Close
About
FAQ
Home
Collections
Login
USC Login
Register
0
Selected
Invert selection
Deselect all
Deselect all
Click here to refresh results
Click here to refresh results
USC
/
Digital Library
/
University of Southern California Dissertations and Theses
/
Real asset liquidity and asset impairments
(USC Thesis Other)
Real asset liquidity and asset impairments
PDF
Download
Share
Open document
Flip pages
Contact Us
Contact Us
Copy asset link
Request this asset
Transcript (if available)
Content
Running head: REAL ASSET LIQUIDITY AND ASSET IMPAIRMENTS 1
Real Asset Liquidity and Asset Impairments
Bryce A. Schonberger
University of Southern California
Author Note
Bryce A. Schonberger, Marshall School of Business, University of Southern California.
Bryce Schonberger is now at the Simon School of Business, University of Rochester.
I am extremely grateful to K.R. Subramanyam for guidance and encouragement on this
project. I benefitted from comments and suggestions from Randy Beatty, Sarah Bonner, Mark
DeFond, David Erkens, John Matsusaka, Jim Manegold, Jeff McMullin, Maria Ogneva, Gordon
Phillips, Mark Soliman, Karen Ton, Bob Trezevant, Paul Zarowin, Jerry Zimmerman, and
workshop participants at Boston College, Columbia University, the George Washington
University, New York University, University of Miami, University of Oregon, University of
Rochester, University of Southern California, and University of Washington. I am grateful to
Alessandro Gavazza for sharing his data on the secondary market for used aircraft. I appreciate
the financial assistance of the Deloitte Foundation. All errors and views remain my own.
Correspondence regarding this article should be addressed to Bryce Schonberger,
Marshall School of Business, University of Southern California, 3660 Trousdale Pkwy, Los
Angeles, CA 90089-0441. Contact: schonber@usc.edu
REAL ASSET LIQUIDITY AND ASSET IMPAIRMENTS 2
Table of Contents
Abstract……………………………………………………………………………………………3
Introduction………………………………………………………………………………………..4
Motivation and Hypothesis Development……………………………………………………….10
Real Asset Liquidity and Impairment Frequency and Magnitude………………………10
Real Asset Liquidity and the Informativeness of Impairments………………………….15
Research Design…………………………………………………………………………………18
Independent Variables – Real Asset Liquidity…………………………………………..18
Dependent Variables and Regression Models…………………………………………...22
Control Variables and Alternative Explanations………………………………………...26
Sample……………………………………………………………………………………………30
Results……………………………………………………………………………………………32
Main Analysis……………………………………………………………………………32
Robustness Tests…………………………………………………………………………39
Conclusion……………………………………………………………………………………….41
References………………………………………………………………………………………..43
Appendix A………………………………………………………………………………………49
Appendix B………………………………………………………………………………………52
Appendix C………………………………………………………………………………………54
Tables…………………………………………………………………………………………….55
REAL ASSET LIQUIDITY AND ASSET IMPAIRMENTS 3
Abstract
I examine how the presence of a more active (liquid) resale market for real assets influences the
frequency, magnitude, and timeliness of asset impairments. Consistent with an available resale
market providing a useful benchmark for evaluating recorded asset values, I find that firms with
more liquid real assets recognize more frequent and timelier impairments, resulting in lower
book-to-market ratios and more conditionally conservative earnings for firms with more liquid
real assets. Impairments are more frequent in tests using both industry-level measures of real
asset liquidity and firm-specific measures of aircraft fleet liquidity for firms in the airline
industry. Real asset liquidity also improves the information content of accounting values,
especially book values. Finally, impairments are associated with decreases in information
asymmetry around earnings announcements for firms with more liquid real assets.
Keywords: real asset liquidity, impairment, information asymmetry, entropy balancing
REAL ASSET LIQUIDITY AND ASSET IMPAIRMENTS 4
Real Asset Liquidity and Asset Impairments
Introduction
Accounting standards dictate that observable market values be used, whenever possible,
to determine recognized impairment amounts for even real (non-financial) assets carried on the
balance sheet (FASB, 1995; 2001a). The ready availability of market values for a firm’s real
assets should therefore simplify and facilitate the accountant’s task of measuring impairments for
such assets. In this study, I examine whether an active resale market for a firm’s real assets,
which I capture through the level of real asset liquidity, influences the frequency, magnitude, and
timeliness of asset impairments, and therefore the information content of accounting numbers. I
find that firms with higher real asset liquidity recognize more frequent, smaller impairments,
resulting in more conservative book values and earnings and more value relevant financial
numbers, especially book values. Consequently, information asymmetry significantly decreases
around earnings announcements for firms with high real asset liquidity that record asset
impairments, even while it increases for those firms with low real asset liquidity that record asset
impairments.
Unlike financial assets, real assets do not trade on organized exchanges. In addition,
managers are not required to disclose details on the liquidity of inputs used in the impairment
measurement process as they are with fair values for financial assets.
1
To capture observable
resale activity capable of acting as a benchmark for impairment estimates, I follow existing
literature in finance (Schlingemann, Stulz, and Walkling, 2002; Ortiz-Molina and Phillips, 2013;
Almeida et al., 2011) and measure a firm’s real asset liquidity for the year as the scaled
aggregate dollar value of annual total asset sales in the firm’s industry. I also use the number of
firms with discontinued operations and the number of firms with merger and acquisition (M&A)
1
See Statement of Financial Accounting Standards No. 157, Fair Value Measurements, for details (FASB, 2006).
REAL ASSET LIQUIDITY AND ASSET IMPAIRMENTS 5
transactions within the firm’s industry each year (measures of the liquidity or thickness of the
resale market in an industry) as additional measures of real asset liquidity. To address concerns
that these proxies do not directly measure the liquidity of specific firms’ real assets, I also
conduct tests in the airline and air courier industries using a firm-specific measure of aircraft
fleet liquidity. I measure the liquidity of each make and model of airplane using a ratio of the
number of aircraft resale transactions scaled by the average number of aircraft in operation in a
given year. Fleet liquidity is a weighted average of the liquidity ratios for each make and model
of aircraft based on the towing weight of each aircraft type within an airline firm’s fleet.
2
Existing standards governing fixed asset impairment mandate a two-step testing process
be conducted in any period where a change in circumstance for the asset indicates the potential
for an impairment (FASB, 2001a). First, firms compare an estimate of future undiscounted cash
flows that can be earned from using the asset in its present capacity, i.e., value-in-use, to its
carrying value on the balance sheet. Standards governing goodwill impairment similarly specify
an impairment trigger in step one, where the fair value of a reporting unit is compared to overall
reporting unit carrying value (FASB, 2001b). However, the test for goodwill recoverability is
mandated on an annual basis, rather than only in periods where a change in circumstance is
present. Should the recoverability tests indicate a fixed asset or goodwill impairment, the amount
of the impairment is then determined by comparing the carrying value of the asset to its fair
value, i.e., value-in-exchange.
3
Thus an asset’s resale value (value-in-exchange) is only relevant
for measuring the amount of any required impairment in step two, after making the decision to
2
See research in finance by Pulvino (1998) and Gavazza (2011) for similar measures of aircraft liquidity. Measures
of aircraft liquidity rely on the dataset used by Gavazza (2011) in his detailed study of market thickness and trading
frictions in the market for used aircraft. I calculate aircraft fleet liquidity both for the entire aircraft fleet and the
portion of the fleet that is owned, not leased, by the airline.
3
The amount of any goodwill impairment is measured as the difference between the carrying value of goodwill on
the balance sheet and the implied fair value of reporting unit goodwill, calculated by allocating the fair value of a
reporting unit to all of the assets and liabilities of that unit (FASB, 2001b).
REAL ASSET LIQUIDITY AND ASSET IMPAIRMENTS 6
record an impairment in step one. Both fixed asset and goodwill impairment standards
recommend the use of market values during the measurement process, where available.
The focus of this study is on whether readily observable resale values act as a benchmark
that improves the process of estimating and recording asset impairments. I expect that observable
resale values will influence the impairment process by supporting more frequent and timelier
asset revaluations. If managers/auditors require sufficient verifiable evidence to support an
adjustment away from carrying value, then greater availability of information on current asset
values will support more frequent and timelier revaluation decisions. Because US GAAP permits
only downward asset revaluations, more frequent and timelier asset revaluations will manifest as
more frequent and timely downward revaluations i.e., impairments.
