFECTIVEit + b2SMLLCit + b3FLEXIBLEit + b4TRANSFERit + b5RATE_DIFFit × EFFECTIVEit + b6REVENUEit + b7W/H_TAXit + b8BACHELORit + b9∆EMPLOYit + b10MANUFACit + b11SERVICEit+ εit

TABLE 3 (continued)


709

having no effect on the startup of LLCs is not surprising because firms will file a return and receive a refund if no tax is ultimately due. These latter findings provide additional support for the notion that federal taxes rather than state taxes are the more important tax determinant of organizing as an LLC. 5.2. CORPORATE TAX REVENUE EQUATION

Results from estimating determinants of state corporate tax revenues using 2SLS are provided in Table 4; we also estimate corporate tax revenues using ordinary least squares (OLS) as a robustness check, although we do not report these estimates. Fixed effects, and particularly state-fixed effects, appear to be very important determinants of corporate tax revenues given the large increase in the R2 that accompanies inclusion of fixed effects. Inclusion of time-fixed effects reduces the importance of some variables, which evidences the difficulty of separately identifying the effects that individual factors have on the pattern of revenue since they are often correlated with time. Still, it is important to recognize that time itself did not cause the erosion of the base, so the coefficient estimates without the time-fixed effects should also be considered. Many of the policy and economic variables have the expected effect on corporate tax revenue. The results show that CORP_RATE is positive and significant, though only when both state- and time-fixed effects are included.25,26 The coefficient is rather small, suggesting that rates have relatively little influence on the revenue that is generated. One explanation is that high tax rates may be levied in states that have narrowed the corporate tax base so that the net effect is that little new revenue is generated. Furthermore, this finding supports the notion that businesses are only willing to pay a certain amount of taxes. At the margin, higher rates may be offset by greater use of tax avoidance schemes, resulting in little or no new revenue for the state. In most cases, the variable measuring the proportion of businesses that selects the LLC form (SHARE) is negative and statistically significant, indicating that state corporate tax revenues have fallen with the rise in the proportion of businesses that is an LLC. The preferred 2SLS estimate with time- and state-fixed effects is insignificant,


710


Two-State Least Squares (2SLS) Corporate Income Tax Regressions (standard errors in parentheses)

VOLUNTARY

REQUIRED

SALES

CORP_RATE

SHARE

INTERCEPT

3.401*** (0.559) –32.219*** (6.797) 0.025 (0.043) 0.851*** (0.194) 0.565** (0.291) 0.636** (0.252)

Without Fixed Effects 2.072*** (0.506) –18.852*** (3.900) 0.041 (0.055) 0.268 (0.183) 0.630*** (0.191) 0.136 (0.166)

With StateFixed Effects


1.258* (0.806) –8.621 (10.569) 0.091* (0.048) 0.697* (0.616) 0.464** (0.186) 0.048 (0.173)

With Stateand TimeFixed Effects

4.777*** (0.813) –31.880*** (10.594) –0.027 (0.066) 0.833*** (0.257) 0.858*** (0.329) 0.882*** (0.296)


2.564*** (0.702) –8.941* (4.395) 0.185* (0.107) 0.293 (0.238) 0.801** (0.328) 0.543* (0.282)

With StateFixed Effects


1.594*** (0.748) –3.671 (6.875) 0.239** (0.098) 1.169 (1.219) 0.478 (0.342) 0.425 (0.302)

With Stateand TimeFixed Effects

REVENUEit = δ0 + δ1CORP_RATEit + δ2SHAREit + δ3REQUIREDit + δ4VOLUNTARYit + δ5SALESit + δ6TAXINCENit + δ7NONTAXINCENit + δ8FEDBASEit + δ9∆PIit + νit

TABLE 4:

