Featured Articles

Black Holes, Tax Shelters, and Other Surprises — Proof by Data

Posted on February 15, 2008 by Joann M. Weiner

Document originally published in Tax Notes International
on February 11, 2008.

Tracking down tax shelter activity is a top priority of the Internal Revenue Service. Unfortunately, identifying tax shelter activity can be as difficult as finding objects trapped inside a black hole — their presence is known only through inference.

"The primary problem with tax shelters from research, investment, tax administration, and enforcement perspectives is detecting them," according to Petro Lisowsky.

Lisowsky, a doctoral candidate in business administration at the Boston University School of Management, made these remarks in his paper "Seeking Shelter: Empirically Modeling Tax Shelters and Examining Their Link to the Contingent Tax Liability Reserve," presented at the 11th annual Tax Symposium held at the University of North Carolina on January 25-26 in Chapel Hill.

The U.S. Treasury Department defines a tax shelter as a transaction or arrangement that generates tax losses without incurring economic losses or risk. Because "it is very hard to measure an absence of income," Treasury does not estimate the amount of revenue lost due to tax shelters. However, in its study The Problem of Corporate Tax Shelters: Discussion, Analysis and Legislative Proposals (July, 1999, p. 31), Treasury notes that other sources estimate that corporate tax shelters may cost the U.S. government about $10 billion each year.

With Lisowsky's research, however, corporate tax shelter detection may no longer be as elusive as it was. Using confidential information from the IRS Office of Tax Shelter Analysis (OTSA), which identifies firms that have used one or more of 64 specific illegal tax-sheltering activities — such as son-of-BOSS, lease strips, partnership straddles, and corporate-owned life insurance — from 2000 through 2004, Lisowsky is able to shine a bit of light on the black hole of tax evasion and tax shelters.

By matching OTSA data with publicly available financial information, Lisowky presents a relatively long list of activities that may reflect the characteristics of tax shelters. Interested parties may then use these data to help identify these shelters. Specifically, firms with tax-haven-based subsidiaries, material foreign operations, complex financial transactions, litigation losses, and relationships with tax promoters are all more likely to be involved in tax shelters than firms without these features.

Lisowsky also establishes an empirical link between the tax cushion, which is the contingent tax liability reserve that firms use to report their uncertain tax benefits in their financial statements, and the use of tax shelters. Lisowsky finds that the tax cushion is not only positively related to tax risk but also is positively related to tax shelter use.

Lisowsky's research confirms what has long been suspected. On average, a nontrivial portion of the tax cushion is likely attributable to tax shelters. In turn, this finding suggests that financial earnings may also be a function of tax shelters to the extent that shelters affect tax expense via the tax cushion.

Since the tax cushion data are publicly available, unlike the tax return data, this evidence is likely to benefit greatly those researchers who must rely on financial statements to uncover tax shelter information. Until now, the evidence that tax shelters affect financial statements through the tax cushion was largely anecdotal. Earlier studies that had examined financial statements for evidence of tax shelter use had inferred, for example, that large differences between book and tax income suggested the use of tax shelters. However, direct evidence for tax shelter use and suggestions to increase the probability of their detection was not available. This evidence, therefore, greatly benefits researchers who must rely on financial statements alone to detect tax shelter use.

Lisowsky has highlighted a number of characteristics that may identify firms involved in tax shelters. It is now up to the tax authorities to identify exactly how much income these firms may be hiding through these tax shelters.


The Consequences of JGTRRA


The evidence keeps rolling in on how U.S. taxpayers took advantage of the favorable tax treatment provided to domestic and certain foreign dividends to reduce their tax liabilities. The Jobs and Growth Tax Relief Reconciliation Act (JGTRRA) of 2003 provided a relatively narrow tax break for certain foreign dividends that may have strongly influenced the international portfolio choices of U.S. individual investors. JGTRRA sharply changed the dividend tax treatment at the individual level by specifying that dividends would be taxed at the capital gains rate. The tax benefit was further enhanced by the corresponding reduction in the capital gains rate to a maximum 15 percent. Dividends paid by domestic corporations and by certain qualified foreign corporations qualified for the reduced rate of taxation.

