Trends in Book-Tax Income and Balance Sheet Differences
William B. Trautman***
May 20, 2002
This paper was presented at the 2002 IRS Research Conference
*, ** Assistant and Associate Professors, respectively, at the University of Arizona
*** Senior Economist, Internal Revenue Service LMSB:SRPP
We especially appreciate the excellent and tireless efforts of Mike Dodd and Donald Lee,
LMSB Research East Senior Program Analysts, in preparing data for this project. We have
benefited from discussions with members of the Internal Revenue Service, LMSB, Research
East, including Charles Beauchain and Bunn Martin. We also wish to thank Charles Boynton,
Fred Cox, David Harris, Fred Naimoli, Thomas Petska and participants in the 2002 IRS
Research Conference for helpful comments. All opinions are our own and do not represent
those of the Internal Revenue Service. Data in this paper are protected by a data non-
disclosure agreement under the Internal Revenue Code and all statistics are presented in the
Trends in Book-Tax Income and Balance Sheet Differences
We use Compustat and tax return data to describe trends from 1991-1998 in differences
between book and tax measures of income and balance sheet amounts. Our primary findings
confirm that book-tax income differences are growing throughout the 1990s. Extending prior
work, we partition the sample to describe the differences by industry, global character and
profitability. Secondly, we compare Compustat financial statement assets and liabilities to the
book balance sheet reported on the tax return and find that the tax return amounts exceed the
financial statement amounts in the aggregate. We plan to investigate suggested explanations for
this excess, including differences in book versus tax consolidation reporting and off-balance
Key words: Book-tax differences, consolidation, off-balance sheet
JEL classifications: H25, H26, M4
As a result of growing interest from Treasury in corporate tax shelters, as well as IRS
interest in incorporating financial reporting data into the tax administration process, LMSB
Research East is conducting a firm-level study of book-tax differences and compliance risks.1
Recent government and academic studies report a growing aggregate gap between book
income and taxable income. The U.S. Treasury (1999) suggested that part of this gap may
result from corporations’ growing use of tax shelters, consistent with concerns raised by some
academics (Bankman 1999) and various members of Congress.2 However, other authors
caution that the increasing use of non-qualified stock option plans (which generate tax
deductions but not book expenses) may be responsible for a large portion of the perceived
growing book-tax gap (Manzon and Plesko, 2002, Hanlon and Shevlin, 2002, Desai, 2002).
Further, book-tax consolidation differences, particularly for multinational corporations, could
generate much of the gap (Mills and Newberry 2000, Manzon and Plesko 2002).
Prior research suggests that book-tax differences relate to firms’ tax and financial
reporting incentives, as well as to mechanical differences caused by known differences between
accounting standards and tax laws. Controlling for simple causes of book-tax differences such
as depreciation and foreign repatriation, Mills (1998) finds that tax deficiencies are higher the
1 We define book-tax differences generally as pre-tax book income less taxable income, or book assets (or
liabilities) less assets (or liabilities) on the tax return. As we discuss later, since there are many ways to
define book income or taxable income, specific definitions are a research design choice. When we refer to
differences between book income and taxable income, we call these book-tax income differences. We also
measure differences between book and tax measures of assets and liabilities for the first time, and we
describe these differences as book-tax balance sheet differences.
2 Representative Lloyd Doggett D-TX, sponsored both the Abusive Tax Shelter Shutdown Act of 1999, HR
2255, introduced in the House, 06/17/99, and the Abusive Tax Shelter Shutdown Act of 2001, HR 2520,
introduced in the House, 07/17/01. While both bills were referred to the House Ways and Means committee,
more book income exceeds taxable income. Mills and Newberry (2001) learn that public firms
(with greater financial reporting pressures) have larger absolute book-tax differences than
private firms: more positive when public firms are profitable and more negative when public
firms are unprofitable. Current research by Manzon and Plesko (2002) and others highlight the
need to carefully investigate sources of book-tax differences to separate explained from
We compare financial statement data to tax return data.3 We merge Statistics of Income
(Form 1120) data with Compustat financial statement data from 1990 through 1999. This paper
summarizes some of the main issues and provides descriptive evidence from the aggregate
book-tax comparisons, based on a panel of 1,579 firms from 1991 to 1998.
Our primary findings confirm results in prior studies that book-tax income differences
are growing throughout the 1990s. Extending prior work, we partition the sample to describe
the differences by industry, global character and profitability. Secondly, we compare Compustat
financial statement assets and liabilities to the book balance sheet reported on the tax return. To
our surprise, the tax return assets and liabilities exceed the financial statement assets and
liabilities. This phenomenon is strikingly large by the end of the sample period: $1.9 trillion of
assets and $900 billion of liabilities are not reported on the Compustat financial statement
compared to the book balance sheet on the tax return.4 We have anecdotal evidence that some
neither bill was passed. Finally, the Tax Haven and Abusive Tax Shelter Reform Act of 2002 S. 2339, is in the
Senate Finance Committee as of May 2002.
