|
Article Excerpt
State corporate income tax (SCIT) revenues have declined significantly over the past 20 years, prompting vigorous debates about the likely causes of, and potential remedies for, this decline. Confounding the analysis is the activism of many states in modifying the structure of their SCITs, particularly the apportionment factor weights and statutory tax rates, to differentiate them from other states' regimes in attempts to attract new investment and/or retain economic activity within their borders. In this paper, we focus on the revenue effects of differences in tax policies among states.
Prior research on the determinants of SCIT revenues is sparse and the studies that have examined the effects of apportionment formula differences on state tax revenues have produced mixed results. Further, state tax codes differ in many other respects, notably in the definition of the tax base and in various accounting policies that affect the measurement of taxable income, which have been examined thus far only in piecemeal fashion. We contribute to the literature in two ways. First, we address the important methodological problem of endogeneity between SCIT revenues and state tax policies, specifically the apportionment formula weights and tax rates, using a two-stage least squares estimation approach. Second, we examine the impact of a broad array of state tax policy choices on SCIT revenues.
Using aggregate state-level panel data from 1982-2002, we estimate fixed-effects regression models of SCIT revenues as a function of state tax policies, controlling for factors such as organizational form, federal tax base changes, and general macroeconomic trends. We find that accounting for the endogeneity of apportionment formula weights in models of SCIT revenues is crucial; once appropriate controls are introduced, states with a double-weighted sales factor experience on average 16-18 percent lower SCIT revenues than states with an equallyweighted sales factor. With respect to tax rates we find a one percentage point higher statutory tax rate is associated with 11-12 percent higher SCIT revenues, with or without endogeneity controls.
We also find that several other tax policies are statistically and economically significant. Specifically, use of a throwback rule is associated with roughly 16 percent higher SCIT collections, and employing a broader definition of business income is associated with 15-21 percent higher SCIT revenues, although combined reporting surprisingly is not significantly associated with SCIT revenues. These findings are robust to a number of alternative specifications and sensitivity tests.
States struggle with the conflicting goals of raising revenue while remaining competitive with other states to attract business. Thus, apportionment formula changes and tax incentives are accompanied by base-broadening measures to thwart strategic tax planning. However, continued shortfalls in SCIT revenues are leading to calls for more drastic policy initiatives, such as repealing the SCIT, or replacing the current system with a uniform nation-wide system administered by the Federal government. Our results on the predicted impact of differences in state tax policies on SCIT revenues suggest that the latter policy prescription has merit.
INTRODUCTION
State corporate income tax (SCIT) revenues have decreased over the past twenty-five years relative to other sources of state tax collections (Wilson, 2006; Sullivan, 2007). As Figure 1 shows, SCIT revenues declined by about 50 percent over the 21-year period, 1982-2002, whether scaled by total state tax collections, federal corporate taxable income, or gross state product. In fact, many states currently raise more revenue from lotteries or tobacco and gasoline excise taxes than from the corporate income tax. Economists note with puzzlement that the decline coincides with increasing corporate profits as a share of national income (Cornia et al., 2005), and deviates from the trend in federal corporate income tax revenues. (1) The decline of the SCIT has sparked an active debate as to its likely causes and potential remedies (Pomp, 1998; Fisher, 2002; Fox and Luna, 2002; Hofmann, 2002; Mazerov, 2003a; Cornia et al., 2005; Wilson 2006; Mazerov, 2007a; Sullivan, 2007).
The leading causes suggested for the relative decline in SCIT revenues include: 1) changes in the federal income tax rules and aggressive federal income tax planning, both of which affect federal taxable income (the typical starting point for computing the SCIT); 2) changes in organizational form, especially the increased use of flow-through entities whose income is typically not included in the corporate income tax base and which are widely used in state tax planning strategies; and 3) changes in the economy resulting from a shift away from manufacturing to services and technology for which the extant design of state tax systems is arguably obsolete (Brunori, 2002; Hofmann, 2002; Tannenwald, 2002; Florida Senate Committee on Finance, 2003; Cornia et al., 2005; Bankman, 2007). A concurrent development over the past quarter century has been the increasing use of the SCIT as an instrument of economic development charged with attracting business and jobs into states, and only secondarily as a means of raising revenue to meet state budget needs. With economic development as an objective, state legislators have actively sought to implement policies to differentiate the structure of their state's SCIT from that of other states, adding another candidate to the list of likely explanations for the revenue decline.