Consistent with verifiable resale values determining impairments, I find that firms with
more liquid real assets record significantly more frequent impairments in earnings. Specifically,
the probability of recognizing an asset impairment during the year increases by 1.7% for a one-
standard deviation increase in a real asset liquidity factor that combines the three separate
liquidity measures discussed above. This translates to an increase of 7.7% in the unconditional
probability of an asset impairment. This effect is concentrated in fixed asset write-offs rather
than goodwill, with the probability of recognizing a fixed asset impairment during the year
increasing by 0.7% (0.9%) for a one-standard deviation increase in the asset sales (discontinued
operations) measure of real asset liquidity, resulting in an increase of 4.4% (5.6%) in the
unconditional probability of a fixed asset impairment. In tests focusing on the airline industry, I
find a larger increase of 5.9% in the predicted likelihood of recording an aircraft impairment,
representing an increase in the unconditional impairment probability of 23% for a one-standard
deviation change in owned aircraft fleet liquidity. In additional tests, I find some evidence in
REAL ASSET LIQUIDITY AND ASSET IMPAIRMENTS 7
favor of smaller fixed asset write-downs for a given carrying value at firms with more liquid
assets. Overall, this evidence is consistent with observable resale values leading
auditors/managers to book impairments more frequently and for somewhat smaller differences
between estimated asset fair value and carrying value on the balance sheet.
I also examine whether real asset liquidity enhances the timeliness and information
content of recorded impairments by prompting auditors and/or managers to revalue assets in
earlier periods. However, a potential issue with existing impairment standards is that firms may
be forced to record impairments caused by temporary fluctuations in asset market values. This
echoes existing evidence showing that fair values calculated using market inputs in less liquid
markets are associated with less informative financials (Altamuro and Zhang, 2012). Despite the
potential for temporary market fluctuations, I find some evidence of timelier impairments for
firms with more liquid real assets. I find that in a regression with controls for non-discretionary
elements of conditional conservatism (Khan and Watts, 2009), firms with liquid real assets
display greater asymmetric timeliness in goodwill impairments when using the M&A measure of
real asset liquidity. For a one-standard deviation move in M&A real asset liquidity, the
asymmetric timeliness of goodwill write-offs increases by 52% of the predicted shift for a one-
standard deviation move in lagged B/M ratio. Total impairments and fixed asset impairments
also display t-stats that are significant in one-tailed tests, indicating weak evidence of greater
conditional conservatism when using the real asset liquidity factor and the discontinued
operations measures of real asset liquidity. In addition, I find that timely loss recognition
cumulates over time on the balance sheet leading to lower book-to-market (B/M) ratios for firms
with more liquid real assets (Roychowdhury and Watts, 2007), including in tests involving
airline firms.
REAL ASSET LIQUIDITY AND ASSET IMPAIRMENTS 8
I next examine whether impairments enhance the information content of financial
numbers for firms with more liquid real assets. Consistent with this conjecture, I find that
explanatory power doubles in regressions of equity prices on earnings and book values when
moving from the lowest to the highest quartile of the real asset liquidity distribution. Further,
book values are more value relevant than earnings for firms with high real asset liquidity,
consistent with more up-to-date recorded asset values for liquid firms.
4
To test whether greater
value relevance is related specifically to impairments, I regress equity prices on book values after
adding back asset write-offs (essentially undoing the impairment) and compare explanatory
power to that from the regression using reported book values. Results show that explanatory
power is significantly higher when using reported book values for firms with liquid real assets. In
contrast, firms with illiquid real assets show generally lower explanatory power for book values
after impairment.
Finally, I investigate changes in information asymmetry around the release of accounting
information for firms recording impairments. Because real asset liquidity is associated with
timelier, more informative accounting information, the quality of publicly available information
for firms with liquid real assets should increase around accounting information releases. If this
publicly available information levels the playing field for unsophisticated investors, then
information asymmetry should decrease around earnings announcements for firms with more
liquid real assets. Indeed, results show that information asymmetry reflected in analyst forecast
dispersion significantly declines around earnings announcements for firms with liquid real assets
that take an impairment. In contrast, information asymmetry actually increases slightly around
4
In an Ohlson (1995) framework, more frequent impairments for firms with more liquid real assets are associated
with a balance sheet concept of earnings, consistent with earnings measuring the change in value of the stock of
assets. This results in a more volatile, less persistent earnings stream and in greater weight on book values in
measuring firm value.
REAL ASSET LIQUIDITY AND ASSET IMPAIRMENTS 9
earnings announcements for illiquid firms that record a write-off.
5
This evidence is consistent
with improved information environments for firms recognizing impairments to liquid real assets.
Measures of real asset liquidity used in the study are relatively exogenous to firms’
individual accounting decisions. However, a potential concern with industry-based measures of
real asset liquidity is that these measures may be merely capturing variation in product market
characteristics across industries, particularly firm performance. The within-industry validation
test for airline firms helps address this concern to some extent. In addition, I conduct tests using
entropy balancing (Hainmueller, 2012)—an arguably superior variant of propensity-score
matching—to control for the effects of performance, volatility, growth, product market
competition, and asset tangibility. Results using this alternative approach to generate a matched
control sample are qualitatively similar to those using multivariate linear regression in my
primary analysis.
This study contributes to existing literature on several dimensions. First, I extend research
examining the determinants of asset impairments (Francis, Hanna, and Vincent, 1996; Riedl,
2004). In contrast to earlier literature that studies economic factors and managerial incentives as
determinants of asset impairments, I examine elements of the process of taking impairments and
suggest that information provided by readily available resale values for a firm’s assets determine
the frequency, magnitude, and timeliness of impairments. Second, I add to research examining
the consequences of requiring complex, potentially unverifiable estimates in financial statements.
Evidence of more frequent and timelier impairments for firms with liquid real assets is consistent
with the information environment influencing impairment decisions. In this sense, judgment
enters into the application of existing accounting standards. However, this judgment is
fundamentally different from managerial discretion. In this case, the focus is on the cost of
5
Results are consistent with work in finance by Gopalan, Kadan, and Pevzner (2012).
REAL ASSET LIQUIDITY AND ASSET IMPAIRMENTS 10
gathering information required to implement existing standards and not on managerial incentives
to bias information. While managerial incentives to bias information may interact with the
availability of verifiable information, the focus here is solely on the availability of this
information. To my knowledge, this is one of the first papers that looks at how the information
environment influences accounting practice. Finally, I extend research examining variation in the
availability of resale values for financial assets. Altamuro and Zhang (2012) examine fair values
of mortgage servicing rights based on managerial inputs (Level 3) vs. market inputs (Level 2)
and find that Level 3 estimates better reflect the cash flow and risk characteristics of the
underlying assets.
6
In contrast to this evidence for financial assets, I find that more liquid real
assets are associated with more informative financial statements for non-financial firms.
Motivation and Hypothesis Development
Real Asset Liquidity and Impairment Frequency and Magnitude
Prior research in accounting focuses on economic factors and managerial incentives that
determine the decision to recognize an impairment. Elliott and Shaw (1988), Francis, Hanna, and
Vincent (1996), and Riedl (2004) find that firm performance significantly determines long-lived
asset impairments. Hayn and Hughes (2006) find that goodwill write-offs are predictable based
on firm performance and characteristics of the original acquisition that gave rise to the recorded
goodwill. In research on incentives, Easton, Eddey, and Harris (1993), Barth and Clinch (1998),
and Aboody, Barth, and Kasznik (1999) examine the determinants of upward long-lived asset
revaluations permitted under Australian and UK GAAP and find that incentives to avoid
violating debt contracts determine asset revaluation decisions. Beatty and Weber (2006) find that
6
Similarly, Lawrence, Siriviriyakul, and Sloan (2013) examine closed-end funds and find that Level 3 fair values
are better predictors of long run intrinsic values for the funds relative to more liquid fair values (Levels 1 and 2).
REAL ASSET LIQUIDITY AND ASSET IMPAIRMENTS 11
firms adopting SFAS No. 142, Goodwill and Other Intangible Assets, elect to take an initial
goodwill impairment charge in response to equity market concerns, debt contracting and bonus
incentives, and length of CEO tenure. Similarly, Ramanna and Watts (2012) find evidence that
firms delay or avoid goodwill impairments in response to motives predicted by agency theory.
In contrast to examining underlying economics and managerial incentives, I examine how
the information environment, in the form of available asset resale values, influences asset
impairments. To my knowledge, extant research has not examined how the information
environment influences the process of measuring and recording asset impairments. I expect that
observable resale values will influence the impairment process by supporting more frequent and
timelier asset revaluations. If managers/auditors require sufficient verifiable evidence to support
an adjustment away from carrying value, then greater availability of information on current asset
values will support more frequent and timelier revaluation decisions. In the absence of evidence
to support a revaluation of the asset, managers and auditors are expected to default to recording
the asset at its current carrying value and recognizing declines in asset value smoothly through
depreciation. Because US GAAP permits only downward asset revaluations, more frequent and
timelier revaluations will manifest as more frequent and timely downward revaluations i.e., asset
impairments.
This straightforward argument for more frequent impairments at firms with liquid real
assets is supported by features of the existing standards governing the impairment process.