711


–0.041 (0.028) 0.032* (0.019) 0.051 (0.047) 0.006*** (0.001) .13

–0.051*** (0.018) 0.038** (0.016) 0.111*** (0.024) 0.001 (0.001) .78

–0.046*** (0.016) 0.035** (0.017) 0.061** (0.026) 0.001 (0.001) .83

–0.067** (0.03i) 0.025 (0.020) –0.077 (0.073) 0.006*** (0.001) .16

–0.098*** (0.022) 0.015 (0.024) –0.023 (0.027) 0.000 (0.001) .89

–0.083*** (0.020) 0.008 (0.023) –0.019 (0.026) 0.000 (0.000) .91

NOTE: REVENUE = corporate tax revenues in state i divided by personal income in state i during time t. CORP_RATE = highest corporate tax rate in state i during time t. SHARE = number of limited liability companies (LLCs) divided by total business in each state during time t. REQUIRED = dummy variable taking a value of 1 if state i requires combined reporting during time t; 0 otherwise. VOLUNTARY = dummy variable taking a value of 1 if state i permits businesses to elect separate or combined reporting during time t; 0 otherwise. SALES = sales weighting for apportionment purposes in state i during time t. TAXINCEN = number of tax incentives offered by state i during time t. NONTAXINCEN = number of nontax incentives offered by state i during time t. FEDBASE = corporate tax collections in state i divided by the maximum federal corporate tax rate during time t. ∆PI = change in personal income in state i during time t and t – 1. *Significant at p < .1. **Significant at p < .05. ***Significant at p < .01.

R2

FEDBASE

∆PI

NONTAXINCEN

TAXINCEN

712


though the OLS estimate with time- and state-fixed effects is significant. Again, the 2SLS results evidence the difficulty of identifying the effect of LLC growth given that it is strongly correlated with time. Nonetheless, there is strong reason to believe that LLC growth has caused considerable base decline because of the significance in other versions of the model and of the effects that an entity-level tax on LLCs, as was enacted in Tennessee, appears to have had on actual collections. The coefficient estimate based on the 2SLS parameter estimate, state-fixed effects only, and average state characteristics indicates that a 10 percent increase in the number of LLCs would decrease the average state’s corporate tax revenues by 3.6 percent, or about $26 million. Any loss must follow either from new firms that select an LLC structure and otherwise would have selected C-Corp status or from existing C-Corps forming LLC subsidiaries for tax avoidance reasons. No revenue loss would have resulted from the choice of the LLC status rather than sole proprietor, S-Corp, or partnership structures since these latter structures would not have resulted in payment of corporate income taxes. The evidence does not indicate that the trend to businesses forming as LLCs has subsided (though there was a small reduction in 2000 in the share of businesses using the LLC form), meaning that the revenue losses could continue to grow. Requiring firms to file combined reports for tax purposes is one means for states to lessen the capacity of large, multistate corporations to engage in some state tax planning techniques (McIntyre, Mines, and Pomp 2001). REQUIRED is positive and significant, supporting the notion that combined reporting partially closes a potential loophole that otherwise allows multistate corporations to shift income to low-tax states.27 The point estimate suggests that adopting required combined reporting increases corporate tax revenues by $95.7 million in the average state. This important finding indicates that combined reporting could offset 44 percent of the revenue loss being attributed to LLCs.28 Of course, this suggests that required combined reporting limits but does not eliminate tax avoidance schemes, which should be expected since some parts of a corporate umbrella may not be required to participate in filing a combined return. Also, water’s edge required combined reporting would not offset tax planning losses available



713

through international holding companies and other cross-country techniques. VOLUNTARY is also positive, indicating that firms may be voluntarily using combined reporting for compliance ease as opposed to a means of tax avoidance. State legislatures are under considerable political pressure to design tax structures that are conducive to business location (Brunori 2001). Changes in the formula for apportioning multistate income have been a target in many states, with the propensity being to raise the weight on the sales factor. In most states, the political rhetoric suggests that the expectation is for revenue losses, albeit relatively small amounts, though in principle the effect could go either direction. The coefficient on SALES has an unexpected positive sign but becomes insignificant when state-fixed effects are added to the model.29 One explanation for the positive coefficient is that politically powerful firms or firms regarded as highly mobile have been effective at getting legislation passed that lowers their tax burdens while aggregate tax burdens rise. Also, increases in the sales factor could reduce revenues in some states, even though no evidence is found that it decreases revenues as a norm. As described above, net importing states should expect an increase in revenue, and the result is consistent with this expectation. Tax incentives are negative in all six model specifications and are highly significant in all but one case.30 The results suggest that the granting of tax concessions lowers tax revenues. Unfortunately, the measure of tax concessions used in this article is simply the number of tax concession programs provided by a state. Other measures such as the tax savings that specific types of firms receive from concessions or the effects of the concessions on marginal rates of return to investments at a particular location would be preferred, but the data are unavailable. The empirical results suggest that the presence of tax incentives lowers state corporate tax revenues, as would be expected since the concessions are often granted against this tax. However, alternative interpretations can be given to this finding. For example, states that grant more types of concessions may be more lax about collecting the corporate income tax. Also, our study does not estimate the effects that tax incentives have on economic activity, so it is not possible to determine whether the attraction of additional economic activity off-


