A Comment on Pleskoe’s “An Evaluation of Alternative Measures of Corporate Tax Rates”

Terry Shevlin

Informations

• Publié par: Irgu
• Nombre de lectures: 119
• Langue: English

Extrait

A Critique of

Plesko’s “An Evaluation of Alternative Measures of Corporate Tax Rates”

Terry Shevlin

University of Washington

October 1999

Abstract

In a recent working paper, Plesko (1999) uses confidential tax return data to evaluate

alternative measures of corporate average and marginal tax rates and concludes “The

results suggest that commonly used measures of average tax rates provide little insight

about annual corporate tax burdens, and may introduce substantial bias into statistical

models.

Marginal tax rate proxies perform better, but there appears to be little, if any,

empirical value added by methods that go beyond easily constructed measures

traditionally used in the literature.”

I caution readers in accepting the above inferences

because of conceptual flaws in Plesko’s evaluation.

In evaluating financial statement

based average tax rates, Plesko assumes researchers are attempting to estimate statutory

tax burdens (tax payable as a percent of taxable income) whereas I would argue that most

researchers are attempting to examine a more general concept of corporate tax burdens

(tax payable as a percent of book income).

Which tax burden is appropriate depends on

the research question but Plesko assumes the statutory tax burden is always the

appropriate measure.

In evaluating financial statement based marginal tax rates, Plesko

uses a single period measure which fails to take into account the effects of carryback and

carryforward rules in calculating taxable income.

Thus his marginal tax rate benchmark

is flawed.

JEL classifications: G3, H25, M4, C200

Email:

[e-mail]

2

A Critique of

Plesko’s “An Evaluation of Alternative Measures of Corporate Tax Rates”

Introduction

As noted in a recent working paper by Plesko (1999), researchers outside the IRS must

rely on alternative sources of information to examine most tax-related hypotheses.

One

source of information used is financial statement information.

Plesko states that “The

maintained hypothesis throughout this paper is that, were it available, researchers would

choose to examine and utilize firms’ tax return information to capture tax attributes rather

than rely on the transmission of such information through the financial statements.” (p.2)

Plesko uses confidential tax return data to evaluate alternative measures of corporate tax

rates and concludes “The results suggest that commonly used measures of average tax

rates provide little insight about annual corporate tax burdens, and may introduce

substantial bias into statistical models.

Marginal tax rate proxies perform better, but there

appears to be little, if any, empirical value added by methods that go beyond easily

constructed measures traditionally used in the literature.”

These are strong claims but

their validity is open to question because of conceptual flaws in Plesko’s evaluation.

Plesko evaluates two types of tax rate measures that are are found in the literature:

average tax rate measures and marginal tax rate measures.

As explained in more detail

below, the conceptual flaw in the evaluation of average tax rate measures is that Plesko

assumes researchers using financial statement data to estimate average tax rates are trying

to estimate statutory tax burdens (which he defines as tax payable as a percent of taxable

income).

However, in many contexts the financial statement based measure is

appropriate.

For example, if the research question is what percentage of the firm’s

reported book income did the firm pay in tax, the financial statement based measure is

(obviously) appropriate.

The conceptual flaw in the evaluation of the marginal tax rate

measures is that Plesko uses a single period measure as his benchmark true marginal tax

rate when marginal tax rates should reflect the present value of taxes payable arising from

earning another dollar of income in the current period.

Marginal tax rates require the

present value concept because of the carryback and carryforward rules affecting taxes

payable on an incremental dollar of income earned in the current period.

Thus his

benchmark is flawed.

I discuss these flaws in more detail for each tax rate measure.

1

Average tax rate measures

As noted by Plesko, average tax rate (ATR) measures are used by researchers to assess

corporate tax burdens. There have been numerous studies and comparison/critiques of

ATR measures,

2

however, Plesko is the first to compare the financial statement based

measures with tax statement measures.

An ATR is calculated as

1

Since I believe the two flaws I discuss are critical to the inferences Plesko draws, I restrict my critique to

these two issues.

That is, my critique is not intended as a comprehensive critique or thorough review of

Plesko (1999).