Dhammika Dharmapala of the University of Connecticut presented a paper, "Taxes, Dividends, and International Portfolio Choice," coauthored with Mihir Desai of Harvard University, that examined how the reduced personal taxation of dividends paid by qualified corporations located in tax treaty countries affected individuals' portfolio choices.

In an unprecedented act, JGTRRA restricted the relief provided to foreign corporations to those corporations that met one of three tests: the "possessions test," the "market test," and the "treaty test." The first two tests essentially covered dividends from corporations resident in a U.S. possession and dividends from corporations whose shares are traded in the United States. Under the treaty test, which is the case Dharmapala discussed, only corporations located in one of 52 IRS-specified treaty countries — those with effective limitation on benefits and exchange of information provisions — qualified for the tax break. Thus, dividends from corporations located in most European countries received favorable tax treatment, while dividends from corporations located not only in "tax haven" locations but also in countries such as Argentina, Brazil, and Chile would not receive the benefit.

Using data from the Treasury International Capital (TIC) reporting system, the authors found that individuals reacted strongly to the up-to-20-percentage point advantage granted to dividends received from certain foreign corporations. Individuals disproportionately increased their portfolio investment in treaty countries relative to nontreaty countries.

Other factors could have explained the changes in portfolio choice. However, the changes are not explained by changes in the underlying trends for foreign portfolio investment, nor are they explained by unobservable nontax factors correlated with treaty status or by the existence of exceptional returns in stock markets located in treaty countries. None of these factors explained the differences.

This paper builds on earlier research Dharmapala presented at the National Tax Association annual meeting in November that also used the TIC data. That paper, "Taxes, Institutions, and Foreign Diversification Opportunities," also coauthored with Desai, compared the pattern of U.S. foreign portfolio investment (FPI) in foreign equities with the pattern of foreign direct investment (FDI) holdings by U.S. firms. Since returns from FDI are subject to an additional layer of U.S. corporate tax on repatriation while the returns from FPI are not subject to U.S. corporate taxation, the researchers expected to find that U.S. FPI would be more sensitive to foreign countries' corporate tax rates than is U.S. FDI.

Using cross-country and panel data from 1994 to 2005, and controlling for U.S. FDI holdings and other variables, the authors found that U.S. individual holdings of equity FPI are highly sensitive to foreign corporate tax rates. Countries with lower corporate tax rates tend to have a higher ratio of U.S. FPI to U.S. FDI. The authors suggest that the high U.S. corporate tax rate may disadvantage U.S. MNCs as a vehicle for investing in foreign countries. U.S. investors may prefer to invest in a foreign economy through portfolio investment in foreign firms rather than through American firms.

The authors did not make this specific point, but their results may provide another explanation for the relatively low corporate income tax receipts despite the relatively high U.S. corporate tax rate. U.S. investors may be able to avoid paying the U.S. corporate tax rate by investing in foreign economies through foreign corporations rather than through U.S. multinational corporations.


Taxes and Stock Ownership


Since World War II, investors in many industrialized countries have significantly reduced their direct investment in stocks, preferring to invest indirectly via mutual funds, pension funds, and other tax-favored institutions. For example, U.S. individuals owned more than 90 percent of the stock market following World War II, compared with 27 percent in 2006. Over all industrialized countries, individuals own directly just 17 percent on average of the stock market.

Ilya Strebulaev of Stanford University presented these results in a paper, "The Evolution of Aggregate Stock Ownership: A Unified Explanation," coauthored with Kristian Rydqvist and Joshua Spizman of Binghampton University. According to the authors, changes in the relative tax advantages offered for owning stocks indirectly rather than directly explains the decline in direct share ownership.

While the explanations differ by country — indirect ownership through financial institutions provides the explanation in the United States and the United Kingdom, while it is indirect ownership through large business groups that provides the explanation in Sweden and Japan — the implications are not different. Four decades of historically high individual income taxation that existed until the major tax reforms in the 1980s drove individuals to invest in the stock market indirectly via tax-favored investment vehicles.