714


set some of the revenue lost through the granting of concessions. The findings seem to suggest that granting of tax concessions is not selffinancing, though the independent effect arising from the stimulation of any new economic activity (not including the revenue costs of the concessions) would generate some additional corporate tax revenues and also greater revenue from other tax sources could result as well. Nontax incentives are positive and significant in the balanced model, but the coefficient becomes insignificant for the unbalanced model. The significant finding is very interesting as nontax incentives cost the states on the expenditure side, but these results provide some evidence that there is a revenue offset through greater corporate tax revenue. Again, the finding must be interpreted with some caution since the variable is only a count of nontax incentive programs. A change in the federal base is positive though only statistically significant without fixed effects, indicating that a decrease in the federal tax base will also lead to a decrease in the state tax base. An accounting relationship exists between the federal and state tax bases since essentially every state uses the federal definition of corporate income as a starting point for determination of state taxable income. Several conclusions can be drawn from the results. First, there is evidence that the federal corporate tax base has fallen relative to corporate profits during the 1990s (Fox and Luna 2002). The regression results support this decline in federal tax base as one cause of the declining state base that has occurred over the past fifteen years and, therefore, of the relative decline in state tax revenues. Second, state governments appear to be becoming increasingly unwilling to follow the federal tax base without modification because of federal policy decisions that otherwise narrow state tax bases without state action. For example, approximately half the states have decoupled their depreciation rules from the federal depreciation bonus legislated under the Job and Worker Assistance Act of 2002. Finally, ∆PI is positive and significant in some specifications, evidencing that corporate tax revenues are volatile and much larger when the state economy is robust.



715

6. SUMMARY, IMPLICATIONS, AND LIMITATIONS

State corporate tax revenues decreased dramatically as a share of the economy during the 1990s, and this article finds evidence that state policies are an important reason. All states enacted legislation permitting LLCs but in many cases did not amend their state tax structure. Businesses responded by rapidly increasing the number of LLCs as a percentage of total businesses from near zero in 1988 to more than 5 percent in 2000 because of the entity’s unique ability to combine flexible ownership and operation, limited liability for owner/managers, and pass-through federal tax treatment. The results indicate that LLCs have had a negative impact on state corporate tax revenues. The granting of tax incentives also lowers corporate tax revenues, though nontax incentives raise revenues. The problem of base erosion could be offset to some extent by states’ requiring combined reporting. The avoidance mechanisms allowed by LLCs and other means, together with failure to require combined reporting, lower state tax revenues, and increase the likelihood that state taxation is nonneutral and distorting. However, the aggregate revenue losses may be less than the estimates provided here. First, the conversion from a corporation to an LLC, or the choice to organize as an LLC instead of a corporation, will in some cases shift taxable income from corporate tax rolls to individual tax rolls because of the pass-through of income to individual members of the LLC. However, ownership of LLCs by large corporations, which is probably the largest source of revenue loss, does not result in additional individual income tax receipts, so this effect is probably small.31 Further research should be done to examine the effect of LLCs on the individual tax base. Second, some states have taken steps to mitigate the revenue losses. For example, several states have increased annual fees on LLCs, imposed withholding taxes, or imposed the corporate tax structure on LLCs. Some states have also responded by imposing add-back provisions for intangible holding companies and by levying alternative minimum tax provisions. The revenue effects of these changes are reflected in the data used in this study, to the extent that they occurred prior to 2000. This study does not quantify either the increases in fees or individual income taxes that mitigate the loss of corporate tax reve-


716


nues resulting from the dramatic increase in LLCs. Such analyses should also be the subject of future research. The study also provides some evidence of the importance that federal tax policy has on state tax bases. Efficient tax policy requires that the federal government consider such vertical externalities when policy is designed (Hayashi and Boadway 2001), though anecdotal evidence suggests that the federal government does not often consider these effects.