2

See, for example, Omer, Molloy, and Ziebart (1991) for a summary of prior studies and an evaluation of

alternative measures.

3

income

year

for the

due

taxes

.

Using financial statement data, the numerator can be either current tax expense or total

tax expense which is the sum of current tax expense and deferred tax expense.

The

denominator is some measure of book income with various measures being used in prior

studies.

Plesko examines five different measures as summarized in his Table 1.

Plesko

compares these financial statement ATRs to two tax-based ATR measures with the

numerator in both measures being taxes due for the year after credits and the denominator

being alternative measures of ‘taxable income.’

As noted by Plesko “Even with tax

return information, defining the ATR is not straight forward as the final measure of

income for a corporation in any year is reached by calculating a number of intermediate

definitions of income.”

(p.7)

3

These latter two measures are estimates of the statutory

tax burden of corporations.

Plesko reports a series of regression models and I have no query or criticism of the

regression models or the estimated coefficients.

Rather, my concern and thus caution to

other readers is in Plesko’s inferences from these results.

“Taken as a whole, the results

suggest average tax rates based upon financial statement information are not very

informative in inferring federal tax burdens.” And “Further, given the effects of

correlated measurement error in the dependent variable, it is not apparent what

conclusions can be made from prior studies which have attempted to estimate the sources

of variation in average tax rates.” (p.19)

If these inferences are correct, we are indeed in

serious trouble!

However, this is where Plesko’s maintained hypothesis is important.

He

assumes the tax–based estimates are the ‘true’ measures (otherwise how can we talk

about measurement error and calibrate the effects of measurement error).

Putting aside

that Plesko examines two alternative tax-based measures suggesting some doubt about

which tax-based measure we will take as being true for purposes of measurement error

issues, the issue really is that the appropriate measure (financial statement based or tax-

return based) of ATRs depends on the context and research question being examined.

It

is not the case that the statutory tax burden is always the correct (or true) measure.

I

discuss some contexts where I believe financial statement ATRs are appropriate.

This list

is not exhaustive.

In a widely cited series of studies, The Citizen’s For Tax Justice (1984, 1985, 1986) using

financial statement data highlighted that many large firms were paying zero or very low

taxes while reporting large earnings to shareholders (that is, the financial statement ATR

was zero or very low).

These reports are credited by some (e.g., Birnbaum and Murray

1987) as directly influencing Congress in the Tax Reform Act of 1986 (TRA 86) to

broaden the tax base and introduce, for several years, a book income adjustment to the

corporate alternative minimum tax (the so-called AMTBIA).

4

Thus any study wishing to

evaluate the CTJ studies can legitimately examine financial statement based ATRs (see,

for example, Gupta and Newberry 1992, Kern and Morris 1992 and Shevlin and Porter

1992).

Further, since the CTJ study appeared to influence tax policy around TRA 86

3

In the first tax-based measure, the denominator is taxable income before net operating loss and special

deductions (referred to as the TAXNI measure) and in the second measure the denominator is income

subject to tax (TACIST).

4

See Gramlich (1991) for a discussion of the AMTBIA.

4

(and I note the

Wall Street Journal

August 4 1999 relates taxes paid to corporate earnings

as well as to taxable income in discussing current tax issues), one can make a reasonable

argument that financial statement based ATRs are relevant in examining corporate tax

burdens (as opposed to statutory tax burdens) for tax policy related issues.

See also

Gupta and Newberry (1992, notes 1 and 2) for references to studies of financial statement

based ATRs by the Joint Committee of Taxation and the U.S. General Accounting Office.

An early motivation for examining ATRs was to provide evidence on the political cost

hypothesis (Zimmerman 1983, Porcano 1986, and Wilke and Limberg 1990).

This

hypothesis predicts that larger firms faced more political scrutiny and thus were more

likely to select income decreasing accounting methods (and accruals).

One measure of

political scrutiny is reflected in the taxes paid by different size firms.

Specifically, do

larger firms face higher tax burdens?

Alternatively stated, do larger firms pay a larger

proportion of their earnings as taxes?

Stated this way, it is very reasonable to compare

the percent of earnings paid in taxes across different size firms and thus to use financial

statement based ATRs.