The authors evaluated the time series and cross-country data of aggregate equity ownership and tax systems since World War II in the United States, the United Kingdom, Sweden, Canada, and Japan. The implications of the authors' research could be very far-reaching. Their evidence suggests that taxes affect not only investor behavior but also affect the development of financial institutions. In particular, the sustained high level of individual income taxation in a range of countries led to the development of tax-favored institutions, such as mutual funds and life insurance companies, and pension funds, as well as to the development of the financial intermediaries.

In the United States, for example, pension funds are taxed under a consumption-tax basis whereas regular savings are taxed under an income-tax basis. Thus, contributions to pensions occur before tax, investment returns accrue tax free, and distributions are taxed as personal income. To obtain these tax benefits, individuals must invest indirectly through pension funds. As consumption tax advocates note, investment in regular savings is subject to double taxation — once when the income is earned and once when it is reinvested — whereas investment via pension funds or other retirement accounts is taxed only once.

The authors' evidence has profound implications for using tax policy to provide tax relief. In particular, with households investing relatively small amounts in taxable forms, such as stocks, the economic effects of modifying marginal tax rates on individuals may be relatively small as well. Tax policies, such as JGTRRA, that attempt to provide tax relief by reducing individual tax rates may provide much less relief than anticipated.

As the authors note, "Government regulation has shaped financial institutions by creating tax clienteles and leading economic agents to invent ways to circumvent their tax obligations."

Furthermore, if individual income taxation is no longer practical or relevant, the authors' research may have implications well beyond the study of tax policy. Classic finance papers written before the sharp reductions in personal income taxation in the 1980s may no longer provide relevant information for purposes of evaluating issues such as the cost of capital or for purposes of the capital asset pricing model.

Strebulaev, Rydqvist, and Spizman recognize that they have additional research to conduct on the systems in other countries. Nevertheless, this research provides a fruitful start into understanding how tax policy may impose significant unintended consequences.


Upsetting the Conventional Wisdom


Some academic conferences are profitable to attend . . . literally.

In a paper, "Tax Expense Surprises and Future Stock Returns," coauthored with Jake Thomas of the School of Management at Yale University, Frank Zhang, also of Yale University, shows that seasonally differenced quarterly tax expenses — which are a proxy for tax expense surprise — predict future stock returns over the next two quarters. They conducted their study using quarterly data available from Standard & Poor's Compustat database and the University of Chicago's Center for Research in Security Prices monthly and daily return files for the period from 1977 through 2005.

In layman's terms, their research shows that it may be possible to make profits systematically by exploiting this anomaly in stock returns. Stock prices incorporate this information with a delay because investors do not recognize fully that surprises in reported tax expenses predict two key variables — future book income and future tax expense — that companies release in the upcoming two quarters.

Since book income, pretax income, and tax expenses are interrelated, it is important to control for surprises in book income to focus on the impact of surprises in tax expenses. Controlling for book income surprises also ensures that the tax expense strategy is not unintentionally mimicking the well-known earnings momentum investment strategy, which is based on earnings surprises.

After controlling for these issues, the researchers found that tax expenses that were higher than expected were "good news" for investors because these unexpectedly high tax expenses indicated that pretax income would also be higher than expected. In all of their specifications, tax expense surprise was consistently and significantly positively related to future stock returns. The information provided by the tax expenses surprise provided incremental explanation beyond the information provided by future book income surprises.

In other words, markets are not fully efficient over this short time period, and traders can profitably exploit these pricing anomalies.

This research goes against the efficient market hypothesis, which says that stock prices fully reflect all available information. Ever since Princeton economist Burton G. Malkiel published his book A Random Walk Down Wall Street (W.W. Norton and Co., first in 1973 and regularly updated), investors have been cautioned that they cannot beat the market. Stock prices follow a random walk, perhaps with a drift, making it impossible to predict future share prices.

Before you race off and call your broker, the Yale researchers warn that their results are not designed to help pick individual stocks. The evidence also does not imply substantial and sustained mispricing. However, it does suggest that stocks with extreme earnings surprises have deviated from their efficient prices by about 5 percent and that this deviation is corrected over about a six-month period.