NOTES 1. For example, New Jersey recently found that thirty of the fifty largest employers in the state paid only the minimum tax of $200 per year. 2. For example, State Tax Review, vol. 61, no. 20 (May 15, 2000) reported on the findings of studies in California and Tennessee. Studies by those states estimated the 1996 revenue losses due to the introduction of limited liability companies (LLCs) to be between $5.4 and $10.1 million per year for Tennessee and to be $6.7 million per year for California. 3. Wyoming became the first state to permit LLC in 1977, followed by Florida in 1982. 4. All states have also amended their partnership statues to add limited liability partnerships (LLPs) to the list of business forms. The current study does not separately examine LLPs because the vast majority of LLPs are formed as an alternative to partnerships and therefore their popularity should not materially influence corporate tax revenues. 5. Tennessee and Kentucky are used as examples here, but no attempt is made to provide an exhaustive list of avoidance mechanisms. 6. States such as California, Georgia, Indiana, Iowa, Louisiana, Minnesota, Missouri, New Jersey, North Carolina, Ohio, South Carolina, and West Virginia have protected themselves from the nexus problem by requiring withholding on a nonresident member’s distributive share of income unless the member’s share is included in the LLC’s composite return. 7. For corporations, Kentucky uses a three-factor apportionment formula with a doubleweighted sales factor. Kentucky generally eliminated the tax advantages for LLCs as part of its 2005 tax legislation. 8. In 2000, revenues increased by at least $92 million because of a one-time boost from requiring quarterly franchise tax payments for the first time. 9. For studies that focus on firm attributes, see Gordon and MacKie-Mason (1994); Plesko (1994); Ayres, Cloyd, and Robinson (1996); Omer, Plesko, and Shelley (2000); and Goolsbee (2002). Also see Gentry (1994); Damadaran, Kose, and Liu (1997); and Regan and Tzeng (1999), which focus on industry classifications. 10. The equation can be thought of as estimating the probability of a firm’s becoming an LLC. Within this framework, only actual businesses (and not potential businesses) are considered in the analysis. See Bartik (1985) for a discussion. 11. State personal income, a broad measure of a state’s economy, is selected rather than gross state product because the income data are updated more frequently and were available for the entire panel when the study began.



717

12. The model takes enactment of laws allowing LLC formation and changing LLC characteristics as exogenous. 13. We also tested SALES using a 0-1 dummy variable. No significant differences are noted. 14. Alternative measures of the macroeconomy were investigated including change in real GDP, change in corporate profits, change in corporate profits before taxes, and change in adjusted corporate profits. None of these variations significantly altered the results. 15. Even after contacting each state directly, the data on the number of LLCs and total businesses are probably measured imprecisely in some cases. For example, for one year Connecticut reported that more than 70 percent of all businesses were LLCs. This observation appears to be in error and has been eliminated from the data. In addition, nine other observations were eliminated because the levels of LLCs were inconsistent with other LLC counts. Data errors result in measurement error in the SHARE variable, but this will not bias the estimates if the measurement error is not correlated with the error term in the revenue equation and if the measurement errors are not correlated over time or across states. 16. Some problems also existed in measuring the corporate tax revenue. For example, Alaska corporate tax data also include some taxes on oil revenue. As a result, corporate tax revenue in Alaska is six to ten times greater than the national average. Alaska was eliminated from the data set because it had a statistically significant effect on the regression results and because the residuals for Alaska were unusually large. 17. All states are included except for Alaska and states without a corporate income tax (Nevada, South Dakota, Texas, Washington, and Wyoming). 18. The approach of using a balanced and unbalanced panel is similar to that followed by Desai and Hines (2002). Minor modifications were required in the estimated equation when the unbalanced panel is used. For example, Rate_Diff was used rather than Rate_Diff × EFFECTIVE in the SHARE equations. 19. The risk involved in using a new entity will decrease over time as businesses and their advisors become more comfortable with operational details and the legal uncertainties associated with new entities are resolved. The time variable helps control for this risk. 20. Random effects estimates were also estimated. The Hausman test for fixed versus random effects showed no evidence of bias in the random effects estimates. Nonetheless, we believe that the fixed effects estimates are preferred because there are sufficient degrees of freedom to estimate the fixed effects model, there is no intent to make predictions outside of the sample, and there is risk of omitted variable bias in the random effects model (which is better accommodated within the fixed effects model). 21. First-stage results are available upon request. 22. The check-the-box (CTB) regulations lessened the importance of bulletproof statutes and potentially other characteristics of LLC statutes. A dummy variable for the years including and following the year when the CTB regulations were in effect was included in a separate set of regressions. The model was run both with and without the other LLC characteristics. CTB was positive and highly significant. It is possible that federal CTB legislation, rather than characteristics of state LLC legislation, accounted for much of the LLC growth after 1997. 23. Alternative measures of state taxes were examined. The results were mixed. 24. It is possible that the federal tax structure has the greater influence on organizational form, but once selected, the LLC form could be used to reduce state tax liabilities. 25. The model was also tested using CORP_RATE2 to see if taxes imposed a nonlinear effect, and no significant differences were observed. See Klassen and Shackelford (1998). 26. The revenue equation was also run by interacting tax rate with all other variables in the equation (based on the arithmetic relationship that tax revenue equals tax rate times tax base) and without an intercept term. Both the tax rate and tax rate squared are in this equation. The results