Statement of Financial Accounting Standards (SFAS) No. 144, Accounting for the Impairment or
Disposal of Long-Lived Assets, maintains a two-step process for recognizing an impairment loss
in settings in which an indicator of impairment is present. Indicators of impairment to an asset
(or asset group) are referred to as changes in circumstance for the asset and are triggered by
REAL ASSET LIQUIDITY AND ASSET IMPAIRMENTS 12
general business conditions or changes in the manner in which the firm uses or expects to use the
asset. If an impairment indicator is present, the firm must first perform a test for recoverability
based on a firm-specific estimate of undiscounted cash flows, i.e., value-in-use. That is, an
asset’s recoverability is estimated within the context of the specific entity, in contrast to
measuring the asset’s fair value, i.e., value-in-exchange, which must rely on market-based
pricing information from outside the firm when this information is available (FASB, 2001a; pg.
40). In the second step, the firm should measure the amount of impairment for assets that fail to
meet the recoverability test as the difference between the asset’s fair value and its carrying value.
Accounting standards governing goodwill impairments are similar in spirit to the above.
The primary differences between SFAS No. 142, Goodwill and Other Intangible Assets, and
SFAS No. 144 are that impairment testing is mandated on an annual basis for goodwill and that
the fair value of a reporting unit is compared to the carrying value of the unit with goodwill
included in the test for recoverability in step one, rather than using an undiscounted cash flow
estimate of the value of goodwill. The amount of any goodwill impairment at a reporting unit is
then measured as the difference between the carrying value of goodwill and the implied fair
value of reporting unit goodwill, calculated by allocating the fair value of a reporting unit to all
of the assets and liabilities of that unit (FASB, 2001b).
I expect available information on resale values to prompt more frequent and timelier
impairments in two specific ways given the structure of existing impairment standards. First,
existing standards require firms to evaluate long-lived assets, other than goodwill, for
impairment only after a change in circumstance for the asset. If observable declines in resale
prices for an asset trigger such a change in circumstance, then I expect more frequent tests for
long-lived asset impairment at firms with more liquid real assets. Given the mandated annual
REAL ASSET LIQUIDITY AND ASSET IMPAIRMENTS 13
testing frequency for goodwill impairment, I do not expect that observable resale values for
firm’s reporting units will result in a difference in goodwill impairment testing frequency across
firms. Second, I expect that readily available resale values will also influence estimates of fair
value and value-in-use. If auditors and/or managers impose a verifiability threshold when
evaluating uncertain impairment estimates, then firms will only recognize an impairment when
the probability is sufficiently high that the true asset value-in-use lies below the carrying value.
Auditors and/or managers may impose a verifiability threshold for recognizing impairments to
avoid recording impairments related to temporary fluctuations in market prices that introduce
noise into earnings, or to counteract managers’ incentives to take big bath write-offs by
downwardly biasing estimates of value-in-use. Indeed, an inability to adequately evaluate
unverifiable fair value estimates is in line with the explanation offered by Ramanna and Watts
(2012) for evidence that firms avoid goodwill impairments in response to motives predicted by
agency theory. Similarly, Christensen and Nikolaev (2013) find that firms reporting under IFRS
do not elect fair value accounting for less liquid asset categories, due presumably to difficulty in
estimating fair value for these assets.
In this sense, judgment enters into the application of accounting standards. However, this
judgment is fundamentally different from managerial discretion. In this case, the focus is on the
cost of gathering information required to implement existing impairment standards and not on
managerial incentives to bias information. While managerial incentives to bias information may
interact with the availability of verifiable information required to comply with a particular
accounting standard, the focus here is solely on the availability of this information. To my
knowledge, this is one of the first papers that looks at how the information environment
influences the practice of accounting and complying with existing standards.
REAL ASSET LIQUIDITY AND ASSET IMPAIRMENTS 14
A non-mutually exclusive alternative explanation for how information provided by more
liquid real assets will affect write-downs is through the experience that auditors/appraisers build
through working with more liquid assets.
7
For instance, professionals working in industries
experiencing more M&A will be used to fair valuing assets through the M&A process, leading to
available expertise when valuing the assets of related firms. Both of these arguments suggest that
the effect of real asset liquidity on the information environment will influence the frequency of
recorded asset impairments, leading to my first hypothesis in alternative form:
H1: Asset impairments will be more frequent for firms with liquid real assets.
In contrast, if auditors/managers impose a sufficiently low verifiability threshold for
recognizing an impairment, then impairments will be more frequent for firms with less verifiable
information to support the impairment. Essentially, auditors/managers will record impairments
for relatively uncertain shifts in asset value, possibly to avoid consequences related to failing to
record an impairment in a timely fashion. Alternatively, if the cost of gathering verifiable
information for impairment estimates is not sufficiently different for firms with available resale
values, then I should observe a similar frequency of impairment for firms with less liquid real
assets. Ultimately, the relation between impairment frequency and real asset liquidity is an
empirical question.
In addition to more frequent impairments, I also expect that conditional on an impairment
being taken, firms with more liquid real assets will record smaller impairments. This prediction
follows from the verifiability threshold that I expect to be imposed on impairment estimates
discussed above. I expect that firms will only recognize an impairment when the probability is
sufficiently high that the true asset value-in-use lies below the carrying value, allowing for
smaller impairments to be recorded for a given difference between estimated fair value and
7
I thank workshop participants at Columbia University for suggesting this point.
REAL ASSET LIQUIDITY AND ASSET IMPAIRMENTS 15
carrying value at firms with more verifiable information supporting an impairment (firms with
liquid real assets). This leads to my second hypothesis in alternative form:
H2: Asset impairments will be smaller for firms with liquid real assets, conditional on an
impairment being recognized.
In contrast, if firms with illiquid real assets delay recording an impairment, then
depreciation and inflation in the nominal value of money will tend to reduce the value of
impairments recorded in future periods by reducing the real carrying value of the asset on the
balance sheet. This will render impairments smaller, not larger, for firms with illiquid real assets.
Real Asset Liquidity and the Informativeness of Impairments
I next examine whether real asset liquidity enhances the timeliness and information
content of recorded impairments. Extant research in accounting examines the influence of
financial and investment asset liquidity on the information content of asset values. Dietrich,
Harris, and Muller (2000) find that mandatory annual fair value estimates for UK investment
property are significantly less biased and more accurate measures of ultimate selling price than
respective historical costs. The authors also show that reliability of fair value estimates increases
when monitored by external appraisers and Big 6 auditors. Similarly, Altamuro and Zhang
(2012) and Lawrence, Siriviriyakul, and Sloan (2013) show that estimates of Level 3 fair values
for mortgage servicing rights and closed end fund investments, respectively, better reflect the
intrinsic value and risk characteristics of the underlying assets relative to market-based fair
values (Level 2). These studies indicate the information content advantages of relying on model-
based valuation techniques for financial assets in less liquid markets. Relatedly, existing
REAL ASSET LIQUIDITY AND ASSET IMPAIRMENTS 16
standards discuss concerns that firms may be forced to record impairments caused by temporary
fluctuations in market prices for assets, introducing noise into earnings.
8
In contrast, Song, Thomas, and Yi (2010) show that the value relevance of bank net
assets estimated using Level 1 and Level 2 fair values is greater than the value relevance of
Level 3 net assets. Riedl and Serafeim (2011) similarly find that Level 3 assets for financial
institutions have higher implied equity betas relative to Level 1 and Level 2 assets, and this
effect is concentrated in firms with poor information environments. Given conflicting evidence
for financial assets, it’s not clear ex ante to what extent the availability of a liquid resale market
for non-financial (operating) assets will influence the information content of financial statements.
Despite evidence of advantages for model-based valuation techniques when valuing less
liquid financial assets, I expect that verifiable resale values for firms’ real assets will allow for
recognition of timelier impairments relative to assets lacking observable resale values, where
impairments are delayed until more verifiable information is accumulated. To examine the
timeliness of asset impairments, I consider properties of earnings and book values that are
influenced by the timely recognition of losses. First, I expect that timelier asset write-offs will
result in greater conditional conservatism (Basu, 1997) in earnings of more liquid firms,
consistent with firms recognizing declines in asset values in a timelier manner relative to the
recognition of gains. This is consistent with variation in the measurement process for losses
driving conditionally conservative reporting.
Because impairments are a significant driver of conditional conservatism (Lawrence,
Sloan, and Sun, 2013), this study notes that conditional conservatism can arise due to the cost of
8
From SFAS No. 144: “[Some respondents to the Discussion Memorandum] favored using either the permanence or
probability criterion to avoid recognition of write-downs that might result from measurements reflecting only
temporary market fluctuations… In their view, a high hurdle for recognition of an impairment loss is necessary to
prevent premature write-offs of productive assets” ( FASB, 2001a).