ATRs are also used to examine tax planning effectiveness (and also tax aggressiveness).

More effective tax planners are expected to exhibit lower ATRs.

But if we examine

statutory tax burdens, the burden can only differ from the top statutory tax rate because of

credits (I am assuming here that the effect of progressivity is minor for the average public

firm usually studied).

That is, any tax planning that defers revenue or accelerates tax

deductions reduces taxable income but has no effect on the statutory tax burden.

Thus,

statutory tax burdens do not reflect the effects of effective tax planning.

On the other

hand, tax planning that defers revenue and accelerates deductions for tax purposes but not

financial reporting purposes (that is, timing differences) will result in lower financial

statement ATRs (where the denominator is current tax expense and not total tax expense).

That is, tax planning that result in timing or permanent differences will have little effect

on statutory tax burdens but will show up in financial statement ATRs.

Also studies

examining the effects of compensation schemes which are based on after-tax accounting

numbers also are also correct in using financial statement based measures (see, for

example, Philips 1999).

Past researchers examing financial statement based measures with any accounting

knowledge were well aware that book income measures differ from taxable income (due

to timing and permament differences) and thus were not purporting to examine firms’

statutory tax burdens.

5

To the extent they were, Plesko’s comments and inferences on

ATRs are valid.

Since the purpose of calculating financial statement income (as a

measure of firms operating performance) is different than the purpose of calculating

taxable income (a base on which to tax the firm), it is arguable as to which income

measure is a better measure of a firm’s performance.

And if in assessing corporate tax

burdens we wish to relate current taxes payable to the firm’s current performance, a

strong argument can be made for financial statement based measure(s) of earnings.

Thus

it is arguable as to which measure of income (tax or book) is a better denominator in

estimating a firm’s corporate tax burden and the answer, as noted above, depends on the

context and research question.

I, and many financial and tax accountants, would argue

5

See Plesko (1999) or Gupta and Newberry (1992) for the definition and a discussion of timing and

permanent differences.

5

that accounting income is a better measure of a firm’s operating performance and that

financial statement based measures of ATRs are better measures of corporate tax burdens

as distinct from being measures of the statutory tax burden.

Most financial and tax

accountants would agree that financial statement ATRs are likely poor measures of

statutory tax burdens (because of the differences in the calculation of the income used in

denominator).

6,7

Marginal tax rate measures

Corporate marginal tax rate (MTR) measures are used in studies examining tax motivated

corporate behavior.

Consistent with Scholes and Wolfson (1992, p.146) the corporate

marginal tax rate is generally defined as the change in the present value of the cash flow

paid to (or recovered from) the tax authorities as a result of earning one extra dollar of

taxable income in the current tax period.

This is a present value concept since U.S.

corporate tax law treats gains and losses asymmetrically by taxing income for the current

period at statutory rates only when positive.

Losses may be carried back to obtain

refunds of previously paid taxes or carried forward to be offset against future taxes

payable.

Because of this asymmetric treatment, gains and losses from other years have

the potential to reduce firms’ current period MTR.

In other words, the current period

MTR is dependent on the firms’ taxable income in prior and future years.

Graham (1996b) evaluates alternative proxies relative to a simulation based method

developed by Shevlin (1987, 1990) and extended by Graham (1996a).

The simulation

approach involves repeated simulation of the firm’s future (financial statement estimated)

taxable income to incorporate the carryback and carryforward rules and thus captures the

present value or multiperiod nature of MTRs.

Based on Graham’s evaluation and the

posting of simulated MTRs on Graham’s website, more and more researchers are using

the simulated rates.

It is important to note that the simulation approach requires

assumptions to implement and the estimates differ depending on the assumptions.

However, we do not know how ‘good’ a proxy these rates are.

Plesko (1999) attempts an

evaluation but uses only one period of taxable income and ignores the multiperiod effects

of the asymmetric treatment of gains and losses.

Thus his measure is a single period

measure.

To see the extent of the problem, if the firm has negative taxable income (a tax

loss) that has to be carried forward, it appears as though Plesko assigns the firm an MTR

of zero when, in fact, if the tax losses are used in the next period the

6

If past researchers were attempting to estimate the statutory tax burden, then an interesting comparison

would be with a financial statement based estimate of taxable income (as per Guenther (1992) where pretax

book income is adjusted for timing and permanent differences to estimate taxable income) with the tax

return taxable income.

