The role of profits and income in the statistical discrepancy.
Rassier, Dylan G.
THE NATIONAL income and product accounts (NIPAs) of the Bureau of
Economic Analysis (BEA) include two alternative measures of economic
output: gross domestic product (GDP) and gross domestic income (GDI).
GDP is an expenditure-based measure and is estimated based on spending
on final goods and services. GDI is an income-based measure and is
estimated based on income generated in the production of goods and
services. Before the recession that began in the fourth quarter of 2007
and ended in the second quarter of 2009, GDI growth was generally lower
than GDP growth, which has generated discussion about whether the source
data and adjustments that underlie GDP reflect enough economic
cyclicality. This article explores an alternative: whether the source
data and adjustments that underlie GDI reflect too much economic
cyclicality and whether this effect may explain a significant share of
the difference between GDP and GDI during the downturn. In particular,
this article identifies and explains the following four factors that
require adjustments to convert financial- or tax-based source data into
economic accounting measures of corporate profits and proprietors'
and partnership income (hereafter, profits and income):
* Misreporting on tax returns comprises a significant portion of
profits and income, and while there may be reasons to think that
misreporting would be cyclical in nature, long-term enforcement efforts
may well offset any cyclicality in reporting, and the overall cyclical
effect of adjustments for misreporting is ambiguous.
* Capital gains and losses may be leaking into measures of profits
and income, which could yield overly cyclical measures of profits and
income.
* Inconsistent measurement of stock options in source data may
generate an overly cyclical measure of GDI relative to GDP.
* Assumptions regarding the capitalization rate of purchased
software may overstate profits and income during cyclical upturns.
For each of these factors, additional research is warranted and
ongoing to understand the potential contribution of each factor to
differences between GDP and GDI.
BEA prepares current quarterly estimates and annual estimates of
GDP and GDI, each of which is subject to several revisions. (1) Each
revision yields a new vintage of estimates. Each successive current
quarterly vintage incorporates newly available source data. Each annual
vintage incorporates newly available source data as well as improved
estimating methodologies. In addition to revising annual estimates, an
annual revision includes revisions to the quarterly estimates.
Approximately every 5 years, BEA prepares a benchmark revision, which
incorporates newly available, comprehensive source data, improved
estimating methodologies, and definition changes.
In concept, GDP and GDI are equal because expenditures by one party
in the economy become income to another party. In practice, all vintages
of GDP and GDI are estimated from largely independent and incomplete
source data, so the errors in each measure are not the same. Vintages
face tradeoffs between timeliness and accuracy. While current quarterly
vintages offer the timeliest look at economic output, the accuracy of
current quarterly vintages is affected more than the accuracy of later
vintages by the completeness and reliability of the underlying source
data. Annual and benchmark revisions improve accuracy, but the resulting
vintages are less timely than the current quarterly vintages. (2)
Regardless of the vintage, discrepancies between GDP and GDI are
introduced through differences in source data, adjustments of source
data to economic concepts and definitions, differences in interpolation and extrapolation techniques, and differences in the timing of quarterly
seasonal adjustments. Thus, even after annual and benchmark revisions
are incorporated, differences between GDP and GDI persist in annual and
quarterly estimates. However, recent work shows that GDP and GDI levels
follow the same trend and rarely drift far from each other
(Greenaway-McGrevy 2011).
The difference between GDP and GDI is known as the statistical
discrepancy. Both GDP and GDI provide a complete picture of economic
output, so the statistical discrepancy provides an indication of net
measurement error. In addition, changes in the degree and direction of
the statistical discrepancy from one period to another reflect
differences in the rates of growth between GDP and GDI. Internationally,
statistical agencies generally choose one of two alternatives to handle
the statistical discrepancy. One alternative is to publish a statistical
discrepancy as a separate line item in the national accounts based on
the relative reliability of the underlying source data. A second
alternative is to allocate the discrepancy to the components of output
where measurement errors are most likely to exist. (3) BEA follows the
first alternative and publishes a statistical discrepancy as a line item
with aggregate GDI. The resulting double-entry accounts yield a
breakdown by component of GDP and GDI in addition to an indication of
the consistency between the two sides of the accounts.
BEA recognizes strengths and weaknesses of both GDP and GDI as
measures used to analyze economic activity and business cyclicality.
However, a decision ultimately has to be made regarding which side of
the national accounts to record the statistical discrepancy. Given the
important role that corporate profits and proprietors' and
partnership income play as components of GDI and given the challenges of
adjusting source data to accord with economic accounting measures of
profits and income, BEA records the statistical discrepancy with GDI in
order to reflect the relative reliability of the source data and
required adjustments underlying GDI relative to GDE Thus, this article
focuses on income-side factors and adjustments that have led to concern
at BEA about possible measurement error. In particular, this article
focuses on the recent behavior of the statistical discrepancy and
potential cyclicality of the measurement error in profits and income.
Table 1 shows the component shares of GDI for the 5-year period
from 2006 to 2010 using quarterly BEA data from the 2011 annual revision
of the NIPAs, which provides the most recent vintage of estimates. This
period includes the most recent recession (from the fourth quarter of
2007 to the second quarter of 2009) as determined by the Business Cycle
Dating Committee of the National Bureau of Economic Research (NBER).
Profits and income generally account for 15 to 20 percent of GDI. Table
1 shows that profits declined sharply through the fourth quarter of
2008, and income declined slightly through the second quarter of 2009.
The declines in profits and income were offset in part by increases in
the shares of compensation, other net operating surplus, and consumption
of fixed capital. Table 1 does not provide insight regarding measurement
error but does indicate profits and income account for a
larger-than-proportionate share of any declines in GDI during the
recession. (4)
In addition to profits and income, other components of GDI are
subject to measurement error. Likewise, measurement error affects all
components of GDP. As a result, some recent studies have questioned
BEA's decision to record the statistical discrepancy with GDI and
the resulting emphasis on GDP in news releases (for example, see Klein
and Makino 2000; Fixler and Nalewaik 2009; Nalewaik 2010). As a steward
of the U.S. national accounts, BEA does not intend to promote one
measure over another, and the recent work generally supports BEA's
conclusion that both GDP and GDI are valid measures of output. However,
how much each measure might be weighted in a combined measure and
whether temporal variation should be assumed to indicate less
reliability or more reliability remains unresolved and, in some cases,
may run contrary to other studies (for example, see Weale 1992; Smith,
Weale, and Satchell 1998; Grimm and Parker 1998; Fixler and Grimm 2002
2005; and Greenaway-McGrevy 2011). In addition, most of the work to date
on combining GDP and GDI focuses on weighting aggregate measures rather
than on weighting the underlying source data. As an alternative, BEA is
currently conducting research on weighting the underlying source data
based on reliability in a model to distribute the statistical
discrepancy before aggregating the component estimates (for example, see
Chen 2006, 2010). While weighting the underlying source data receives
strong support from a theoretical perspective (for example, see Stone,
Meade, and Champernowne 1942; Byron 1978), the practicality and
feasibility of weighting the underlying source data have yet to be
determined. (5)
The first section presents an accounting framework to describe the
role of profits and income in the statistical discrepancy. The section
also discusses empirical evidence to explain the focus on profits and
income rather than on other components of GDI and GDE The next section
identifies and explains the factors of profits and income that are most
likely to contribute to the statistical discrepancy. The final section
offers some concluding observations.