REAL ASSET LIQUIDITY AND ASSET IMPAIRMENTS 17
gathering and processing information to comply with existing accounting standards. By
examining the relation between verifiable information provided by asset resale values and the
timeliness of recorded impairments, I seek to provide evidence on the precise link between
accounting standards and conditional conservatism in earnings. In essence, this study examines
how impairment standards are applied in practice. This prediction for the effect of information
associated with real asset liquidity on conditional conservatism in firm earnings differs from
prior literature that focuses on discretionary vs. non-discretionary explanations for conservative
reporting, as discussed by Roychowdhury and Martin (2013). I focus on the effect of information
availability on reporting conservatism and not on the presence or absence of managerial
incentives for conservative reporting.
Second, because current US GAAP allows downward revaluations of non-financial assets
to reflect fair value but prohibits upward revaluations, I expect that firms with more liquid real
assets will have lower book values, consistent with timely loss recognition cumulating in firms
conservatively valuing their existing asset base. This prediction follows Roychowdhury and
Watts (2007), who demonstrate that conditional conservatism on the income statement cumulates
over time in lower B/M ratios. These predictions are summarized in my third hypothesis in
alternative form:
H3: Asset impairments will be timelier for firms with liquid real assets, resulting in more
conditionally conservative earnings and lower book values.
I also expect that firms with more liquid real assets will have significantly more value
relevant accounting information, consistent with timelier impairments improving the information
content of book values and earnings. I further expect that greater value relevance for firms with
more liquid real assets will be concentrated in more accurate book values following impairments.
REAL ASSET LIQUIDITY AND ASSET IMPAIRMENTS 18
Because impairments update book values, firms with more liquid real assets should maintain a
less persistent, more volatile earnings stream with book values receiving a greater weight in
measuring firm value. These predictions are consistent with the residual income valuation
framework in Ohlson (1995) and follow evidence in Collins, Maydew, and Weiss (1997) of an
increase in value relevance of book values for firms recognizing one-time items in earnings:
H4: Summary accounting information, particularly book values, will be more value
relevant for firms with liquid real assets.
Finally, I explore whether real asset liquidity influences information asymmetry through
its effect on timelier and more informative impairments. If real asset liquidity is associated with
timelier, more informative accounting information, then the quality of publicly available
information about the firm should increase around accounting information releases. If this
publicly available information serves to level the playing field for unsophisticated investors, then
information asymmetry should decrease around earnings announcements for firms with more
liquid real assets:
H5: Information asymmetry will decrease around earnings announcements for firms with
liquid real assets.
Research Design
Independent Variables – Real Asset Liquidity
To capture the presence of observable resale values, I follow research in finance.
Williamson (1988) notes that the ability to redeploy an asset, such as commercial land, to an
alternative use is a key driver of real asset liquidity. These general-use assets will have
liquidation values that approach the asset’s value in best use given the large set of potential
REAL ASSET LIQUIDITY AND ASSET IMPAIRMENTS 19
buyers. In contrast, research by Shleifer and Vishny (1992) notes that most assets fail to meet
Williamson’s definition of redeployable. For instance, oil rigs and steel plants are specialized to
the particular function for which they were created. To sell these assets at a value approaching
the value in best use, a buyer must be located that will use the assets in approximately the same
way as the current owner. Assets sold to a buyer outside of the firm’s industry will face adverse
selection costs due to a lack of familiarity with the assets themselves and will experience agency
costs if the buyer is forced to hire an outside manager for the assets. Research by Ramey and
Shapiro (2001) examining aerospace plant closures provides empirical evidence consistent with a
costly process both in terms of time and discounts to price for transferring real assets to
alternative uses outside the industry.
I seek to capture observable asset resale activity that will act as a useful benchmark for
recorded asset values in a manner that reflects the industry equilibrium concept of real asset
liquidity emphasized by Shleifer and Vishny (1992) and in a subsequent refinement by Gavazza
(2011). To capture high valuation buyers that have a working knowledge of the assets being
transferred, I focus on the 3-digit SIC level. I examine 3-digit SIC industries to balance concerns
that the industry definition is too selective, while being specific enough to result in meaningful
resale activity for a firm’s assets. In addition, I rely on SIC industries in place of alternative
industry definitions as SIC industries are defined according to production technology, which is
critical to identifying a set of homogenous real assets across firms.
9
SIC industries are also
readily available in Compustat and SDC Platinum and research by Bhojraj, Lee, and Oler (2003)
9
From the Bureau of Labor Statistics discussion of industry classifications: “An industry consists of a group of
establishments primarily engaged in producing or handling the same product or group of products or in rendering the
same services.” Available at http://www.bls.gov/bls/naics.htm.
REAL ASSET LIQUIDITY AND ASSET IMPAIRMENTS 20
finds little difference between SIC industries and updated versions, such as NAICS, in most
research settings.
10
My first measure of real asset liquidity is similar in spirit to the measure of resale activity
developed by Almeida and Campello (2007) and Almeida, Campello, and Hackbarth (2011).
11
These authors rely on US Bureau of Census Economic Census data that tracks the portion of used
vs. new assets employed by manufacturing firms to capture the degree of resale activity within
an industry. I rely instead on cash flow statement data available in Compustat to avoid a
requirement for US Census data, which ceases to track the portion of used assets employed
following the 1992 Economic Census. I use the aggregate dollar value of asset sales captured on
the cash flow statement under investing activities scaled by book value of industry assets to
capture asset sales that do not require the sale of entire divisions in the M&A market.
12
As a second measure of the extent of observable asset sales in an industry, I take the
natural log of the count of the number of firms reporting gains or losses from discontinued
operations on the cash flow statement in a given industry-year. Firms discontinuing operations
must dispose of assets through a resale or through the scrap market. The presence of
discontinued operations will therefore generate observable prices for assets in the discontinued
segments. To my knowledge, this is the first study to rely on firms reporting discontinued
operations as a measure of the extent of asset resale activity.
Similarly, I calculate the thickness of the merger and acquisition (M&A) market by
taking the log of the number of successful merger and acquisition transactions in SDC Platinum
10
Indeed, results are qualitatively similar when using 4-digit NAICS codes to define real asset liquidity measures.
11
Related work by Alderson and Betker (1995) measures real asset liquidity using discounts calculated in
bankruptcy proceedings for asset liquidation values relative to going-concern values. I do not follow this approach
as the evidence involves a small sample (88 firms) and involves highly variable estimates for firms within the same
industry. The authors acknowledge that generalizing liquidity discounts to other firms may be problematic.
12
Alternatively, I examine a measure that relies on PPE sales tracked on the cash flow statement in place of total
asset sales. Results are similar but weaker (untabulated) when using this alternative measure, due in large part to the
high incidence of missing values for PPE sales in Compustat.
REAL ASSET LIQUIDITY AND ASSET IMPAIRMENTS 21
within each industry-year. This measure is similar to Schlingemann, Stulz, and Walkling's (2002)
and Ortiz-Molina and Phillips' (2013) measures of the extent of asset sales in an industry using
the aggregate dollar value of M&A scaled by the book value of industry assets. I use the number
of transactions in place of dollar values to alleviate concerns that the measure of M&A activity is
dominated by a small number of large dollar value deals in a given year.
To account for multi-segment firms, I follow Schlingemann, Stulz, and Walkling (2002)
and weight all real asset liquidity measures by the share of identifiable segment assets for each
firm’s distinct 3-digit SIC segments. Segment-weighting means that measures of real asset
liquidity will depart from a strictly industry-level definition for multi-segment firms. In addition,
I follow Ortiz-Molina and Phillips (2013) and use 3-year averages for all three real asset liquidity
measures based on transactions occurring over years t-2 through year t in order to capture resale
information available at time t. In addition, I conduct a principal component analysis to identify a
single real asset liquidity factor common to the three liquidity measures. Appendix A provides
detailed calculations for all variables.
To alleviate concerns that industry-based measures of real asset liquidity may be merely
capturing variation in product market activity across industries, I also focus on the airline
industry to calculate a firm-specific measure of real asset liquidity. Pulvino (1998) and Gavazza
(2011) utilize data on transactions in the secondary market for used aircraft to measure the
liquidity of each make and model of aircraft in a given period. I use the aircraft history dataset
from Gavazza (2011) detailing worldwide commercial jet operators from 1963 through April
2003 to calculate an aircraft liquidity ratio, measured as the number of planes that are resold on
the secondary market scaled by the number of aircraft in operation for each make and model of
REAL ASSET LIQUIDITY AND ASSET IMPAIRMENTS 22
aircraft.
13
This measure of resale activity is matched with fleet information on the number and
type of planes operated as of fiscal year-end for firms in the scheduled and non-scheduled air
transportation (SIC 4512 and 4522, respectively) and air courier (SIC 4513) industries with 10-K
reports available on SEC’s EDGAR database and with underlying data in Compustat. Aircraft
fleet liquidity is then calculated using a firm-specific weighted average of the aircraft liquidity
ratio based on the towing weight of each aircraft type within the airline firm’s fleet, both for the
entire fleet and the portion of each firm’s fleet that is owned and not leased. Measures used for
airline industry tests are detailed in the notes to Table 4.