7

There are other problems with calculating financial statement ATRs which are not germane to this

comment.

One of these problems occurs when tax payable is positive and income is negative or tax

payable is negative (the firm obtains a refund) and income is negative or positive.

The traditional ATR is

not well defined in these situations.

Wilkie and Limberg (1993) discuss this problem and suggest an

alternative measure of the corporate tax burden.

Also the ATRs here focus only on explicit taxes and

ignore implicit taxes.

Finally, the numerator of the ATR includes only current taxes payable, and it can be

argued that the present value of deferred taxes should be included to obtain a better measure of the percent

of taxes due on the current period accounting earnings.

However, calculating present value is not an easy

task.

6

MTR

t

=

r)

(1

str

1

t

+

+

.

If the statutory tax rate in period t+1, str

t+1

, is 35% and the after-tax discount rate, r, is

10%, then MTR

t

= 32% which is far greater than the 0% assigned by Plesko.

Thus the

tax-based one-period MTR used by Plesko as the benchmark (which he assumes

represents the true marginal tax rate) is conceptually flawed and thus any inferences

based on this measure are flawed (and misleading).

Plesko could (partially) address the conceptual flaw by using a time-series of tax data for

the firm (with the time-series being a function of the carryback and carryforward rules,

which were changed in the Taxpayer Relief Act of 1997 from a 3 to 2 year carryback

period and from 15 to 20 year carryforward period).

Using a time-series would take into

account the effects of the carryback and carryforward rules on the current period MTR.

However, a remaining problem with all attempts to evaluate MTR proxies is that a firm’s

MTR is truly unobservable for reasons pointed out in Shevlin (1990, note 8).

The reason

for this is that any taxable income realizations beyond period t are the result of actions

taken by the firm in period t in response to its tax status.

That is, the future period

taxable income realizations are endogenous to the period t MTR.

To the extent the researcher is willing to assume this endogenity is of second order

magnitude, then future taxable income realizations could be used to calculate a present

value MTR measure.

With this measure, an evaluation of the alternative MTR proxies

could be conducted and if results similar to Plesko were observed (that two binary

variables seem to capture most of the variation in the tax-based present value MTR) then

I think this would be a valuable contribution to the literature.

That is, use of simple

binary variables, or even a combination of the two binary variables, in lieu of the more

complex simulation approach would be easier to implement.

Finally any measurement error really should be defined with respect to the concept or

MTR measure that firms actually use in decision making, and a worthwhile endeavor

would be to document (possibly by field study) how firms incorporate their tax status into

their decisions.

It is possible that managers use a simple binary measure based on the

sign of taxable income or a slightly more sophisticated measure that combines the sign of

taxable income and NOL status.

In this context, the simulation approach by estimating a

continuous rate would contain much measurement error.

Conclusion

To summarize, Plesko attempts to provide an evaluation of financial statement based

average and marginal tax rates.

The inferences he draws from both sets of evaluations

are flawed (or at best, must be tempered) by conceptual errors underlying his analysis.

With respect to the average tax rate evaluation, financial statement based estimates of

corporate tax burdens are fine conditional on recognizing that they are not estimates of

the statutory tax burden.

Whether tax-based or financial statement based ATRs are

appropriate depends on the context and research question and to assume that the tax-

based measure is always the desired or correct measure is incorrect.

I have provided

several scenarios in which financial statement based ATRs are desired (and thus the tax-

based measures contain measurement error in these settings!).

With respect to the

7

marginal tax rate evaluation, Plesko uses a single period measure of the marginal tax rate

which ignores the effects of the carryback and carryforward rules which in my opinion

are key drivers of cross-sectional variation in corporate marginal tax rates.

Thus the

benchmark is incorrect.

Thus I disagree with the following conclusions by Plesko (1999):

“Given the extent to which tax rates appear to be mismeasured by publicly available data,

and the econometric implications of mismeasurement of variables, any prior finding of

tax effects may be viewed as surprising.” and

“The differences in calculated tax rates across the two sources of data are striking and

have a number of obvious, and not so obvious, interpretations and implications.