Profits and Income and the Statistical Discrepancy
Accounting framework
To provide some conceptual context, we follow Klein and
Makino's (2000) construction of the national accounting identity.
(6) The expenditure-based measure of output can be written as the sum of
consumer expenditures (C), investment (I), government expenditures (G),
and exports (X) less imports (M). Likewise, the income-based measure of
output can be written as the sum of wages (W), profits and income (P),
rents and interest (R), and taxes on production (T) less subsidies (S).
Thus, if aH these variables are measured in accordance with economic
accounting principles, the accounting identity for output is as shown in
the following equation:
[C.sup.*] + [I.sup.*] + [G.sup.*] + [X.sup.*] - [M.sup.*] =
[W.sup.*] + [P.sup.*] + [R.sup.*] + [T.sup.*] - [S.sup.*] (1)
The left side of equation (1) captures all final expenditures on
goods and services in the economy, and the right side captures all
income accruing to the input factors used for the production of the
goods and services. The asterisks in equation (1) indicate components
measured without error. In practice, each of the components is usually
estimated from independent and incomplete source data. In addition, at
least some of the components in equation (1) are estimated from source
data that are not consistent with economic accounting concepts and thus
require adjustments. Thus, the identity is inevitably not satisfied,
resulting in a statistical discrepancy (SD) as follows:
SD = (C + I + G + X - M) - (W + P + R + T - S) (2)
In equation (2), asterisks are removed to reflect measurement error
in each of the components.
Klein and Makino (2000) point out that firm-level profits and
income ([PI]) are never directly estimable but are merely a residual
between sales and costs as follows: (7)
[PI] = Sales - Costs (3)
From a financial accounting perspective, the results of equation
(3) may vary across firms because of flexibility in the application of
financial accounting rules. Likewise, the results of equation (3) may
vary between financial and tax accounting records within a firm because
of differences between financial and tax accounting rules.
From an economic accounting perspective, a measure of profits and
income from equation (2) can be obtained by calculating the residual
between the measured expenditure-based components and the measured
income-based components other than profits and income as follows:
[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] (4)
In equation (4), the expenditure-based components in the first set
of parentheses correspond to sales in equation (3), and the income-based
components in the second set of parentheses correspond to costs. If the
economic accounting measure of profits and income is determined by
equation (4), the statistical discrepancy is allocated entirely to
profits and income, which is most likely incorrect.
Alternatively, a measure of profits and income for equation (2) can
be obtained by aggregating profits and income for each firm j from
equation (3) as follows:
P = [summation over (j)] [[PI].sub.j] + Adjustments (5)
The adjustments in equation (5) are required in order to obtain an
economic accounting measure of profits and income based on source data
that might be inconsistent with economic accounting concepts. In
contrast to equation (4), equation (5) provides a check on the accuracy
of the other measured components in equation (2), and uncertainty
remains as to the allocation of the statistical discrepancy.
Practical considerations
In practice, GDP and GDI in the NIPAs are estimated from largely
independent and incomplete source data. In addition, profits and income
in the NIPAs are estimated according to equation (5) rather than
equation (4). As summarized in equation (2), the statistical discrepancy
reflects the completeness and reliability of the underlying source data
and the consistency of the underlying source data with economic
accounting concepts for each of the measured expenditure-based and
income-based components. In equation (5), adjustments are required to
manage incomplete and less reliable source data and source data that are
inconsistent with economic accounting concepts. Adjustments are
generally required for all expenditure-based and income-based components
in the NIPAs, but all quarterly and annual vintages of GDI are generally
subject to more adjustments than corresponding vintages of GDP (Grimm
and Weadock 2006; Holdren and Grimm 2008). With some exceptions, the
source data underlying current quarterly estimates and annual estimates
of GDP are collected by the Census Bureau in a set of surveys that are
designed to provide data consistent with economic accounting concepts.
In contrast, the source data underlying current quarterly estimates and
annual estimates of profits and income are generally collected from
financial- and tax-based source data, which are often inconsistent with
economic accounting concepts. In addition to conceptual inconsistencies,
tax-based source data underlying profits and income are less timely than
the source data underlying GDE As a result, current quarterly estimates
and the first annual estimates of GDP are based on a more complete set
of source data than the estimates of profits and income for the same
vintages. (8) Given the important roles that profits and income play as
components of GDI and the challenges of adjusting financial- and
tax-based source data for completeness, reliability, and economic
accounting concepts, BEA records the statistical discrepancy with GDI,
reflecting the relative reliability of the source data and required
adjustments underlying GDI relative to GDP.
To be clear, components of GDI other than profits and income also
pose estimation challenges that likely contribute to the statistical
discrepancy. Likewise, estimation challenges exist for components of GDE
Related research has used aggregate BEA data and applied statistical
analyses to determine the contributions of GDI components and GDP
components to the statistical discrepancy. Klein and Makino (2000) use
regression analysis to estimate the statistical discrepancy for the
period 1947 to 1997; they found that in addition to profits and income,
the discrepancy is affected significantly by exports and government
expenditures. However, Klein and Makino's (2000) results were not
supported when using more recent vintages of estimates in Grimm (2007),
which used regression analysis to estimate the statistical discrepancy
for 1970 to 2004. Grimm (2007) found that the effect of any GDI or GDP
component on the statistical discrepancy is indeterminate for the period
because of multicollinearity.
Additional related research used aggregate BEA data to determine
whether GDI, GDP, or a combination of GDI and GDP offers a better
measure of true economic output. Fixler and Nalewaik (2009) made the
reasonable assumption that revisions to GDP and GDI add news to the
estimates (Mankiw and Shapiro 1986) and applied a revision decomposition for 1984 to 2005 to find that the idiosyncratic variation in GDI growth
was higher than the idiosyncratic variation in GDP growth after
revisions. Fixler and Nalewaik (2009) attributed the increased variation
in GDI growth to news and concluded that GDI should be weighted more
than GDP in a combined measure of output without suggesting a
combination of appropriate weights. In addition, Nalewaik (2010) applied
statistical tests to determine whether growth in GDP or GDI better
reflects business cyclicality in output growth for 1978 to 2009, and the
study concluded that GDI growth is a better measure of cyclicality
without providing a rigorous analysis for a weighted measure of output.
Greenaway-McGrevy (2011) applied a Kalman filter to determine true
economic output for the period 1983-2009 and concludes that the
measurement error of GDP is smaller than the measurement error of GDI.
Greenaway-McGrevy (2011) suggested that GDP should be weighted
approximately 60 percent and GDI should be weighted approximately 40
percent in a combined measure of output.