It’s important to note that I assume that write-down decisions are made at the same asset
group level as my measures of real asset liquidity. So if write-downs occur on a business unit
level, then M&A activity measured at the business unit level is the appropriate benchmark for the
impairment decision. Examples in Appendix B of language in 10-K reports for firms recording
write-offs indicates some variation in the level of asset groups at which firms discuss recorded
impairments. As a result, I examine multiple real asset liquidity measures to identify appropriate
benchmarks for the impairment decision across firms.
Dependent Variables and Regression Models
Because real asset liquidity is determined by the resale market for a firm’s assets, real
asset liquidity is relatively exogenous to a firm’s individual accounting decisions. As a result,
primary tests rely on pooled Logit and OLS regression models. Hypothesis 1 predicts a link
between real asset liquidity and the frequency of asset impairments. Prior to 2000, fixed asset
and goodwill impairments are generally included as negative special items in earnings and are
only separately tracked in Compustat after 2000. To give an idea as to the assets that are
involved in fixed asset write-offs tracked in Compustat, Appendix C details results of hand
13
I am grateful to Alessandro Gavazza for sharing his aircraft fleet liquidity data.
REAL ASSET LIQUIDITY AND ASSET IMPAIRMENTS 23
coding 200 randomly selected write-offs with underlying data able to be matched to 10-K reports
on SEC’s EDGAR database.
14
Fixed asset write-offs tend to involve PPE assets (at least 50% of
write-offs involve PPE), with long-term equity investments (26%) and intangible assets (35%)
comprising a significant portion of write-offs, as well. In contrast, short-term assets that are
governed under separate impairment standards such as accounts receivable and inventory
comprise a minority of fixed asset write-offs in Compustat.
To measure the relation with impairment frequency, I use a version of the determinants
model for long-lived asset impairments in Riedl (2004). While Riedl uses a Tobit model to
examine both the likelihood and magnitude of impairments jointly, I conduct a Logit regression
to focus on the likelihood of a write-off. Riedl (2004) includes economic factors and earnings
management incentives as determinants of the decision to write-off assets. I add to the Riedl
(2004) model controls for volatility, asset tangibility, and product market competition, discussed
below. In place of controlling for GDP and industry median return on assets (ROA), I include
fiscal year dummies and detailed controls for firm performance.
15
The final model employed to
examine the determinants of asset impairments is a Logit model of the following form:
Indicator(Write-off
t
=1) = α + β
1
*Real Asset Liquidity
it
+ Σγ
i
* Economic factors
it
+
Σδ
i
*Competition controls
it
+ Σλ
i
*Earnings management incentives
it
+ ϵ
it
(1)
To examine the magnitude of impairment conditional on a write-off being recorded, I
employ a similar determinants model to that specified in equation (1) with one primary
difference. To capture the extent to which impairment amounts are conditional on the existing
carrying value of the assets, such that assets that are more fully depreciated will generally require
14
Note that 9 write-offs were selected from Compustat, but were unable to be matched to an underlying 10-K.
15
In addition, Riedl requires Execucomp data to track changes in senior management. I examine requiring this
control in robustness tests, but do not require this in my main analysis due to the reduction in number of
observations available for Execucomp firms. Results are qualitatively similar for the Execucomp subsample.
REAL ASSET LIQUIDITY AND ASSET IMPAIRMENTS 24
a lower impairment, I add an interaction term between each measure of real asset liquidity and
the unlevered B/M ratio as of the end of prior year. I expect that conditional on the beginning of
period B/M ratio, firms with more liquid real assets will record smaller write-down amounts. The
sample used for tests of the relation between real asset liquidity and the magnitude of recorded
impairments is limited to those firm-years with a fixed asset or goodwill write-down, depending
on the measure of real asset liquidity considered. Write-down amounts are scaled by beginning
of year market capitalization, measured by the total market value of common equity plus the
book value of total debt.
To test Hypothesis 3, I examine the effect of real asset liquidity on the timely recognition
of losses. I first examine the conditional conservatism of write-off amounts using a Basu (1997)
regression model with interaction terms for beginning of period B/M ratio, firm size, and
leverage added as determinants of conditional conservatism documented by Khan and Watts
(2009). I use write-off amounts as the dependent variable in the Basu model in place of net
income in order to focus on the timeliness of write-offs separately from remaining components
of earnings. Lawrence, Sloan, and Sun (2013) refer to the additional controls in the Basu model
as determinants of non-discretionary or “normal” conservatism. I add to this model an indicator
variable, LIQDum, tracking firms with high vs. low real asset liquidity, where I code firms in the
top quartile of real asset liquidity as +0.5 and firms in the bottom quartile of real asset liquidity
as -0.5, with remaining quartiles coded as 0. The γ
7
coefficient in the following regression
captures the difference in asymmetric timeliness for write-off amounts when moving from the
lowest to the highest quartile of real asset liquidity, while controlling for determinants of
conditional conservatism documented in prior literature:
REAL ASSET LIQUIDITY AND ASSET IMPAIRMENTS 25
WO
it
= α
0
+ α
1
*LIQDum
it
+ γ
2
*D
it
+ γ
3
*D
it
*LIQDum
it
+ γ
4
*ARET
it
+ γ
5
*ARET
it
*D
it
+
γ
6
*ARET
it
*LIQDum
it
+ γ
7
*ARET
it
*D
it
*LIQDum
it
+ Σλ*CScore_Controls+ ε
it
(2)
To alternatively capture the cumulative effect of timely loss recognition, I examine end of
period book-to-market (B/M) ratios as a measure of asset valuation after the impairment decision
in year t. These tests are similar in spirit to those used by Bini and Penman (2013) to examine
differences in goodwill impairment standards and B/M ratios in Europe relative to the US. While
B/M is used as a proxy for risk in asset pricing work (Fama and French, 1992; 1993) and for
growth opportunities (see work in finance by Lindenberg and Ross, 1981) among other
constructs, the B/M ratio is a natural measure of asset valuation. Indeed, Roychowdhury and
Watts (2007) show that the cumulative effect of conditionally conservative reporting is reflected
in lower end of period B/M ratios. To avoid capturing the effect of financial leverage, I calculate
B/M ratio using the book values of common equity and total debt scaled by the market value of
common equity and the book value of total debt. This unlevered B/M ratio assumes that the book
value of total debt adequately captures market value.
To examine Hypothesis 4 concerning the information content of accounting values, I
follow the approach taken by Hann, Heflin, and Subramanayam (2007) in their examination of
the value relevance of book values and earnings under alternative pension accounting schemes.
These authors conduct price-level value relevance regressions of the following form:
P
it
= α
0
+ β
1
*BVPS
it
+ β
2
*EPS
it
+ Σγ
t
*I
t
+ ϵ
it
(3)
where I
t
is an indicator for fiscal year to control for time varying characteristics affecting the
value relevance relation. Using price per share as the dependent variable rather than returns has
specification advantages despite suffering from heteroskedasticity when estimating the
regression model (see Kothari and Zimmerman, 1995 for a discussion). I compare the R-squared
REAL ASSET LIQUIDITY AND ASSET IMPAIRMENTS 26
for the above regression run on observations in the highest vs. lowest quartile of the real asset
liquidity distribution. In addition, I compare the R-squared within each subsample (high and low
real asset liquidity separately) using a Vuong test for significance of the R-squared differences
when using EPS, BVPS, and (BVPS + asset write-offs) as alternative independent variables in the
regression model.
Finally, Hypothesis 5 predicts an association between real asset liquidity and changes in
information asymmetry around earnings announcements. Barron et al. (1998) show that analyst
forecast dispersion has a component that captures information asymmetry in addition to
uncertainty. To disentangle these effects, the authors propose a model for decomposing forecast
dispersion into its relevant components. The intuition underlying the Barron et al. (1998)
decomposition stems from the fact that dispersion in forecasts around the mean and error in the
mean forecast differentially reflect error in analysts’ private vs. common information sets,
respectively. Barron et al. assume that analysts’ private information sets reflect the degree of
information asymmetry between informed and uninformed investors.
16
I examine changes in the
information asymmetry component of this measure during the month around the earnings
announcement using the calculation in Barron, Stanford, and Yu (2009), detailed in Appendix A.
Analyst forecast dispersion should decrease for firms taking write-offs to liquid real assets,
consistent with more informative impairments for these firms.
Control Variables and Alternative Explanations
Control variables are included from prior research on impairments. Perhaps the most
important alternative explanation to control for is the effect of underlying firm performance. If
poor performance is correlated with asset sales in an industry (which may be the case if firms sell
16
This assumption is consistent with viewing analysts as relatively informed investors who are likely to have more
information than uninformed investors when a greater proportion of analysts’ information is private in nature.