3 We are aware of a concurrent project by the Office of Tax Analysis, U.S. Department of Treasury to match
Statistics of Income data with Compustat data, and we look forward to further discussions with the SOI_CS
(Compustat) Matched File Project Team.
4 This difference could be even larger if the balance sheet on the tax return excludes the gross assets and
liabilities of controlled foreign corporations that are separately reported on Forms 5471. We are perfecting
these data and inquiring about common reporting practice.
companies are not as careful in preparing the consolidated balance sheet for the tax return as
they are for their public financial statements, possibly resulting in double-counting of assets and
liabilities on the tax return. In addition, based on preliminary discussions with IRS personnel, we
also understand that part of this difference could be related to off-balance sheet financing
resulting from structured transactions or special purpose entities. We intend to explore all
explanations in future research.
Financial versus tax reporting rules and incentives
Financial accounting standards and tax laws frequently provide specific, and often
different, rules for how to report income for book and tax purposes, even though both income
reports are based on the same underlying fundamental transactions. Some book-tax reporting
differences may be viewed as mechanical because they relate to clear differences in rules.
Examples of material book-tax differences generated by clear differences in rules are
depreciation, stock options and consolidation. We discuss the latter two in detail because they
present particular measurement challenges.
Hanlon and Shevlin (2002) provide a detailed discussion of the accounting treatment for
nonqualified stock options, and the difficulty such treatment presents in controlling for the book-
tax difference caused by stock option deductions. There is typically little book expense
recorded for stock options, but the company receives a tax deduction when the employee
exercises the option. The deduction is equal to the difference between the fair market value of
the stock and the option price at the date of exercise. The benefit for the deduction is not
recorded in tax expense, but is treated as an offset to the stock transaction in the stockholders’
equity account. Thus, both the difference between book income and taxable income and the
difference between tax expense on the books and tax on the return are similarly affected. Since
neither Compustat financial data nor the tax return delineate the stock option deduction,
constructing a large-sample control is difficult.
Manzon and Plesko (2002) avoid the difficulty by using financial statement data to
estimate taxable income before an option deduction. Thus, their comparison of book income to
derived taxable income is free of the stock option difference. Hanlon and Shevlin (2002)
conduct a small-sample study using hand-collected footnote data. Desai (2002) extrapolates
employee option exercises from Compustat’s Executive Compensation database (Execucomp),
which are available for 2000 firms since 1992. He compares these computations to a detailed
analysis of 150 firms and concludes that the estimates from Execucomp “are reliable estimates
for the aggregate levels of the impact of option exercises on the corporate tax base” (Desai,
16). We intend to use Execucomp to estimate the stock option deduction for our sample of
firms in future work.
Many U.S. corporations own part or all of other corporations. Financial reporting
standards and tax laws provide different rules for when related corporations should be
combined for reporting. The combined reporting is called consolidation, in which the individual
lines of income and expense are totaled across all consolidated entities, net of transactions
between related parties.
Financial consolidation is governed by Statement of Financial Accounting Standard No.
94, with numerous administrative interpretations by the Financial Accounting Standards Board.
Generally, the consolidated reporting group includes the parent corporation and all subsidiaries
(both domestic and foreign) in which the parent has more than 50% ownership. If the parent
corporation does not own 100% of the subsidiary, it subtracts from net income the portion of
the subsidiary’s earnings that is allocable to the minority shareholder interest. When a
corporation owns between 20 and 50 percent of another corporation, the parent’s financial
reports include its percentage interest in the net income of that entity as “net equity of
unconsolidated subsidiaries.” If the parent owns 20 percent or less of a corporation, then it only
includes the dividends of such corporation in book income.
These general rules have flexibility related to the control exercised by the parent.
Special Purpose Entities (SPEs) have recently received great publicity as a mechanism to avoid
financial consolidation.5 The corporation can benefit by excluding the assets and the associated
debt and equity of the SPE from the consolidated balance sheet, because such entities are
typically designed with higher leverage ratios. By keeping debt out of the consolidated balance
sheet, the company protects its credit rating. Further, SPE losses appear to remain outside the
consolidated income statement. Since we do not yet know whether companies typically treat
the SPEs as corporations or as partnerships for tax purposes (the check-the-box rules would
permit either treatment), we cannot definitively comment on the tax effect. However, if the SPE
is treated as a partnership, foreign SPE losses will be deductible on the U.S. consolidated tax
return and the high leverage typical of SPEs would generate large interest deductions.