718


are essentially the same as those reported in Table 4, except that the tax rate was positive and statistically significant and tax rate squared was negative and statistically significant in the balanced panel version of the model. The balanced panel results indicate that revenues rise at a diminishing rate with the tax rate until the tax rate reaches 20.2 percent with fixed effects and 28.0 percent without fixed effects. Signs on the tax rate variables reverse in the unbalanced panel version. This leads to the nonintuitive result that revenues initially decline with tax rate increases and then rise. 27. Presence of a withholding tax on LLCs was tested in the equation, but it was statistically insignificant and had no influence on other variables. A likely explanation is that a variable that more precisely measured the structure of withholding taxes is necessary to measure their implications for revenues. 28. This calculation is based on a linear extrapolation of the point estimate of the coefficient on SHARE using the estimates only with state-fixed effects. 29. The model was also tested with SALES*CORP_RATE, but no significant differences were observed. See Klassen and Shackleford (1998). 30. TAXINCEN was negative but not statistically significant in the balanced panel without fixed effects. The model was also run using TAXINCEN2 and NONTAXINC2 (Gupta and Hoffman 2002). NONTAXINCEN2 becomes insignificant when fixed effects are included in the model. No other differences are noted. 31. Another equation with individual tax revenues as the dependent variable and SHARE, individual tax rate, a progressivity measure, capital gain income/adjusted gross income (AGI), and personal income as right-hand-side variables was estimated. When run as a two-stage least squares (2SLS) model, SHARE is statistically insignificant. In addition, some businesses will choose to form the LLC over the S Corporation. In general, this choice should not result in any change in overall revenues. We do not attempt to decompose all the revenue effects.

REFERENCES Anand, Bharat N., and Richard Sansing. 2000. The weighting game: Formula apportionment as an instrument of public policy. National Tax Journal 53 (2): 183-99. Ayres, Benjamin C., C. Bryan Cloyd, and John R. Robinson. 1996. Organizational form and taxes: An empirical analysis of small businesses. Journal of the American Taxation Association 18 (Suppl.): 49-67. Bartik, Timothy J. 1985. Business location decisions in the United States: Estimates of the effects of unionization, taxes, and other characteristics of states. Journal of Business and Economic Statistics 3 (1): 14-22. Brunori, David. 2001. State tax policy: A political perspective. Washington, DC: Urban Institute Press. Cornia, Gary, Kelly D. Edmiston, David L. Sjoquist, and Sally Wallace. 2005. The disappearing state corporate income tax. National Tax Journal 58 (1): 115-38. Damadaran, Aswath, John Kose, and Crocker H Liu. 1997. The determinants of organizational form changes: Evidence and implications from real estate. Journal of Financial Economics 45 (2): 169-92. Desai, Mihir A. 2002. The corporate profit base, tax sheltering activity, and the changing nature of employee compensation. National Bureau of Economic Research Working Paper, Cambridge, MA.