Other related research used disaggregated BEA data to determine the
distribution of the statistical discrepancy to GDI and GDP components.
Chen (2010) applied a generalized least squares (GLS) model to
distribute the statistical discrepancy in 2002 to the components of
expenditure-based GDP in the NIPAs, the components of value added in the
income-based GDP by industry accounts, and the gross output and
intermediate inputs of the input-output accounts based on the relative
reliabilities of underlying source data in all the accounts. She found
that the optimal adjustments to gross output, intermediate inputs, and
GDP components are relatively small, while the optimal adjustments to
value added are relatively large due to the relatively low reliability
of tax-based source data and adjustments included in the gross operating
surplus component of value added. (9) In earlier work, Chen (2006)
applied a GLS model to distribute the statistical discrepancy in 1997 to
the components of value added in the income-based GDP by industry
accounts and to gross output and intermediate inputs of the input-output
accounts. In contrast to Chen (2010), the components of
expenditure-based GDP were held fixed in Chen (2006). Similar to Chen
(2010), Chen (2006) found relatively small adjustments to gross output
and intermediate inputs and relatively large adjustments to value added
given the relative reliabilities of the underlying data.
In sum, the studies that used disaggregated data with statistical
analyses have yielded a consistent set of results and conclusions,
whereas studies that used aggregated data with statistical analyses have
so far yielded a mixed set of results and conclusions. In other words, a
bottom-up approach may be necessary to draw conclusions about the extent
to which the statistical discrepancy is likely to be attributable to
expenditure-based and income-based components. Thus, from a practical
perspective, BEA must rely on its experience with underlying source data
in its decision to record the statistical discrepancy with GDI.
Cyclicality
The behavior of the statistical discrepancy may look different on a
quarterly basis than on an annual basis because quarterly variation is
netted out in annual estimates. (10) The behavior of the statistical
discrepancy during cyclical turning points is particularly important to
policymakers and other decisionmakers because differences between GDP
and GDI can complicate the decisionmaking process. Likewise, when it
comes to making real-time decisions, current quarterly vintages of GDP
and GDI are timelier than vintages based on annual and benchmark
revisions. Regardless of the vintage, the cyclicality of underlying
source data and required adjustments affecting GDP and GDI is important
to consider in the decision about where to record the statistical
discrepancy because source data and adjustments that are overly cyclical
or not cyclical enough are likely to yield a less accurate measure of
economic output.
Chart 1 presents the quarterly statistical discrepancy as a percent
of GDP for the 5-year period 2006-2010. This period includes the
recession that began in the fourth quarter of 2007 and ended in the
second quarter of 2009, as determined by NBER's Business Cycle
Dating Committee. Before the recession that began in 2007, the quarterly
statistical discrepancy changes from relatively large and negative in
the first quarter to relatively large and positive by the last quarter.
With the exception of the period from the first quarter of 2006 to the
first quarter of 2007, the quarterly statistical discrepancy as a
percent of GDP is less than 1 percent.
Chart 2 presents real GDP levels and real GDI levels. (11) The
difference between the two series reflects the variation in the
statistical discrepancy as shown in chart 1. As shown in chart 2, real
GDI is generally higher than real GDP before the third quarter of 2007
but relatively flat for the three quarters leading to the NBER peak in
the fourth quarter of 2007. Real GDI in creases slightly in the first
quarter of 2008 before decreasing significantly during the remaining
quarters of the recession. At both the NBER peak and the NBER trough,
real GDI is lower than real GDE Real GDP increases consistently in the
quarters preceding the NBER peak in 2007 but decreases slightly for the
first quarter of 2008 before increasing slightly and then decreasing
significantly during the remaining quarters of the recession.
[GRAPHIC 1 OMITTED]
[GRAPHIC 2 OMITTED]
Chart 3 presents the percent changes from the preceding period in
real GDP and real GDI. As shown in chart 3, both GDP and GDI growth
begin to slow in the second quarter of 2006. Thus, the measures of
output demonstrate weakness prior to the NBER peak. The percent change
in real GDI seesaws for the three quarters before the NBER peak and
increases in the first quarter of 2008 before decreasing significantly
in the remaining quarters of the recession. The percent change in real
GDP generally decreases in quarters preceding and immediately after the
NBER peak before increasing and then decreasing significantly during the
recession. Thus, chart 2 and chart 3 show similar patterns for GDP and
GDI for the last half of the recession but slightly different patterns
for the quarters leading up to the NBER peak and immediately after the
NBER peak.
Factors of Profits and Income That Contribute to the Statistical
Discrepancy
This section explains the following factors that require
adjustments to convert financial- or tax-based source data into economic
accounting measures of profits and income: (1) misreporting, (2) capital
gains and losses, (3) employee stock options, and (4) produced
intangibles. BEA considers potential measurement error in these factors
to be likely contributors to the statistical discrepancy regardless of
economic cyclicality. However, some factors may be more likely than
others to contribute to the statistical discrepancy during cyclical
turning points because of cyclicality in the related measurement error.
While ongoing work at BEA and other federal agencies attempts to address
and mitigate the related measurement error, the work is limited by
conceptual differences and regulatory reporting requirements underlying
the financial- and tax-based source data.
[GRAPHIC 3 OMITTED]
Misreporting
Studies conducted at the Internal Revenue Service (IRS) have
determined that taxpayers make significant tax misreporting errors.
Because tax-based source data are used to estimate profits and income
for annual and benchmark revisions, adjustments are required for
misreporting by taxpayers. BEA makes separate misreporting adjustments
for corporate profits and for proprietors' and partnership income.
Any inaccuracy in the misreporting adjustments affects the cyclicality
of measured profits and income, but as discussed below, it is unclear
how any errors in the misreporting adjustments may affect the
cyclicality of profits and income.
Misreporting data for proprietors' and partnership income come
from two sources. First, the IRS provides industry-level tabulations of
underreported taxable income based on a study conducted for the 2001 tax
year under the National Research Program (NRP). Because 2001 is the most
recent year for which NRP data are available, BEA extrapolates
underreporting amounts from the 2001 data. Second, the Census Bureau
provides industry-level estimates of nonreporting based on annual
exact-match studies. The nonreporting portion of the misreporting
adjustment is small relative to the underreporting portion.
Misreporting data for corporate profits come primarily from annual
IRS corporate audit reports, which provide estimates of additional tax
amounts owed as determined through audits. BEA supplements the audit
reports with IRS tabulations of the amounts actually collected versus
the amounts recommended in the audit reports. To determine misreported
profits, BEA makes judgments regarding marginal tax rates. In addition,
given the nonrandom nature of the audit sample, BEA makes judgments
regarding the application of the audit amounts to the universe of
corporations.