Tax consolidation is governed by IRC Section 1501, under which affiliated groups may
elect to file a single consolidated return. Affiliated groups may consist of corporations that are
related through ownership of at least 80%. Only domestic corporations may be included in the
affiliated group. Corporations owned less than 80% are excluded completely from the
consolidated return and file their own separate returns.
Thus, several types of entity differences arise due to book-tax consolidation
differences.6 The following differences make financial statements more inclusive than tax returns:
1. The financial statement consolidates > 50% owned foreign subsidiaries that are
excluded from the tax return.
2. The financial statement consolidates > 50 to < 80% domestic subsidiaries that are
excluded from the tax return
3. For companies owned 20 to 50%, the financial statement includes the percentage
ownership in the net equity of companies.
On the other hand, the tax return is more inclusive than the financial statement in the
4. The consolidated tax return includes 100% of the income for all domestic subsidiaries
owned at least 80%, with no reduction for minority interest.7
5. The tax return includes dividends from unconsolidated subsidiaries, reduced by the
dividends received deduction for dividends from domestic corporations.
6. The tax return may include special purpose pass-through entities that are excluded from
the financial statement.
Consistent with consolidation differences existing for foreign subsidiaries, Mills and
Newberry (2000) show that average book income reported on the tax return, Form 1120,
5 See Financial Executives International (2002) for a discussion of SPE guidelines.
6 See Dworin (1985) for an early discussion of consolidation differences.
7 We assume that U.S. parents elect to file a consolidated return with all of their 80% owned subsidiaries.
While corporations may file separately, our experience with the Coordinated Industry Cases suggests that
affiliated groups elect to file consolidated returns.
Schedule M-1, lines 1 plus 2, falls between Compustat worldwide consolidated pretax income
and Compustat U.S. pretax income. Their finding for large industrial firms is consistent with
some, but not all, foreign income being repatriated and included in taxable income. We hope to
use information on dividend schedules and Form 5471 related to controlled foreign corporations
to construct a proxy for the unrepatriated foreign earnings.
There are no easy solutions for detecting and measuring consolidation differences for
domestic subsidiaries. Our anecdotal understanding from agents in the large-case audit program
(Coordinated Industry Cases) is that very large taxpayers do not have many > 50 to < 80%
owned domestic subsidiaries. We are less sure whether this is equally true for the full large and
midsize business (LMSB) program. For minority interest differences, we can adjust for the items
disclosed in the financial statements, including minority interest and equity in net earnings or loss
of nonconsolidated subsidiaries.
A recent trend is the increasing use of check-the-box regulations to choose freely
whether an entity is a corporation or a pass-through entity (like a partnership) for tax purposes.
The book use of special purpose entities to exclude the income, losses, assets and liabilities of
pass-through entities will be difficult to identify. We do not yet know how to detect book-tax
differences due to differing classification of corporation versus pass-through status, and we
welcome suggestions on how to address this issue.
Book and tax incentives to manage reporting
We discuss above how certain known differences in accounting standards versus tax
laws generate book-tax differences. However, both accounting standards and tax laws permit
flexibility in reporting decisions. This flexibility means that book-tax differences could be viewed
as potential indicators of either opportunistic financial reporting or tax compliance risk.
Although financial reporting principles are designed to provide relevant and reliable
information to financial statement users, managers may prepare such reports opportunistically.
Financial reporting principles emphasize consistency over time within a firm, but they permit
considerable flexibility in the choice of methods and discretion in estimation, particularly when
the information is not deemed to be ‘material’, i.e. of sufficient magnitude to affect a user’s
decision.8 Independent auditors are necessary because managers may opportunistically use the
discretion granted by financial accounting principles. Typically, users are concerned that
managers will overstate income and assets.9
In contrast, the IRS must audit tax returns to detect and deter underreporting. Tax laws
exist primarily to raise government revenues. IRC Section 446(a) states that “taxable income
shall be computed under the method of accounting on the basis of which the taxpayer regularly
computes his income in keeping his books,” but IRC Section 446(b) permits the IRS to
disallow accounting methods that do not ‘clearly reflect income.’10 Tax law only requires
conformity with financial reporting in the case of last-in first-out (LIFO) inventory. Since firm
managers generally prefer to pay less tax (to increase cash flows) and report lower tax expense
(to increase reported financial earnings), potential underreporting represents a compliance risk.
8 See Manzon and Plesko (2002) for an extended discussion of the application of the Statements of Financial
Accounting Concepts No. 1 and 2.
9 Managers may also face incentives to decrease book income opportunistically. For example, firms with
higher income than projected may use discretionary accruals to smooth income downward, building
reserves (called ‘cookie jar’ accounting by the SEC) to use in the future to manage earnings upward.
Alternatively, firms in loss years may further decrease earnings (called taking a ‘big bath’), also creating
additional reserves for future use.