719

Desai, Mihir A., and James R. Hines. 2002. Value-added taxes and international trade: The evidence. University of Michigan Working Paper, Ann Arbor. Edmiston, Kelly D. 2002. Strategic apportionment of the state corporate income tax. National Tax Journal 55 (2): 239-62. Fay, Elisa, and Sharlene Amitay. 2001. Attention LLC members: Is nexus a foregone conclusion? State Tax Notes 67 (37): 821-26. Fox, William F., and LeAnn Luna. 2002. State corporate tax revenue trends: Causes and possible solutions. National Tax Journal 55 (3): 491-508. Gentry, William. 1994. Taxes, financial decisions and organizational form: Evidence from publicly traded partnerships. Journal of Public Economics 53 (2): 223-44. Goolsbee, Austan. 2002. The impact and inefficiency of the corporate income tax: Evidence from state organizational form data. National Bureau of Economic Research Working Paper, Cambridge, MA. Goolsbee, Austan, and Edward L. Maydew. 2000. Coveting thy neighbor’s manufacturing: The dilemma of state income apportionment. Journal of Public Economics 75 (1): 125-43. Gordon, Roger H., and Jeffrey K. MacKie-Mason. 1994. Tax distortions to the choice of organizational form. Journal of Public Economics 55 (2): 279-306. Gordon, Roger, and John D. Wilson. 1986. An examination of multijurisdictional corporate income taxation under formula apportionment. Econometrica 54 (6): 1357-73. Guenther, David. 1992. Taxes and organizational form: A comparison of corporations and master limited partnerships. Accounting Review 67 (1): 17-45. Gupta, Sanjay, and Mary Ann Hoffman. 2003. The effect of state income tax apportionment and tax incentives on new capital expenditures. Journal of the American Taxation Association 25 (Suppl.): 1-25. Haas, June S. 2003. Passthrough nexus? Nexus issues for nonresident corporate partners, members, and shareholders. State Tax Notes 28 (6): 561-66. Hayashi, Masayoshi, and Robin Boadway. 2001. An empirical analysis of intergovernmental tax interaction: The case of business income taxes in Canada. Canadian Journal of Economics 34 (2): 481-503. Klassen, Kenneth J., and Douglas A. Shackelford. 1998. State and provincial corporate tax planning: Income shifting and sales apportionment factor management. Journal of Accounting and Economics 25 (3): 385-406. Lowy, Peter A., and Juan F. Vasquez Jr. 2003. When is it constitutional for states to tax nonresident members of limited liability companies? State Tax Notes 28 (7): 633-44. Luna, LeAnn. 2004. Corporate tax avoidance strategies and states’ efforts to prevent abuses. Journal of Multistate Taxation and Incentives 14 (2): 6-17, 46-47. McIntyre, Micheal J., Paul Mines, and Richard D. Pomp. 2001. Designing a combined reporting regime for a state corporate income tax: A case study of Louisiana. Louisiana Law Review 61:699-761. Omer, Thomas C., George Plesko, and Marjorie K. Shelley. 2000. The influence of tax costs on organizational choice in the natural resource industry. Journal of the American Taxation Association 22 (1): 38-55. Plesko, George A. 1994. Corporate taxation and the financial characteristics of firms. Public Finance Quarterly 22 (3): 311-34. Pomp, Richard D. 1998. The future of the state corporate income tax: Reflections (and confessions) of a tax lawyer. In The future of state taxation, ed. David Brunori, 49-72. Washington, DC: Urban Institute Press. Regan, Lareen, and Larry Y. Tzeng. 1999. Organizational form in the property-liability insurance industry. Journal of Risk and Insurance 66 (2): 253-73.


720


Scholes, Myron S., and Mark A. Wolfson. 1992. Taxes and business strategies: A planning approach. Englewood Cliffs, NJ: Prentice Hall. Terando, William D., and Thomas C. Omer. 1993. Corporate characteristics associated with master limited partnership formation. Journal of the American Taxation Association 15 (1): 23-45. Wasylenko, Michael. 1997. Taxation and economic development: The state of the economic literature. New England Economic Review, March/April, pp. 37-52. Watkinson, Troy S. 1989. LIFO recapture on conversion from C corporation to S corporation: An S corporation penalty and failure in tax policy. The Tax Lawyer 42 (3): 841-58.

William F. Fox is William B. Stokely Distinguished Professor of Business and director of the Center for Business and Economic Research at the University of Tennessee. He is past president and recipient of the Steven D. Gold Award from the National Tax Association. LeAnn Luna is an assistant professor in the Department of Accounting and Information Management and in the Center for Business and Economic Research at the University of Tennessee. Prior to receiving her Ph.D. in accounting from the University of Tennessee, she worked as a tax consultant for an international public accounting firm.