Given the patchwork of misreporting source data and the age of some
misreporting source data, BEA considers the misreporting adjustments to
be of relatively low reliability for assessing year-to-year changes. In
addition, the misreporting adjustments comprise a significant amount of
corporate profits and proprietors' and partnership income. As a
percent of proprietors' and partnership income in the NIPAs, the
misreporting adjustments for proprietors and partnerships have been
approximately 50 percent since 1970. As a percent of profits before
taxes in the NIPAs, the misreporting adjustments for corporations have
generally fluctuated between 15 and 25 percent. Thus, the misreporting
adjustments for proprietors and partnerships are generally larger as a
percent than the misreporting adjustments for corporations.
BEA does not make any explicit cyclical adjustments to its overall
misreporting adjustments. This is in part due to uncertainty about the
potential effect of cyclicality on misreporting. For example, if
misreporting increases during a downturn as businesses attempt to retain
a larger after-tax share of their business income, the decline in
profits and income could be overstated during the downturn. However,
long-term efforts by the IRS to increase the number of examinations and
audits overall, including audits on high-income individuals, and to
increase the number of audits of sole proprietors and partnerships may
result in a trend decrease in overall misreporting.
Confronted with this uncertainty, the effect of both procyclical
and countercyclical misreporting are assumed in simulating the effect on
the statistical discrepancy for the most recent recession. In order to
simulate the change in the statistical discrepancy in the case of
countercyclicality, we assume a 10 percent increase in annual
misreporting for 2008. If total misreporting increased 10 percent in
2008, the statistical discrepancy would change from -$2.4 billion to
$73.3 billion, and the percent change in real GDI would increase from
-0.4 to 0.1. Likewise, if misreporting is procyclical and decreased 10
percent in 2008, the statistical discrepancy would change from -$2.4
billion to $68.3 billion, and the percent change in real GDI would
decrease from -0.4 percent to -0.9 percent.
Capital gains and losses
Capital gains and losses reflect changes in prices rather than
changes in quantities or economic production. In other words, capital
gains and losses do not reflect profits and income arising from
production and should be excluded from the measures of profits and
income in the NIPAs. However, both financial- and tax-based source data
include capital gains and losses, which requires BEA to make adjustments
to remove them. There are two areas where BEA has concerns regarding the
accurate removal of capital gains and losses due to a lack of data:
gains and losses attributable to corporate partners and gains and losses
associated with mark-to-market (or fair value) accounting. To the extent
that BEA cannot identify capital gains and losses attributable to
corporate partners or mark-to-market accounting, measured profits and
income in the NIPAs may be affected.
Corporate partners
Capital gains and losses attributable to corporate partners may be
leaking into measures of partnership income, which could result in an
overly cyclical measure of partnership income.
Annual tax-based source data on both corporate profits and
partnership income include partnership income attributable to corporate
partners. To prevent double-counting, BEA removes the corporate share
from the NIPA measure of partnership income. Source data on the
corporate share of NIPA partnership income are not available, but data
on the corporate share of tax-based partnership income are available.
However, the tax-based partnership income attributable to corporate
partners includes capital gains and losses. In order to be consistent
with NIPA partnership income, the capital gains and losses must be
removed from the corporate share of tax-based partnership income. (12)
Thus, the adjustment to remove the corporate share from the NIPA measure
of partnership income is determined by subtracting an approximation of
capital gains and losses from the corporate share of tax-based
partnership income.
Table 2 displays net capital gains and losses attributable to
partnerships as a proportion of net partnership income published by the
IRS's Statistics of Income (SOI) for the 10-year period 1999-2008.
(13) (Data for table 2 are not available by type of partner). The
proportion of net capital gains and losses in table 2 appears to be
procyclical with relatively high proportions in years preceding NBER
peaks and relatively low proportions in years following NBER peaks.
Given the unavailability of data by type of partner, BEA approximates
the corporate share of capital gains and losses with a combination of
SOI tabulations of capital gains and losses and assumptions regarding
the corporate share of capital gains and losses. If capital gains and
losses are disproportionately high relative to the chosen assumptions,
the corporate partner adjustment would yield a measure of partnership
income that is too high.
If capital gains and losses are disproportionately low relative to
the chosen assumptions, the corporate partner adjustment would yield a
measure of partnership income that is too low. Thus, while BEA makes the
best estimate possible of partnership income given the available data,
the measure of partnership income is subject to the procyclicality of
capital gains and losses based on the corporate partner adjustment.
Mark-to-market accounting
In addition to partnership income, capital gains and losses may be
leaking into measures of corporate profits through mark-to-market
accounting practices, which could yield an overly cyclical measure of
corporate profits.
While the mischaracterization of capital gains and losses as
ordinary income and losses and vice versa is possible, BEA has no
evidence to suggest systematic mischaracterization in either financial-
or tax-based source data. However, capital gains and losses associated
with mark-to-market accounting rules may be required to be characterized and reported as ordinary income and losses in financial statements and
tax returns.
In this case, ordinary income and losses refer to income and losses
arising from production and used by BEA to derive economic accounting
measures of profits.
Under mark-to-market accounting rules, an asset held at the end of
a reporting period may be treated as sold at its fair value even if the
asset was not actually sold. Likewise, a liability held at the end of a
reporting period may be treated as transferred even if the liability was
not actually transferred. As a result, capital gains and losses may be
recognized in financial statements or on tax returns, but they need to
be excluded from profits for economic accounting purposes. However,
because of a lack of data, BENs adjustments for capital gains and losses
in financial- or tax-based source data may not capture mark-to-market
gains and losses that are required to be reported as ordinary income and
losses. While methods for valuing the asset or liability may differ
under financial accounting rules and tax accounting rules, the
application of mark-to-market accounting may differ considerably under
financial accounting rules and tax accounting rules.
Tax accounting rules. Under tax accounting rules, gains and losses
associated with hedging transactions that are conducted in the normal
course of a taxpayer's business are generally required to be
characterized and reported on an income tax return as part of ordinary
income and losses. (14) For example, a taxpayer that manages future
input costs with a hedge would recognize ordinary income or losses
associated with the hedge. Likewise, a taxpayer that uses a hedge to
manage interest-rate risk related to a future debt issuance would
recognize ordinary income or losses associated with the hedge. While the
taxpayer has some flexibility in choosing an accounting method for
recognizing gains and losses associated with hedging transactions, these
gains and losses must generally be recognized in the same period as
gains and losses associated with the underlying asset or liability. (15)
In cases where a hedge and the underlying asset or liability are both
disposed of in the same year, recognizing the gains and losses may
satisfy the recognition requirement. However, a hedging transaction may
also be accounted for under mark-to-market accounting in order to
satisfy the recognition requirement, which results in capital gains and
losses recognized on the taxpayer's income tax return as ordinary
income and losses. No separate line item is required on the tax return
for the gains and losses associated with the mark-to-market accounting.
Thus, BEA's adjustments to tax-based source data for capital
gains and losses does not account for mark-to-market gains and losses
associated with hedging transactions.