REAL ASSET LIQUIDITY AND ASSET IMPAIRMENTS 27
assets to raise funds in down years), then the use of resale activity to measure real asset liquidity
may result in merely capturing the relation between poor performance and impairment. To
address this issue, I control for underlying firm performance using linear controls in the main
analysis and using entropy balancing tests to generate a matched control sample (discussed in
section 5.2). In addition, to control for asset dispositions themselves driving the results in place
of an information effect related to real asset liquidity, I re-run impairment frequency and
magnitude tests after excluding firms with asset sales and/or discontinued operations on the cash
flow statement, or where the firm is listed as a target or an acquirer in SDC in a given year.
17
In terms of controls for the effect of firm performance on impairments, work by Barth,
Beaver, and Landsman (1998) demonstrates that financial distress will influence the relevance of
book value vs. earnings components. To capture economic performance of the firm that may
indicate financial distress and therefore asset impairment, I include changes in the firm’s pre-
impairment return on assets from period t-1 to t (d_PreIB) and the percentage change in sales
over this period (Rev_Growth). In addition, I include equity returns for year t and year t-1 (RET
t
and RET
t-1
, respectively) to capture the market’s evaluation of firm performance. In robustness
checks, I also include Altman's (1968; 2000) credit-risk score (Z_Score).
A second alternative explanation that may account for a relation between real asset
liquidity measures and financial reporting is related to operating risk. Ortiz-Molina and Phillips
(2013) find that real asset illiquidity is associated with higher firm-level implied cost of equity
capital. To control for operating risk and volatility that may drive both impairments and asset
sales, I include the standard deviation of 12 months of equity returns ending 3 months after the
end of fiscal year t (sd_RET) and the natural log of the unlevered market value of the firm
(log_MCAP), as I expect that larger firms are able to weather a given operating shock more
17
I also include fiscal year dummy variables to control for the potential effect of merger waves on the results.
REAL ASSET LIQUIDITY AND ASSET IMPAIRMENTS 28
easily than a smaller firm. In a related vein, I also include the unlevered B/M ratio measured at
the end of year t-1 (BM
t-1
). While the B/M ratio as of year t is a dependent variable listed above,
lagged B/M ratio will act as a control for growth options at the firm and the effect of past
impairment and depreciation activity on the book value of assets. I expect that firms operating in
more volatile industries will record more depreciation to assets in a given period, resulting in
lower beginning of period B/M ratios, all else equal.
Some research also indicates an effect of financial leverage on impairment. Research by
Easton, Eddey, and Harris (1993) and Aboody, Barth, and Kasznik (1999) show that asset
revaluations under Australian and UK GAAP vary with debt-to-equity ratios. In studies on real
asset liquidity, Sibilkov (2009) notes a relation between real asset liquidity and leverage ratios.
To control for debt contracting effects on asset impairments, I include financial leverage
(Leverage) and an indicator for whether the firm has publicly rated debt (CR_Dum) as controls.
Third, I control for differences in asset tangibility. In contrast to measures of real asset
liquidity, measures of asset tangibility capture the stock of liquid assets held by firms, such as
cash and equivalents. I include an indicator variable for whether the firm operates in a high-
technology industry, given that firms in these industries generate relatively more value from
intangible assets (HiTech). In addition, I include measures of cash holdings (Cash) and in
robustness tests, I add a measure of depreciation rate for property, plant, and equipment assets
(DepRate) as an alternative to measuring high-technology industries.
18
Fourth, differences in product market competition may drive impairment decisions. For
instance, firms in more competitive industries may record more write-offs if competition erodes
the returns to an investment project more quickly relative to firms in less competitive industries.
18
Results are qualitatively similar and stronger in most cases when an alternative measure of intangible asset
intensity using expenditures on research and development and advertising scaled by total assets is used.
REAL ASSET LIQUIDITY AND ASSET IMPAIRMENTS 29
If measures of real asset liquidity are correlated with competition, then the relation documented
between real asset liquidity and impairments may be spurious. To address this concern, I include
determinants of competition identified by Karuna (2007) as the size of the market (Comp1),
product substitutability (Comp2), and barriers to entry (Comp3). Karuna (2007) provides
evidence consistent with these multi-dimensional measures capturing competition more
completely relative to a uni-dimensional industry concentration ratio. I also include a measure of
industry concentration measured by a Herfindahl-Hirschman Index calculated using the ratio of
firm sales to total industry sales (HHI_Sale) for completeness. Computation of these variables is
detailed in Appendix A.
Finally, it’s important to control for the effects of managerial discretion over the
recording of asset impairments. To the extent that managerial incentives and opportunities to
manage reported earnings vary across industries in line with measures of real asset liquidity, I
may falsely attribute the association between impairment and asset liquidity to information
effects rather than managerial discretion. To address this concern, I follow Bartov (1993),
Francis, Hanna, and Vincent (1996), and Riedl (2004) by including controls for incentives to
record big baths in earnings and to smooth earnings by accelerating impairment charges. To
proxy for these reporting incentives, these authors include separate proxies for when earnings are
“unexpectedly” high and when they are “unexpectedly” low. The EMI_down (EMI_up) variable
captures changes in pre-impairment earnings that are in the bottom (top) portion of the earnings
distribution to focus on performance ranges where managers are more likely to have incentives
to engage in big bath (smoothing) reporting behavior.
REAL ASSET LIQUIDITY AND ASSET IMPAIRMENTS 30
Sample
Data for the main analysis comes from Compustat and CRSP over the 2000-2012 period.
I exclude ADR firms and any firms or firm segments operating in the financial (SIC 6000-6999),
payroll (SIC 872), and regulated utility (SIC 4900-4999) industries due to capital restrictions
placed on the assets and due to regulatory restrictions placed on asset dispositions, respectively.
19
The sample begins in December 2000 as this is when Compustat begins tracking fixed asset and
goodwill write-offs separately from special items in earnings. To calculate measures of real asset
liquidity, I require at least 5 firms in the same 3-digit SIC industry with data available to
calculate industry asset sales and discontinued operations from the cash flow statement and
M&A activity available from SDC Platinum.
20
Finally, I gather analyst forecast dispersion for a
subsample of firms from I/B/E/S. The final sample covers 191 unique 3-digit SIC industries
comprising 33,641 firm-year observations. Sample selection procedures are detailed in Table 1.
All measures including control variables are described in Appendix A. I winsorize all
continuous variables (except for industry averages and variables that are ranks or natural logs) at
the extreme one percent levels. In regressions, I cluster standard errors on two dimensions to
control for time-series correlation (3-digit SIC industry clusters) and to control for cross-
sectional correlation (fiscal year clusters) among standard errors.
21
Table 2, Panel A provides descriptives for all variables considered in the analysis. Panel
A indicates that measures of real asset liquidity are slightly positively skewed with means higher
19
The following industries are measured at 2-digit in place of 3-digit SIC because the bulk of Compustat firms
operating in these industries are classified at the 2-digit level: 01, 02, 07, 08, 10, 14, 16, 17, 41, 47, 52, 56, 72, 76,
82, and 83. In addition, I combine the following industries in each set of parentheses because they operate in similar
product markets. This follows a similar approach taken by Hoberg and Phillips (2010) in their classification of firms
into 3-digit SIC industries: (311, 315, 316, 317, 319), (551, 552, 554, 559), (571, 572), and (752-754).
20
This requirement is weakened from the requirement for 10 firms in the Schlingemann, Stulz, and Walkling (2002)
study as I examine 3-digit SIC industries in place of 2-digit SIC.
21
Logit and Tobit model standard errors are clustered one-way by industry given the difficulty in estimating two-
ways clustered standard errors for these non-linear models.
REAL ASSET LIQUIDITY AND ASSET IMPAIRMENTS 31
than medians. While the real asset liquidity factor is normalized to minimize the effect of
outlying observations, the asset sales measure is scaled by the book value of total industry assets
and the remaining two measures, discontinued operations and M&A activity, employ logs to
minimize the effect of extreme observations. Measures of real asset liquidity are highly
correlated with industry membership but will differ across time and across firms in the same 3-
digit SIC due to the presence of multi-segment firms. Indeed, I find in untabulated analysis that
industry membership accounts for 91% of the variation in the real asset liquidity factor, while
industry membership accounts for 76%-89% of the variation in the three remaining measures of
real asset liquidity. Standard errors are clustered by 3-digit SIC industry to address correlations
in standard errors within industry groups.
Table 2, Panel B provides correlations among key variables used in the analysis. For
brevity, only the real asset liquidity factor is included in Panel B. However, to ensure robust
results, I run all analyses using all four measures of real asset liquidity. Correlations in Panel B
indicate that real asset liquidity is indeed negatively correlated with firm performance, positively
correlated with return volatility, and positively correlated with high-technology industries during
the sample period. Controlling for these factors is important in subsequent analyses.