[FIGURE 1 OMITTED]
In this study we focus on this development and examine the role of tax policies used by states to measure, allocate/apportion and tax corporate income in explaining the pattern of SCIT collections over the two decades from 1982-2002. During this period, while SCIT revenues relative to overall economic activity measures have generally trended downwards, experiences between states have differed markedly, with some states recording stable revenues as a share of total tax collections, and some even showing revenue increases (Cornia et al., 2005). These differing state experiences suggest that factors likely to affect all states similarly, such as changes in the federal tax base or macroeconomic fluctuations, probably cannot fully explain the SCIT revenue decline. Further, during this period many states changed their tax structures, particularly the factor weights in their apportionment formula and the statutory tax rate.
We believe that focusing on the role of state--level differences in the structure of the SCIT is critical to the policy debate, in part because the extant empirical research on the declining SCIT phenomenon and its potential causes is sparse. The studies that exist focus on the apportionment formula and the effects of changing the weights placed on the factors used in this formula on economic activity, such as employment (e.g., Goolsbee and Maydew, 2000) and investment (e.g., Gupta and Hofmann, 2003), and on firms' incentives to engage in tax planning (e.g., Gupta and Mills, 2002). To be sure, some studies have also examined the effect of apportionment formula differences on state tax revenues (e.g., Klassen and Shackelford, 1998; Edmiston, 2002; Edmiston and Arze, 2006). Although this attention is well deserved given the primacy of formula apportionment in determining the SCIT and the frequency of changes made in the formula's factor weights, state tax codes differ in many other respects as well, notably in the definition of the tax base and in various accounting policies that affect the calculation of taxable income. Recent studies have begun to broaden the enquiry to include factors such as tax incentives (Fisher, 2002) and the increased use of flow-through entities (Fox and Luna, 2005) in explaining the revenue decline. Yet, ambiguities remain because of conflicting results, failure to consider all of the relevant structural differences, and methodological issues, prompting calls for further research (Cornia et al., 2005).
We contribute to this line of research in two ways. First, we address one of the largest methodological challenges facing all studies in this area--endogeneity between SCIT revenues and state tax policies, specifically apportionment formula weights and tax rates. Tax policies are not randomly assigned. Instead, each state's policymakers identify
revenue needs and sources differently depending on a variety of variables, including historical factors, natural resources, economic opportunities, education systems, and political affiliations of elected officials. Over time these variables have combined to shape each SCIT system, while each SCIT system has also likely had an impact on some or all of these variables. More immediately, policymakers consider revenue needs when defining the tax base, the accounting policies used to determine the base, and the tax rates used to compute state corporate income tax liability. Thus, state tax policies are likely endogenous with state corporate tax revenues. Unlike previous studies, we explicitly attempt to mitigate this problem with a two--stage least squares approach to account for the endogeneity of apportionment formula weights and tax rates in SCIT revenue regressions.
Second, we comprehensively examine whether, and to what extent, a broad array of state tax policy choices affects SCIT revenues. Thus, our analysis also considers cross-state differences in the definition of the tax base, such as whether the state utilizes the throwback rule or requires unitary (or combined) reporting, and in various other tax policies, such as net operating loss carryback provisions, definition of business income, deductibility of federal income taxes, imposition of the alternative minimum tax, and state policies relating to passive investment companies.
We use aggregate state-level panel data covering the years 1982-2002 from a variety of public sources to estimate fixed-effects regression models of SCIT revenues that control for unquantifiable and/or unobservable state- or year--specific characteristics that may influence the level of SCIT revenue collected. The models we employ also include controls for tax planning through organizational form, federal tax base changes and other general macroeconomic trends. We find that controlling for the endogeneity of apportionment formula weights in models of corporate tax revenues is crucial; once appropriate controls are introduced, higher weights on the sales factor in the apportionment formula are associated with lower SCIT revenues. With respect to tax rates we find that controlling for endogeneity does not materially affect inferences; higher statutory corporate income tax rates are associated with higher SCIT collections in models with or without such controls. Specifically, we find that states with a sales factor weight of one-half (a "double-weighted" sales factor) have 16-18 percent lower SCIT revenues on average than do states with a sales factor weight of one-third (an "equally-weighted" sales factor), while a one percentage point higher statutory tax rate is associated with 11-12 percent higher SCIT revenues.