IRS schedule M-3 is a recent information form that corporations
with $10 million or more in assets are required to file. Schedule M-3
provides previously unavailable details regarding receipts and
deductions reported on a corporate income tax return; one of the line
items on schedule M-3 is for hedging transactions. SOI has recently
published tabulations of schedule M-3 for 2008, showing a hedging
transactions loss of $95.1 billion. Assuming hedging transactions
include some mark-to-market gains and losses, not adjusting for the
mark-to-market gains and losses could yield an overly cyclical annual
measure of profits and income and contribute to the statistical
discrepancy. However, without more data and further study, BEA has no
direct evidence regarding the degree or cyclicality of the
mark-to-market gains and losses included in hedging transactions.
Financial accounting rules. Under financial accounting rules,
mark-to-market accounting is required on a recurring basis (that is,
periodically) for some financial assets and liabilities and may be
elected for other financial assets and liabilities. Examples of
financial assets and liabilities include investment securities,
derivative instruments, loans and other receivables, notes and other
payables, and debt instruments issued. Nonfinancial assets and
liabilities are generally accounted for at historic cost with fair value
gains or losses recognized as ordinary gains or losses only when the
value of an asset or liability is considered to be
"other-than-temporarily" impaired. Since gains or losses
associated with other-than-temporary impairment are only recognized on a
nonrecurring basis, nonfinancial assets and liabilities are outside the
current scope. Thus, the focus here is on financial accounting rules
that require or allow mark-to-market accounting for financial assets and
liabilities. (16) Financial accounting rules distinguish three classes
of debt and equity investment securities: (1) debt securities intended
to be held to maturity, (2) debt and equity securities bought primarily
for short-term trading purposes, and (3) debt and equity securities that
are available for sale but not classified in the previous two classes.
(17) Held-to-maturity securities are accounted for at historic cost.
Mark-to-market accounting is required on a recurring basis for the
second of the three classes--trading securities--and the third
class--available-for-sale securities. Trading securities include
mortgage-backed securities that are held for sale in conjunction with
mortgage banking activities. Mark-to-market gains and losses generated
by trading securities are required under financial accounting rules to
be included with earnings in the income statement (that is, ordinary
income or losses) with a separate disclosure of the amount in the notes
to the financial statements.
No separate line item is required in the financial statements for
the gains and losses associated with the mark-to-market accounting.
Mark-to-market gains and losses generated by available-for-sale
securities are required to be included directly in shareholder's
equity rather than in earnings. Thus, earnings reported in financial
statements may include capital gains and losses associated with trading
securities but not available-for-sale securities.
Financial accounting rules also require mark-to-market accounting
on a recurring basis for derivative assets and liabilities, including
derivatives that qualify as hedges. (18) Mark-to-market gains and losses
generated by derivative assets and liabilities and derivatives qualified
as hedges of changes in fair value of an asset or liability are required
to be included with earnings in the income statement (that is, ordinary
income or losses) with no separate line item to distinguish the
mark-to-market gains or losses. In the aggregate, gains or losses
associated with derivative assets should be offset by gains or losses
associated with derivative liabilities. However, earnings available in
disaggregated source data may include capital gains and losses
associated with derivative instruments. Likewise, gains or losses
associated with hedged assets or liabilities are presumably offset only
to the extent of the gains or losses on the qualified derivative. Thus,
earnings reported in financial statements may include capital gains and
losses associated with derivative instruments and financial assets and
liabilities that have not been offset by hedges.
In addition to requiring mark-to-market accounting for investment
securities and derivative instruments, financial accounting rules allow
companies to elect mark-to-market accounting for other financial assets
and liabilities, such as receivables, payables, and debt instruments.
(19) A mark-to-market election is generally applied to an individual
instrument and is irrevocable. In addition, similar to trading
securities and derivative instruments, mark-to-market gains and losses
associated with an election are required to be included with earnings in
the income statement (that is, ordinary income or losses) with no
separate line item. The mark-to-market election has been broadly
available since 2008, so earnings reported in recent financial
statements may include capital gains and losses associated with the
financial assets and liabilities covered by the accounting rules.
BEA uses financial-based source data for quarterly indicators of
corporate profits in some industries. In particular, BEA uses quarterly
financial reports (QFRs) provided by the Census Bureau for mining,
manufacturing, wholesale trade, and retail trade industries. QFRs
include a sample of publicly owned and privately owned companies and
also include adjustments to remove capital gains and losses for use in
the NIPAs. In addition to QFRs, BEA uses Compustat data for some
utilities, transportation, information, real estate, and finance and
insurance industries. For the finance and insurance industries,
quarterly indicators come from Compustat for nondeposit credit
intermediaries, securities dealers, life insurance, and real estate
investment trusts. The Compustat database only includes publicly owned
companies and does not provide a variable to distinguish mark-to-market
gains and losses included in earnings. Thus, for quarters with
substantial changes in market values of securities, BEA can only resort
to a small sample of quarterly financial reports filed with the
Securities and Exchange Commission by individual companies to adjust for
mark-to-market gains and losses. (20) Assuming mark-to-market gains and
losses are procyclical, overadjusting based on the chosen sample would
yield a quarterly measure of profits that is not cyclical enough, and
underadjusting based on the chosen sample would yield a quarterly
measure of profits that is too cyclical.
Finance and insurance industries. Given the inclusion of
mortgage-backed securities with trading securities and given the
concentration of debt and equity securities purchased and sold for
finance-related activities, financial institutions are particularly
affected by mark-to-market accounting. Financial accounting rules for
mark-to-market accounting have been under increasing scrutiny since the
most recent recession (fourth quarter of 2007 to the second quarter of
2009) and the related subprime mortgage crisis because of the volatile
impact that the rules have on earnings during times of market
volatility. For NIPA purposes, the removal of mark-to-market gains and
losses was particularly important but challenging in the finance and
insurance industries leading up to and following the NBER peak in the
fourth quarter of 2007 because of the lack of adequate data on
mark-to-market gains and losses included in earnings reported in
financial statements and compiled in the Compustat database. Thus,
declines in profits in the finance and insurance industries may reflect
mark-to-market losses to the extent that the losses were not identified.
If so, profits in the finance and insurance industries would be
understated.
Chart 4 presents quarterly estimates of corporate profits with
inventory valuation adjustment (IVA) and capital consumption adjustment
(CCAdj) published in the NIPAs. Separate series are shown for the
finance and insurance industries and all other industries. In addition,
the chart includes a series that combines all domestic industries and,
for reference to patterns of potential capital gains and losses, the
chart includes a series for the S&P 500 Index measured on the right
axis. (21)
As shown in chart 4, measured corporate profits with IVA and CCAdj
generally dropped consistently from one quarter to the next for all
domestic industries leading up to the NBER peak. The series for all
domestic industries continued to decline during the recession, but the
decline was driven primarily by the finance and insurance industries,
which dropped considerably more than the nonfinance industries. In
addition, in quarters outside of the recession, corporate profits in the
finance and insurance industries were generally as high as at least 40
percent of corporate profits in nonfinance industries; however, during
the recession, corporate profits in the finance and insurance industries
dropped to less than 5 percent of corporate profits in nonfinance
industries for some quarters. The S&P 500 Index increased steadily
until the third quarter of 2007 and then decreased steadily until it
reached a low in the fourth quarter of 2008 and started another steady
increase.