Table 2, Panel C presents the most and least liquid industries based on average real asset
liquidity over the entire sample period. Items in bold are those industries ranked as most and
least liquid that overlap across at least two of the real asset liquidity measures. The ranking of
industries appears intuitive. Several of the most liquid industries involve mobile assets such as
automotive rentals (SIC 751) or contain numerous participants such as computer and data
processing services (SIC 737). In contrast, illiquid industries are generally those requiring large
investments in concentrated industries, such as pipelines, except natural gas (SIC 461), local and
REAL ASSET LIQUIDITY AND ASSET IMPAIRMENTS 32
interurban passenger transit (SIC 410), and broadwoven fabric mills (SIC 222). Few buyers are
likely present for these assets.
Results
Main Analysis
Consistent with real asset liquidity influencing impairment decisions, Panel A of Table 3
indicates that write-offs are significantly more frequent for firms with more active asset resale
markets. After controlling for economic, competition, and earnings management factors
associated with impairment in model 2, the real asset liquidity factor remains significantly
positively associated with the likelihood of an asset impairment, with a predicted increase of
7.0% in the probability of recognizing a fixed asset or goodwill write-off when moving from the
lowest to the highest factor value. Fixed asset or goodwill write-offs occur in 21.4% of firm-
years during the sample window, indicating a predicted change of 32.5% relative to the baseline
write-off frequency when moving from the minimum to the maximum real asset liquidity factor
value. Results in model 3 show that impairments remain significantly more frequent for firms
with more liquid real assets even after removing firm-year observations where asset sales occur
and/or where the firm is a target or acquirer during the year. Overall, results provided in Panel A
show that information provided by a more active asset resale market significantly determines
impairment frequency in addition to economic factors included in the Riedl (2004) model.
Panel B of Table 3 provides results of tests for the three separate industry-level measures
of real asset liquidity that underlie the real asset liquidity factor in Panel A. Results in model 3
for the asset sales and discontinued operations measures of real asset liquidity show that the
positive relation between real asset liquidity and overall write-offs is concentrated in more
REAL ASSET LIQUIDITY AND ASSET IMPAIRMENTS 33
frequent fixed asset write-offs in the sample of firms with no asset dispositions during the year.
In contrast, goodwill write-offs show no significant differences in frequency for the M&A
measure of real asset liquidity in the rightmost columns of Table 3, Panel B.
22
This difference
may be due to mandatory annual testing frequency for goodwill impairments. Despite the
absence of more frequent goodwill write-offs, goodwill write-offs may still be timelier for firms
with more liquid real assets, as explored in tests below.
Impairment results in Table 3 are also in line with language in 10-K reports for firms
recording impairments. Appendix B details excerpts from 10-K reports selected from the sample
of firms with fixed asset write-offs in Compustat during the sample period. For the two example
firms using discounted cash flow valuation techniques to measure the impairment, language in
the 10-K refers to several of the assumptions used in valuing the assets. Language in the 2008
10-K report for L-1 Identity Solutions, Inc. indicates the following: “The use of the discounted
cash flow method requires significant judgments and assumptions of future events many of
which are outside the control of the Company, including estimates of future growth rates, income
tax rates, and discount rates, among others.” In contrast, language in the 2007 10-K report for
Steelcase, Inc. utilizes comparable market data to compute impairment estimates: “The MVA
[market value analysis] was only calculated for International and North America excluding
consolidated dealers because these are the two reporting units where we could obtain comparable
market data.” These examples are consistent with the availability of observable resale
information influencing impairment decisions and estimates.
Table 4 provides results of impairment frequency tests within the airline and air courier
industries. Panel A of Table 4 provides descriptive statistics for the sample of 69 firm-years with
22
Frequency results are unchanged when controlling for the amount of goodwill recorded on the balance sheet at the
beginning of the period in place of requiring observations to have positive goodwill at the start of the period.
REAL ASSET LIQUIDITY AND ASSET IMPAIRMENTS 34
available aircraft fleet liquidity from the aircraft history dataset in Gavazza (2011), with 10-K
reports available after 1995 on SEC’s EDGAR database, and with financial information available
in Compustat. Table 4, Panel B shows that aircraft impairments are significantly more frequent
for firms with more liquid aircraft fleets after controlling for differences in asset carrying values,
firm performance, volatility, and size. This positive relation is present for a measure of fleet
liquidity using the entire aircraft fleet at each firm and is slightly stronger when measured using
only those aircraft in the fleet that are owned and not leased by the firm. When using measures of
owned aircraft liquidity, two firm-year observations with entirely leased fleets are dropped from
the analyses. Aircraft fleet liquidity is associated in model 3 (6) with a predicted 1.0% (5.9%)
increase in the likelihood of an aircraft impairment for a one-standard deviation change, relative
to an unconditional aircraft impairment frequency of 26%. Results in Table 4 using firm-specific
aircraft fleet liquidity measures are consistent with evidence using industry-level measures in
Table 3 of more frequent impairments for firms with more liquid real assets.
Table 5 details results of impairment magnitude tests for the sample of firms recording a
fixed asset and/or goodwill write-off during the year. OLS regression results for the real asset
liquidity factor show that impairments are actually larger, not smaller, in model 1 for firms with
more liquid real assets.
23
In contrast, results in model 2 show that impairments are indeed smaller
for firms with greater real asset liquidity after including an interaction term between B/M and
real asset liquidity, as indicated by the significantly positive main effect of the RAL_Factor in
model 2. This is consistent with real asset liquidity allowing for more precise estimates of asset
value, permitting firms to record impairments for a smaller given shift in estimated asset value.
23
When using a Tobit model on the full sample of firm-years to measure the magnitude of the write-down while
accounting for the right censoring of write-down amounts, results show a smaller magnitude of impairments for
goodwill write-offs when using the M&A measure of real asset liquidity. Remaining models show no difference in
write-off magnitude across real asset liquidity measures.
REAL ASSET LIQUIDITY AND ASSET IMPAIRMENTS 35
The .005 coefficient on the RAL_Factor indicates that for a one-standard deviation increase in
the real asset liquidity factor, impairments are predicted to be smaller by 0.5% of the firm’s total
market capitalization. In addition, the significantly negative coefficient on the interaction term
with beginning B/M indicates that impairment amounts commove more strongly with asset
carrying values for firms with higher real asset liquidity. This greater comovement with
beginning B/M is present for all three remaining measures of real asset liquidity Table 5, while
impairments are significantly smaller for firms in industries with greater total asset sales.
To examine the timeliness of impairments for Hypothesis 3, I examine the conditional
conservatism of reported write-offs after controlling for non-discretionary determinants of
conservatism in Table 6 (Khan and Watts, 2009). Results in Table 6 shows that firms operating
in industries with more liquid M&A markets display greater conditional conservatism in
goodwill write-offs, consistent with these firms recognizing impairments to goodwill in a
timelier manner than firms with illiquid real assets. The significantly positive coefficient in
model 4 for the real asset liquidity interaction term in the modified-Basu model ( γ
7
) indicates
that for a one-standard deviation move in M&A real asset liquidity, asymmetric timeliness of
goodwill write-offs increases by 52% of the predicted shift for a one-standard deviation move in
lagged B/M ratio. Results in models 1 and 2 for the real asset liquidity factor and the asset sales
measures, respectively, display t-stats for the γ
7
coefficient that are significant in one-tailed tests.
Overall, there is some evidence of greater conditional conservatism in write-offs for firms with
more liquid real assets, particularly for goodwill write-offs. This is consistent with information
provided by resale markets allowing for timelier recognition of impairment losses.
As an alternative to examining the timeliness of write-offs in capturing returns in a given
period, Table 7 details results for the cumulative asset valuation tests to confirm that timely
REAL ASSET LIQUIDITY AND ASSET IMPAIRMENTS 36
recognition of impairments in earnings cumulates in more conservative book values for firms
with liquid real assets. These tests are similar in spirit to those used by Bini and Penman (2013)
to examine goodwill impairment differences in Europe relative to the US. Panel A of Table 7
shows that end of period unlevered B/M ratios are generally lower for firms with greater real
asset liquidity, both before and after controlling for fundamental determinants of B/M ratios.
Coefficients are significantly negative for the real asset liquidity factor, discontinued operations,
and M&A count measures, consistent with more conservative asset values relative to firm size.
Table 7, Panel B examines unlevered B/M ratios within the airline and air courier industries
using measures of aircraft fleet liquidity. Results in Panel B show that B/M is significantly lower
on average for firms operating more liquid aircraft fleets, both for overall fleet liquidity and for
owned aircraft liquidity. Overall, results in Table 7 show that as real asset liquidity increases,
assets are recorded at lower, more conservative values on the balance sheet.