Although the results go against conventional wisdom, they are built on a solid logical foundation and one that anyone who has tried to decipher a Fortune 500 company's financial statements will understand. As the researchers note, "We believe that some of this information contained in tax expense may be reflected in stock prices with a delay because tax disclosures are too complex to be fully understood by most investors."

In particular, since the tax expense reported for book purposes reflects both a current component, which is based on income defined under tax rules, and a deferred component, which reflects attempts to manage book income, there may be information in the reported tax expense that reflects pretax income but not book income.

Many tax conferences merely provide participants with valuable investment knowledge. This conference potentially provided its participants with profitable investment returns.


More Surprises From FIN 48


In July 2006 the Financial Accounting Standards Board introduced new rules for accounting for uncertainty in income taxes that became effective in 2007. These new rules, known as FIN 48, "Accounting for Uncertainty in Income Taxes," are intended to improve the accuracy and visibility of a publicly traded firm's reporting of these uncertain tax benefits.

Under the old standard, firms would establish an account that represented their expected tax benefits. The rule firms generally followed was based on how much of the tax benefit the firm expected to retain if it were audited. For example, if a firm engaged in R&D and claimed a tax credit for its expenses, there was some probability that the IRS would audit the taxpayer and reject some or all of that claim.

The taxpayer would interpret the tax law — which can be subject to significant ambiguities — and then determine its filing position. This tax filing position would reflect the firm's own judgment about whether the IRS would challenge the claim and, if so, how much of the claim the IRS would sustain.

Because the firm had more information about its tax position than the IRS did, the firm was likely to make a larger claim of its expected tax benefit than it might have if the IRS had had the same information. Since auditing is costly to the IRS, firms may have won the audit lottery more often than they should have.

The new rules for accounting for income tax uncertainty have dramatically changed this situation.

The new rules require publicly traded firms to follow a two-step approach — recognition and measurement — to account for the tax uncertainty. Before beginning this new process, the firm identifies an uncertain tax position, which may, for example, involve the R&D credit. The firm then determines whether the court is "more likely than not" to recognize the claim, based on the technical merits of the tax position. Then, for tax positions that the court is likely to recognize, the firm measures the uncertain tax position as the largest amount that, on a cumulative probability basis, is "more likely than not" to be realized upon ultimate settlement.

Tuck School of Business at Dartmouth Prof. Leslie Robinson presented research in "FIN 48 and Tax Compliance," a paper coauthored with Lillian Mills of the University of Texas at Austin and Richard Sansing of the Tuck School of Business at Dartmouth that explores a new aspect of the FIN 48 rules. The authors modeled how the new rules for reporting uncertainty in income tax benefits, which they call "the most significant change in financial accounting for income taxes over the past decade," affect the strategic relationship between publicly traded corporations and the government.

They were particularly interested in how the new rules might affect not only how much uncertainty a taxpayer reveals through its tax filing position, but also how that disclosure might affect the taxpayer's strategic interaction with the government. They examined whether the new rules would give the IRS previously unreported information concerning potentially aggressive tax positions.

This research exploits the fact that ambiguities in the tax law create uncertainty over the facts and circumstances that apply to a particular tax position. This ambiguity may encourage taxpayers to report an "aggressive" tax position knowing that the facts and circumstances of the case may lead the IRS to accept none, some, or all of that position on audit. In other words, tax law ambiguity may often work to the taxpayer's advantage.

The authors' model presents some surprising implications, all of which are contrary to popular impression. First, FIN 48 may actually "protect" taxpayers who establish small tax reserves for large tax benefits. In the post-FIN 48 world, taxpayers that have a strong factual case may have a higher expected payoff from reporting a relatively large tax benefit than in the pre-FIN 48 world. The large benefit claimed for an uncertain tax position may signal a strong factual position.

Second, the firm's disclosed tax liability does not necessarily overstate the actual tax expense the firm expects to incur. The reported tax expense depends on how the taxpayer evaluates the likelihood that it will be audited and the distribution of its postaudit tax benefits.