The variation in corporate profits in chart 4 is highly correlated with the variation in potential capital gains and losses reflected by
the S&P 500 Index. In 2006-2010, the correlations between the
S&P 500 Index and corporate profits in the finance and insurance
industries and nonfinance industries are 0.41 and 0.49, respectively.
The correlation with all domestic industries is 0.52 for the full
period. For the recessionary period, the correlations of the S&P 500
Index with finance and insurance and nonfinance are 0.48 and 0.33,
respectively. The correlation with all domestic industries is 0.79 for
the recessionary period.
Chart 4 does not lead to a conclusion that corporate profits
include capital gains and losses. However, the pattern of corporate
profits in the finance and insurance industries over the period raises
the issue of whether mark-to-market gains and losses play a role in the
pattern of the statistical discrepancy given the relatively high
corporate profits in the finance and insurance industries leading up to
the NBER peak and following the NBER trough and dramatically low
corporate profits during the recession. In order to simulate the change
in the statistical discrepancy in the case of the inclusion of capital
gains and losses during the most recent recession, 10 percent of
quarterly corporate profits are assumed to be attributable to mark-to
market losses. If corporate profits increase by 10 percent for each
quarter of the recession, the statistical discrepancy would improve for
five of the seven recessionary quarters, and the difference between the
percent change in GDP and the percent change in GDI would decline for
five of the seven recessionary quarters. Further study is warranted to
better understand how mark-to-market gains and losses may affect
corporate profits and the statistical discrepancy.
[GRAPHIC 4 OMITTED]
Employee stock options
Inconsistent measurement of stock options in source data for
corporate profits and compensation may generate an overly cyclical
measure of GDI relative to GDE
Differences between the measurement of stock options in source data
can generate significant differences between stock options expense
included in corporate profits from stock options included in wages and
salaries. (22) Financial-based source data for current quarterly
estimates and quarterly interpolations of corporate profits generally
measure stock options expense as the fair market value of the options
allocated over the vesting period on the date that options are granted.
Tax-based source data for annual estimates of corporate profits
generally measure this expense as the difference between the market
price of the stock and the strike price of the options on the date that
the options are exercised. Source data for wages and salaries estimates
initially come from the Current Employment Statistics (CES) program at
the Bureau of Labor Statistics (BLS). The CES data exclude income from
stock options. Five months after the reference quarter, BEA incorporates
data into wage and salary estimates from the BLS's Quarterly Census
of Employment and Wages (QCEW). The QCEW includes income from stock
options measured consistently with the annual tax-based source data.
Given the consistent measurement of stock options in the annual
tax-based source data underlying corporate profits and the QCEW data
underlying wages and salaries estimates, the measurement and timing
differences should not affect the annual statistical discrepancy by the
second annual revision because QCEW data and tax-based source data are
fully incorporated into the NIPAs by then. However, the measurement and
timing differences are likely to contribute to the statistical
discrepancy in current quarterly estimates, and the effect is likely to
persist in the quarterly interpolations after the first annual revision
because stock options are measured inconsistently in quarterly
financial-based source data and in the QCEW. The procyclical nature of
stock prices and the incentive for employees to exercise stock options
when stock prices increase as well as the disincentive when stock prices
decrease may yield an overly cyclical measure of quarterly wages and
salaries. In contrast, quarterly corporate profits as measured by
financial-based source data would be less affected by changes in stock
prices because stock options expenses in quarterly financial data is
measured when stock options are granted and are distributed evenly over
the vesting period. Thus, GDI may be overstated relative to GDP during
stock market increases but understated during stock market declines.
Moylan (2008) provides a comprehensive discussion regarding the
inclusion of stock options in measures of corporate profits and
compensation.
Produced intangibles
Any error in assumptions regarding the capitalization rate of
produced intangibles results in inaccurate measures of profits and
income.
In the year that produced intangibles are acquired, the seller of
the intangibles recognizes revenue and the buyer recognizes expense for
tax purposes if intangibles are not capitalized and depreciated. In this
case, revenues offset expenses, and the statistical discrepancy is
unaffected. When produced intangibles are capitalized and depreciated
for tax purposes, BEA adds the depreciation back to tax-based receipts
less deductions, which is the starting point for profits and income, and
includes the depreciation in consumption of fixed capital, which is
BEA's measure of depreciation included in GDI.
In the case of purchased computer software, BEA assumes a low rate
of capitalization for tax purposes. (23) As a result, the depreciation
for produced intangibles that is added back to tax-based receipts less
deductions includes only a small amount of depreciation for software. In
the year software is purchased, tax-based receipts less deductions
overstates profits and income to the extent that software is capitalized
and not depreciated for tax purposes beyond BEA's assumed rate of
capitalization (that is, when aggregate receipts from software sales are
greater than aggregate deductions from software purchases). Thus, the
statistical discrepancy may be affected. In the years that software is
depreciated, the statistical discrepancy is unaffected because the
capital consumption adjustment absorbs the difference between the actual
depreciation and the assumed depreciation. Assuming software purchases
are procyclical, failure to accurately adjust for capitalized software
would yield a measure of profits and income that may be too high during
cyclical upturns but less affected during downturns.
Summary and Conclusions
This article explains the significant role that profits and income
play in BEA's decision to record the statistical discrepancy as a
separate line item on the income side of the NIPAs and an overview of
the factors of profits and income that are most likely contributing to
the statistical discrepancy.
BEA's decision to record the statistical discrepancy with GDI
reflects BEA's experience and careful consideration of the
reliability of the underlying source data. Source data underlying GDP
are generally consistent with economic accounting concepts and thus
considered more reliable than source data underlying GDI. In contrast,
data underlying the profits and income components of GDI are generally
collected from financial and tax-based sources, which can be
inconsistent with economic accounting concepts and thus require
adjustments for economic accounting purposes. While BEA works to reduce
measurement error related to the source data and required adjustments,
the work is limited by conceptual differences and regulatory reporting
requirements underlying the financial- and tax-based source data.
This article specifically discussed four factors that require
adjustments to convert financial- or tax-based source data into economic
accounting measures of profits and income. First, adjustments for
misreporting are likely factors contributing to the statistical
discrepancy, and the direction of the effect is ambiguous without
further study. Second, capital gains and losses may be leaking into
measures of profits and income and contributing to the statistical
discrepancy through corporate partner adjustments and mark-to-market
accounting practices, which could yield overly cyclical measures of
profits and income. Measurement of profits in the finance and insurance
industries was particularly challenging during the most recent
recession. Third, inconsistent measurement of stock options in source
data for corporate profits and wages and salaries may generate an overly
cyclical measure of GDI relative to GDE Finally, any error in
assumptions regarding the capitalization rate of purchased software may
overstate profits and income during cyclical upturns.
These issues lend support to BEA's practices of not promoting
one output measure over another and of recording the statistical
discrepancy in a transparent manner on the income side of the NIPAs.