Given

the inability to interpret both coefficients and significance in these cases, financial data

does not appear to provide a strong foundation on which to judge the effects of taxation

or build arguments for tax policy.” (p.27)

Strong claims indeed but which unfortunately do not hold up to close scrutiny for the

reasons outlined in this critique.

8

References

Birnbaum, J. H. and S. A Murray. 1987.

Showdown at Gucci Gulch, Lawmakers,

Lobbyists, and the Unlikely Triumph of Tax Reform

. Random House.

Citizens for Tax Justice. 1984.

Corporate Income Taxes in the Reagan Years

.

Washington, D.C: Citizens for Tax Justice.

Citizens for Tax Justice. 1985.

Corporate Taxpayers & Corporate Freeloaders

.

Washington, D.C: Citizens for Tax Justice.

Citizens for Tax Justice. 1986.

130 Reasons Why We Need Tax Reform

. Washington,

D.C: Citizens for Tax Justice.

Graham, J. 1996a. “Debt and the Marginal Tax Rate,”

Journal of Financial Economic

s,

41-74.

Graham, J. 1996b. 1996, “Proxies for the Marginal Tax Rate,”

Journal of Financial

Economics

, 187-221.

Gramlich, J. 1991. “The Effect of the Alternative Minimum Tax Book Income

Adjustment on Accrual Decisions.”

Journal of the American Taxation Association,

Spring, 36-56.

Guenther, D. A. 1992. “Taxes and Organizational Form: A Comparison of Corporations

and Master Limited Partnerships,”

The Accounting Review

, January, pp.17-45.

Gupta, S and K. Newberry. 1992. “Corporate Average Effective Tax Rates After the Tax

Reform Act of 1986,”

Tax Notes

. May 4 689-702.

Kern, B.K. and M.H. Morris. 1992. “Taxes and Firm Size: The Effect of Tax Legislation

During the 1980s,”

Journal of the American Taxation Association.

Spring

80-96.

Omer,T

.,

K. Molloy and D. Ziebart. 1991. “Using Financial Statement Information in the

Measurement of Effective Corporate Tax Rates,”

Journal of the American Taxation

Association.

Spring

57-72.

Phillips, J.D. 1999. “Corporate Tax Planning Effectiveness: The Role of Incentives,”

Working paper (University of Connecticut, Storrs, CT, August).

Plesko, G.A. 1999. “An Evaluation of Alternative Measures of Corporate Tax Rates,”

Working paper (MIT, Boston, MA, July).

Porcano, T. 1986. “Corporate Tax Rates: Progressive, Proportional, or Regressive,”

Journal of the American Taxation Association.

Spring

17-31.

9

Scholes, M. and M. Wolfson. 1992.

Taxes and Business

Strategy: A

Planning Approach.

Engelwood Cliffs, NJ: Prentice-Hall, Inc.

Shevlin, T. 1987. “Taxes and Off-Balance-Sheet Financing: Research and Development

Limited Partnerships,”

The Accounting Review,

July, 480-509.

Shevlin, T. 1990. “Estimating Corporate Marginal Tax Rates with Asymmetric Tax

Treatment of Gains and Losses,”

Journal of the American Taxation Association,

Spring,

51-67.

Shevlin, T. and S. Porter. 1992. “The Corporate Tax Comeback in 1987: Some Further

Evidence,”

Journal of the American Taxation Association,

Spring, 58-79.

Wilkie, P. and S. Limberg. 1990. “The Relationship Between Firm Size and Effective

Tax Rate: A Reconciliation of Zimmerman (1983) and Porcano (1986),”

Journal of the

American Taxation Association,

Spring, 76-91.

Wilkie, P. and S. Limberg. 1993. “Measuring Explicit Tax (Dis)Advantage for Corporate

Taxpayers: An Alternative to Average Effective Tax Rates.”

Journal of the American

Taxation Association,

Spring, 46-71.

Zimmerman, J. 1983. “Taxes and Firm Size,”

Journal of Accounting and Economics

,

August, 119-149.