With respect to the effects of other state tax policies, we find that the use of a throwback rule, disallowance of net operating loss carrybacks, and the use of a broader definition of business income have a positive and generally significant association with SCIT collections. However, requiring combined reporting or imposing an alternative minimum tax are not significant; and rules nullifying the use of passive investment companies surprisingly have a negative (marginally significant) association with SCIT revenues. Our results suggest that some of these policies are also economically significant. For example, the use of a throwback rule is associated with roughly 16 percent higher SCIT collections, and expanding the tax base using a broader definition of business income is associated with 15-21 percent higher SCIT revenues.
The rest of the paper proceeds as follows. The next section briefly describes state tax policies generally and variations enacted in some of those policies over the past two decades. This is followed by a review of prior research and a description of our research design and the results, including an analysis of the economic significance of the key SCIT policies on revenues. A brief summary concludes the paper.
INSTITUTIONAL BACKGROUND
State Tax Policies
As illustrated in Figure 2, states make a number of tax policy choices that can influence the amount of corporate income tax revenues collected. First are items that affect the measurement of the tax base. For organizations taxed as corporations, most states begin the computation of taxable income with federal taxable income. (2) Idiosyncratic state-level policies then apply so that specific types of income are included or excluded, and certain deductions are allowed or disallowed. For example, all states require the add-back of state income taxes deducted on the federal return. Likewise, a number of states do not permit net operating loss (NOL) carry-backs to offset prior year income. These adjustments tend to increase the state's tax base. On the other hand, some states allow a full or partial deduction for federal income taxes in arriving at state taxable income, which significantly reduces the SCIT base.
For multi-state firms, an important distinction exists between the types of income that are apportioned among the states in which the firm operates (i.e., typically "business income"), and the types of income that are allocated to the state in which the income is earned (i.e., interest income, certain non-operating gains and losses, etc.). We discuss the apportionment of business income in more detail later.
Another policy that affects the size of the tax base is the treatment of members of an affiliated group, particularly when one or more members are located in different states. So-called "unitary" states treat all affiliated corporations as a single entity and mandate combined reporting, under which allocable and apportionable income (as well as the apportionment factors) is summed across all firms in the group and the income taxable to the state is computed as though the group were one taxpayer. In 2002, at the end of our sample period, 14 states required combined reporting of all unitary businesses, 21 states prohibited combined reporting, and the remaining states allowed it at the option of the taxpayer or the state (CCH, 2004). The combined reporting requirement is designed to remove the advantage that a firm might gain by shifting profits to affiliates located in low-tax states, mostly Nevada or Delaware, using techniques such as transfer pricing, passive investment companies (PICs), real estate investment trusts (REITs), (3) or inter-company loans (Smith, 2000; McIntyre, Mines, and Pomp, 2001; Mazerov, 2007b; Sheppard, 2007; Sontag, 2008). Combined reporting potentially brings more out-of-state income into the tax base although, if out-of-state affiliates incur losses, it could also reduce the amount of income apportioned to a unitary state.
A less sweeping policy aimed at preventing income shifting is denying deductions for interest and royalties paid to affiliated corporations, particularly PICs. In recent years, some states have enacted such a prohibition. Furthermore, some states have enacted an alternative minimum tax similar to that at the federal level. (4)
The Apportionment Formula and the Throwback Rule
The apportionment formula is a key feature of the SCIT that is used to subdivide multi-state firms' income among jurisdictions with which the firms have sufficient contact (i.e., "substantial nexus"). (5) In general, business income is apportioned among states based on the proportions of sales and payroll that occur, and the property that exists, within each state. The logic of the formula is that a weighted application of these factors produces a fair reflection of income, and hence the SCIT, attributable to each state. Under the formula, a multi-state firm's income tax liability, x, in state i is computed as:
[1] [x.sub.i] = [([w.sup.P.sub.i] x [p.sub.i]/P)+([w.sup.L.sub.i] x [l.sub.i]/L]) + ([w.sup.S.sub.i] x [s.sub.i]/S)] x [pi] x [r.sub.i]
where [pi] is apportionable income; [r.sub.i] is the statutory tax rate in state i; [p.sub.i] [l.sub.i], and [s.sub.i] are the firm's property, payroll, and sales within state i, while P, L, and S are the firm's total property, payroll,...
|