However, more attention should be given to describing the GDI estimates
in a manner that will inform the public about this alternative source of
macroeconomic information. Furthermore, additional research is warranted
on factors contributing to the statistical discrepancy, on a framework
for weighting underlying source data in an effort to distribute the
statistical discrepancy, and on a framework and appropriate weights for
a combined output measure.
References
Bureau of Economic Analysis. 2002. Corporate Profits: Profits
Before Tax, Profits Tax Liability, and Dividends. Methodology Paper.
Washington, DC: Bureau of Economic Analysis, September;
www.bea.gov/scb/pdf/ national/nipa/methpap/methpap2.pdf.
Byron, Ray E 1978. "The Estimation of Large Social Account
Matrices." Journal of the Royal Statistical Society, Series A 141,
part 3 (March): 359-367.
Chen, Baoline. 2006. "A Balanced System of Industry Accounts
for the U.S. and Structural Distribution of Statistical
Discrepancy." BEA Working Paper WP2006-8. Washington, DC: Bureau of
Economic Analysis; www.bea.gov.
Chen, Baoline. 2010. "Reconciling the System of U.S. Accounts
and Distribution of the Aggregate Statistical Discrepancy."
Unpublished. Washington, DC: Bureau of Economic Analysis.
European Commission, International Monetary Fund, Organisation for
Economic Co-operation and Development, United Nations, and World Bank.
2009. System of National Accounts 2008. New York: United Nations.
Financial Accounting Foundation, Financial Accounting Standards
Board. Statement of Financial Accounting Standards. Numbers 115 (May
1993), 133 (June 1998), 157 (September 2006), and 159 (February 2007).
Financial Accounting Foundation, Financial Accounting Standards
Board. Accounting Standards Codification. Topics 320, 815, 820, and 825.
Fixler, Dennis J., and Bruce T. Grimm. 2002. "Reliability of
GDP and Related NIPA Estimates." SURVEY or CURRENT BUSINESS 81
(January): 9-27.
Fixler, Dennis J., and Bruce T. Grimm. 2005. "Reliability of
the NIPA Estimates of U.S. Economic Activity." SURVEY OF CURRENT
BUSINESS 85 (February): 8-19.
Fixler, Dennis J., and Jeremy J. Nalewaik. 2009. "News, Noise,
and Estimates of the "True" Unobserved State of the
Economy." BEA Working Paper WP201004. Washington, DC: Bureau of
Economic Analysis; www.bea.gov.
Fixler, Dennis J., Ryan Greenaway-McGrevy, and Bruce T. Grimm.
2011. "Revisions to GDP, GDI, and Their Major Components?'
SURVEY OF CURRENT BUSINESS 91 (July): 9-31.
Greenaway-McGrevy, Ryan. 2011. "Is GDP or GDI a Better Measure
of Output? A Statistical Approach?' BEA Working Paper WP2011-08.
Washington, DC: Bureau of Economic Analysis; www.bea.gov.
Grimm, Bruce T. 2007. "The Statistical Discrepancy." BEA
Working Paper WP2007-01. Washington, DC: Bureau of Economic Analysis;
www.bea.gov.
Grimm, Bruce T., and Robert P. Parker. 1998. "Reliability of
the Quarterly and Annual Estimates of GDP and Gross Domestic
Income." SURVEY OF CURRENT BUSINESS 78 (December): 12-21.
Grimm, Bruce T., and Teresa L. Weadock. 2006. "Gross Domestic
Product: Revisions and Source Data." SURVEY OF CURRENT BUSINESS 86
(February): 11-15.
Holdren, Alyssa E., and Bruce T. Grimm. 2008. "Gross Domestic
Income: Revisions and Source Data." SURVEY OF CURRENT BUSINESS 88
(December): 14-20.
Hodge, Andrew W. 2011. "Comparing NIPA Profits With S&P
500 Profits?' SURVEY OF CURRENT BUSINESS 91 (March): 22-27.
Internal Revenue Service, U.S. Department of the Treasury. Code of
Federal Regulations, Title 26. "Chapter I, Subchapter A, Part 1,
sections 446, 475, and 1221." Washington, DC: Internal Revenue
Service.
Klein, Lawrence R., and J. Junichi Makino. 2000. "Economic
Interpretations of the Statistical Discrepancy." Journal of
Economic and Social Measurement 26 (January), no. 1, 11-29.
Landefeld, J. Steven. 2010. "Comments on 'The Income- and
Expenditure-Side Estimates of U.S. Output Growth." Brookings Papers
on Economic Activity 1 (Spring): 112-123.
Mankiw, N. Gregory, and Matthew D. Shapiro. 1986. "News or
Noise: An Analysis of GNP Revisions." SURVEY OF CURRENT BUSINESS 66
(May): 20-25.
Moylan, Carol E. 2008. "Employee Stock Options and the
National Economic Accounts." SURVEY OF CURRENT BUSINESS 88
(February): 7-13.
Nalewaik, Jeremy J. 2010. "The Income- and Expenditure-Side
Estimates of U.S. Output Growth." Brookings Papers on Economic
Activity 1 (Spring): 71-106.
Rassier, Dylan G., Thomas E Howells III, Edward T. Morgan, Nicholas
R. Empey, and Conrad E. Roesch. 2007. "Implementing a
Reconciliation and Balancing Model in the U.S. Industry Accounts."
BEA Working Paper WP2007-05. Washington, DC: Bureau of Economic
Analysis; www.bea.gov.
Smith, Richard J., Martin R. Weale, and Steven E. Satchell. 1998.
"Measurement Error with Accounting Constraints: Point and Interval
Estimation for Latent Data with an Application to U.K. Gross Domestic
Product." The Review of Economic Studies 65, no. 1 (January):
109-134.
Stone, Richard, James E. Meade, and David G. Champernowne. 1942.
"The Precision of National Income Estimates." The Review of
Economic Studies 9, no. 2 (Summer): 111-125.
Weale, Martin. 1992. "Estimation of Data Measured with Error
and Subject to Linear Restrictions." Journal of Applied
Econometrics 7, no. 2 (April-June): 167-174.
(1.) There are three current quarterly vintages for each quarter:
advance, second, and third. Advance, second, and third quarterly
vintages are released approximately 1, 2, and 3 months, respectively,
after the end of the reference quarter. Likewise, there are three annual
vintages for each year: first, second, and third. Annual vintages are
released at the end of July for the previous 3 years. The most recent
vintage of annual estimates comes from the 2011 annual revision, which
was released on July 29, 2011. The 2011 annual revision includes the
first annual revision for 2010, the second for 2009, and the third for
2008. In addition, the 2011 annual revision was the first
"flexible" annual revision, which includes revisions to
current-dollar GDP and some components back to the first quarter of
2003.
(2.) However, revision studies generally conclude that annual and
benchmark revisions do not substantively change BEA's measures of
long-term growth, pictures of business cycles, and trends in major
components of GDP (Fixler, Greenaway-McGrevy, and Grimm 2011).