Finally, FIN 48 does affect tax reporting for positions that the taxpayer believes the IRS may not sustain. In other words, if the tax position does not meet FIN 48's new "more likely than not" standard, then the firm is less likely to report an aggressive tax position than it would have been under the old rules.

Therefore, FIN 48's biggest impact may be on the subset of firms that have a weak set of facts. These firms are unlikely to report a large tax reserve if they estimate they have a less than 50 percent chance of prevailing on audit. FIN 48, however, does not lead taxpayers to overstate their expected tax liabilities associated with uncertain tax positions.

In responding to the implications of their model, Mills explained that the model does a good job of explaining behavior of "companies that are involved with only one tax authority and do not have multinational issues."

"We'll be obtaining FIN 48 data over the next few years and will be able to test our theoretical model on real data at that time," Mills said.


Understanding Marginal Tax Rates


Analysts use marginal tax rates (MTRs) to understand why corporations undertake activities ranging from where they choose to invest, why they pay dividends, how they compensate their employees, and how they decide to finance that investment. Unfortunately, there is much disagreement on exactly which "tax rate" is important for these types of decisions.

Jennifer Blouin of the Wharton School at the University of Pennsylvania attempted to shed some light on this issue in her paper "Improved Estimates of Marginal Tax Rates: Why They Are Needed, Approach, and Implications," coauthored with John E. Core and Wayne Guay, also of the University of Pennsylvania.

Blouin first establishes that the MTR she is discussing is the present value of current and expected future taxes paid on an additional dollar of income earned today. The MTR is neither the statutory tax rate nor the average tax rate, which is the ratio of taxes paid to income. The MTR depends on both the tax code and on expectations of future taxable income. Because firms can use negative taxable income in one period to offset taxable income in prior and future periods, taxes owed are not a linear function of taxable income. Thus, the MTR depends on estimates of the distribution of future taxable income.

Using Compustat data from 1980 to 1994, the authors simulate future expected income using the "random walk with a drift" income simulation method. They find that this method introduces important measurement errors into estimates of future taxable income, which in turn significantly affect the MTR calculations. In particular, the standard deviation of future taxable income is, on average, too low under the random walk income simulation (the random walk standard deviation is only about half as large as the actual standard deviation).

Using an alternative nonparametric estimation method, the authors identify the rough direction of the mismeasurement. Compared with conventional measures of MTRs, the revised estimated MTRs are higher for firms with relatively low current profitability and are lower for firms with relatively high current profitability.

These revised MTRs have important implications for understanding how taxes affect firms' financial decisions, among other issues. For example, the authors note that their MTR estimates indicate that very few firms could increase their debt by more than a factor of two and still earn tax benefits at the top statutory rate. Thus, firms may not necessarily be underleveraged.


A Tar Heel Tradition


Prof. Douglas A. Shackelford, Meade H. Willis Distinguished Professor of Taxation at the University of North Carolina's Kenan-Flagler Business School, organized the 11th annual UNC Tax Symposium, which brought together accountants, economists, and lawyers from academe, practice, and the government who share a common interest in tax research.

Douglas ShackelfordAs Shackelford said:


      The purpose of the UNC Tax Center is to provide a point of connection for tax scholars, practitioners, and policymakers. The three groups often work in parallel universes, interested in the same issues, but approaching them differently and in isolation. The Center provides a neutral setting for the three groups to interact to their mutual betterment. My experience has been that the interaction leads to mutual understanding, respect, and cooperation with a goal of better tax education, practice, and policy.

Shackelford also announced the UNC Tax Doctoral Seminar, which will hold its inaugural meeting January 5-9, 2009. The seminar is designed to encourage students who might not otherwise have considered tax research and education to develop an interest in undertaking a tax dissertation and entering the job market as a tax professor. In so doing, the Tax Center hopes to turn around the recent decline in tax faculty in business schools.

Copies of the symposium papers and information about the doctoral seminar are available at /taxbasehttp://areas.kenan-flagler.unc.edu/Accounting/taxsym/Pages/default.aspx.



Joann M. Weiner is a contributing editor to Tax Analysts.
E-mail:
jweiner@tax.org