(3.) For the first alternative, the System of National Accounts
2008 (SNA) suggests "... it is usual to attach [the discrepancy] to
the variant of [national output] the office feels is least accurate. The
aim is to show users something about the degree of reliability of the
published data" (SNA paragraph 18.16). For the second alternative,
the SNA suggests "The alternative is for the office to remove the
discrepancy by examining the data in the light of the many accounting
constraints in the SNA, making the best judgment possible about where
the errors are likely to have arisen and modifying the data
accordingly" (SNA paragraph 18.17).
(4.) A similar perspective of GDP for the same period indicates a
sharp decrease in private investment, which is offset in part by
increases in net exports and government expenditures and gross
investment.
(5.) Weighting underlying source data in a statistical framework
has been successfully implemented at BEA to reconcile and balance the
gross operating surplus component of the 2002 input-output accounts and
GDP by industry accounts (Rassier et al. 2007).
(6.) Klein and Makino (2000) argue that BEA's decision to
record the statistical discrepancy with GDI is tenable but may result in
nonrandom error in the NIPAs. As a result, the authors argue that the
statistical discrepancy should be distributed among the components of
GDP and GDI. While their conclusions are subject to question (see Grimm
2007), their analytic flamework is uncontroversial and useful to explain
BEA's decision to record the statistical discrepancy with GDI.
(7.) We use different notation for profits and income in equation
(3) than in equation (1) because P in equation (1) denotes aggregate
profits and income that are consistent with economic accounting concepts
while ([PI]) in equation (3) denotes firm-level profits and income that
are consistent with financial or tax accounting concepts.
(8.) See Grimm and Weadock (2006), Holdren and Grimm (2008), and
Landefeld (2010) for further discussion.
(9.) Corporate profits, proprietors' income, and partnership
income constitute a large proportion of gross operating surplus.
(10.) While first annual estimates are generally based on the same
source data used for quarterly interpolations and extrapolations, second
and third annual estimates are based on tax-based source data. Thus,
quarterly variation related to the second and third annual estimates
comes from the source data used for quarterly interpolations.
(11.) Real GDP and real GDI are published in chained 2005 dollars.
We deflate current-dollar GDI using the implicit price deflator for GDP
because there is no price deflator specifically for GDI.
(12.) Capital gains and losses are included in tax-based
partnership income as part of portfolio income and losses. In addition
to capital gains and losses, portfolio income and losses includes
interest, dividends, and royalties. BEA removes all portfolio income and
losses. However, the focus here is on the capital gains and losses
portion because of the effect on partnership income.
(13.) Net capital gains include short term and long-term capital
gains and losses. SOI data for more recent years have not yet been
published.
(14.) Tax rules regarding the definition and identification of
hedging transactions and regarding the treatment of gains and losses
associated with hedging transactions are provided in Treasury
Regulations [section] 1.1221-2.
(15.) Tax rules regarding the accounting methods for hedging
transactions and regarding the recognition of gains and losses
associated with hedging transactions are provided in Treasury
Regulations [section] 1.446-4 for most taxpayers except securities
dealers. Accounting methods for securities dealers are provided in
Treasury Regulations [section] 1.475.
(16.) Financial accounting rules for fair value measurement are
provided in Statement of Financial Accounting Standards (SFAS) number
157 or topic 820 in the new Accounting Standards Codification (ASC).
(17.) Financial accounting rules for investments in debt and equity
securities are provided in SFAS number 115 or ASC topic 320.
(18.) Financial accounting rules for derivative instruments and
hedging activities are provided in SFAS number 133 or ASC topic 815.
(19.) Financial accounting rules for the fair value option for
financial assets and liabilities are provided in SFAS number 159 or ASC
topic 825.
(20.) For more information on corporate profits in the NIPAs, see
Hodge (2011) and Bureau of Economic Analysis (2002).
(21.) The S&P 500 Index series is determined by the monthly
average closing value adjusted for dividends and stock splits.
(22.) The focus here is on nonqualified stock options (NSOs) rather
than incentive stock options (ISOs) because NSOs are more common than
ISOs and because NSOs give rise to ordinary income and losses while ISOs
give rise to capital gains and losses, which are excluded from the NIPA
concepts of corporate profits and compensation (Moylan 2008).
(23.) U.S. tax law allows taxpayers to deduct the cost in the year
of acquisition rather than to capitalize and depreciate the cost of
qualifying property, including purchased computer software, subject to
deduction limitations and other restrictions.
Table 1. Component Shares of Gross Domestic Income, 2006-2010
[Percent]
Compensation Taxes less Proprietors'
subsidies income
2006: I 55.0 6.9 8.4
II 54.9 6.9 8.4
III 54.7 6.8 8.3
IV 55.3 6.8 8.2
2007: I 55.7 6.9 7.9
II 55.7 6.9 7.8
III 56.1 6.9 7.7
IV 56.4 6.9 7.7
2008: I 56.6 6.8 7.8
II 56.2 6.9 7.8
III 56.2 6.9 7.7
IV 57.1 6.9 7.4
2009: I 56.6 6.9 6.9
II 56.8 7.0 6.7
III 56.4 6.9 6.7
IV 55.8 7.0 6.8
2010: I 55.0 6.9 6.9
II 55.1 6.9 7.1
III 55.0 6.8 7.2
IV 54.7 6.8 7.3
Corporate Other net Consumption
PAN oprating of fixed
surplus capital
2006: I 10.0 7.5 12.1
II 9.9 7.8 12.2
III 10.2 7.8 12.2
IV 9.5 7.8 12.3
2007: I 8.8 8.2 12.5
II 8.9 8.2 12.5
III 8.0 8.5 12.7
IV 7.3 8.9 12.7
2008: I 6.6 9.5 12.7
II 6.4 9.9 12.8
III 6.2 10.1 13.0
IV 4.4 10.8 13.4
2009: I 5.9 10.1 13.6
II 6.7 9.3 13.6
III 7.8 8.8 13.4
IV 8.5 8.7 13.3
2010: I 9.5 8.7 13.0
II 9.6 8.4 12.9
III 9.8 8.3 12.9
IV 10.1 8.1 12.9
NOTES. The shaded area is the date of the recession determined by
the Business Cycle Dating Committee of the National Bureau of
Economic Research. GDI components are from NIPA table 1.10. The
data are from the 2011 annual NIPA revision.
Table 2. Proportion of Net Capital Gains
in Net Partnership Income, 1999-2008
[Percent]
Net capital
gains proportion
1999 34.6
2000 37.4
2001 11.0
2002 1.6
2003 19.3
2004 34.9
2005 37.0
2006 39.3
2007 46.7
2008 1.6
NOTES. Net capital gains include short-term and long-term capital
gains and losses.
The shaded areas are approximate dates of recessions determined
by the Business Cycle Dating Committee of the National Bureau of
Economic Research.
The data are from the Internal Revenue Service's Statistics of
Income.
Net capital gains include short-term and long-term capital gains
and losses.