Table 8 examines the value relevance of book values and earnings to provide evidence on
the information content of accounting values as predicted by Hypothesis 4. Panel A of Table 8
shows that prices of firms with liquid real assets tend to place greater weight on book values and
less weight on earnings relative to firms in the bottom quartile of real asset liquidity, as indicated
by significant coefficients on the interaction terms for BVPS and EPS variables. Panel B of Table
8 further shows that explanatory power doubles in regressions of equity prices on earnings and
book values when moving from the lowest to the highest quartile of the real asset liquidity
distribution across all four asset liquidity measures. To examine whether greater value relevance
for firms with liquid real assets is concentrated in book values, Panel B of Table 7 conducts a
Vuong test comparing the explanatory power for book values vs. earnings within each asset
liquidity quartile. Results show that firms with liquid real assets have more value relevant book
REAL ASSET LIQUIDITY AND ASSET IMPAIRMENTS 37
values when compared with earnings, consistent with book values explaining greater value
relevance for liquid firms. In contrast, firms with illiquid real assets have earnings and book
values that are similarly value relevant, with only marginal differences in explanatory power.
Table 8, Panel C further examines whether impairments account for the greater value
relevance of book values among firms with liquid real assets. Using a Vuong test, I compare
explanatory power from regressions of equity prices on reported book values vs. book values
with fixed asset and goodwill write-offs added back (essentially undoing the write-off). If asset
impairments improve the information content of book values, explanatory power should be
greater for reported book values relative to book values with write-offs added back. Indeed,
significantly negative Vuong test statistics in Table 8, Panel C show that reported book values
are significantly more value relevant for firms with liquid real assets. Surprisingly, book values
before write-offs are actually more strongly associated with equity prices for firms with illiquid
real assets when using real asset liquidity measures other than the total asset sales measure. This
evidence is consistent with theory offered in Ohlson (1995) and with results in Collins, Maydew,
and Weiss (1997) showing an increase in value relevance of book values for firms recognizing
one-time items in earnings. However, I find that write-offs improve the information content of
book values only among the subsample of firms with liquid real assets.
Table 8, Panel D provides value relevance coefficients for firms in the airline and air
courier industries. Results show that the weight on book value per share is significantly greater
for firms with above median total aircraft fleet liquidity as indicated by the significantly positive
coefficient on the interaction term with BVPS. In contrast, book values and earnings display
similar weights for the liquidity of the owned aircraft fleet. These results are weakly consistent
with results for the industry-level measures of real asset liquidity in Table 8, Panel A.
REAL ASSET LIQUIDITY AND ASSET IMPAIRMENTS 38
Finally, Hypothesis 5 predicts that information asymmetry between informed and
uninformed investors will decrease around earnings announcements for firms with liquid real
assets, consistent with timelier impairments leveling the playing field for less sophisticated
investors. Table 9 displays results of information asymmetry tests using the dAssym measure of
information asymmetry, calculated as the change in the information asymmetry component of
analyst forecast dispersion between the last monthly consensus analyst forecast of one-year
ahead earnings before the earnings announcement in year t and the first monthly consensus
forecast after the earnings announcement (Barron et al., 1998). As a result, changes in
information asymmetry are measured for the one-month period surrounding the earnings
announcement. Regressions in Table 9 add controls for the effect of trading volume (log_VOL)
and the number of analysts following the firm (Numest) relative to prior tables.
The baseline model in Table 9 shows that changes in the Assym measure of information
asymmetry are not significantly related to write-offs for the full sample of firms. Model 1 for
each asset liquidity measure shows that real asset liquidity is marginally associated with
decreases in information asymmetry. In contrast, Model 2 shows that information asymmetry
significantly decreases for firms with liquid real assets that record a write-off during the period
as measured by the significantly negative coefficient on the interaction between real asset
liquidity and the write-off indicator variable. The coefficient on the write-off main effect for tests
using measures other than the RAL_Factor may be interpreted as the effect of a write-off on
changes in information asymmetry for firms with zero values (the minimum value in the sample)
for each of the real asset liquidity measures. For these firms with illiquid real assets, write-offs
marginally increase information asymmetry as indicated by the significantly positive coefficient
on the write-off indicator when using discontinued operations and M&A activity to measure real
REAL ASSET LIQUIDITY AND ASSET IMPAIRMENTS 39
asset liquidity. These results show generally lower information asymmetry for firms with liquid
real assets, in line with work in finance by Gopalan, Kadan, and Pevzner (2012) showing a
positive relation between real asset liquidity and stock liquidity. Evidence provided here points
to the possibility that real asset liquidity results in greater liquidity (lower information
asymmetry) for the firm’s equity securities as a result of more informative asset impairments.
Robustness Tests
Entropy balancing is discussed in Hainmueller (2012) as an alternative to propensity-
score approaches to ensure that treated and control groups have similar distributions of key
control variables, known as covariate balance. In contrast to propensity-scores, entropy balancing
focuses directly on covariate balance by weighting control group observations to achieve balance
on the specified moments of the distribution.
24
To ensure that regression estimates are adequately
controlling for non-linear differences across firms with high vs. low real asset liquidity, I use
entropy balancing to weight observations in the bottom quartile of the real asset liquidity
distribution (control group) to achieve balance on the first (mean) and second (variance)
moments of the distribution relative to firms in the highest quartile of real asset liquidity
(treatment group) on key controls for performance, volatility, asset carrying value, competition,
and asset tangibility used in prior tests. This approach avoids manually iterating through
propensity-score models to examine whether balance is achieved. Table 10, Panel A shows that
covariate balance across high and low real asset liquidity subsamples improves significantly after
running the entropy balancing program.
Table 10, Panel B provides results of weighted least squares regressions run using the
entropy balancing weights calculated for control group observations. Panel B displays
coefficients for the entropy-balance weighted regression models in addition to coefficients from
24
See work by McMullin (2013) for an example of research in accounting that employs entropy balancing.
REAL ASSET LIQUIDITY AND ASSET IMPAIRMENTS 40
the main analysis. The sign and magnitude of coefficients is broadly consistent across the two
approaches, although the level of significance is generally weaker when using entropy balancing.
This may be due in part to the smaller number of observations, as observations in the middle
quartiles of the real asset liquidity distribution are excluded from entropy balancing tests.
Interestingly, goodwill write-offs are now significantly more frequent in entropy-balanced
regression tests when using the M&A measure of real asset liquidity.
As additional robustness tests, I examine the following. First, I include the balance sheet
liquidity index used by Gopalan, Kadan, and Pevzner (2012) based on earlier models by Berger,
Ofek, and Swary (1996) as an additional asset tangibility control. This approach weights cash
holdings at full value, other current assets at 75% of full value, and net tangible property at 50%
of full value. All other assets including intangibles receive a weight of zero. Second, I include a
control for goodwill scaled by lagged total assets, to separately control for goodwill carrying
values. Third, I replace the indicator for high-technology industries with a control for the
depreciation rate of property, plant, and equipment (DepRate) as a measure of the tangibility of
the firm’s assets. Finally, I control for Z-scores (Altman, 1968) in value relevance tests (Table 8)
by sorting firms into high and low Z-score samples to control for greater value relevance of book
values among distressed firms (Barth, Beaver, and Landsman, 1998). I find that my main results
(untabulated) are unchanged in response to these modifications, with the exception that the
difference in the value relevance of book values in Panel B of Table 8 is concentrated among
non-distressed firms.
REAL ASSET LIQUIDITY AND ASSET IMPAIRMENTS 41
Conclusion
I examine the effect of real asset liquidity on asset impairments and on the information
content of accounting values. Results show that write-offs are more frequent and somewhat
smaller for firms with liquid real assets, consistent with firms adjusting assets in response to
information provided by resale markets. Consistent with more timely loss recognition for firms
with liquid real assets, goodwill write-offs are more conditionally conservative and book values
are lower relative to firm size. These features of real asset liquidity are associated with higher
information content of accounting values, measured by value relevance of book values and
earnings, and with decreases in information asymmetry around earnings announcements for
firms with liquid real assets.
Future research may further examine the underlying reasons for why verifiability
influences complex estimates, such as impairment. If auditors are optimally requiring a
verifiability threshold in order to avoid including temporary fluctuations in fair value from the
financials or to counteract downward bias in estimates, then a verifiability threshold may actually
improve earnings quality for firms lacking verifiable benchmarks for complex estimates, such as
firms with low real asset liquidity. In contrast, auditors may be prone to biased unverifiable
estimates provided by management, leading to lower earnings quality for these firms.
In addition, existing research in corporate finance focuses on capital structure effects of
real asset liquidity. However, little evidence exists as to how real asset liquidity influences the
use of accounting information in debt contracting. Some initial evidence in Benmelech and
Bergman (2008) examines the role of collateral redeployability in the airline industry and its
effect on loan rates and on lease payment concessions during contract renegotiation. However,
this research does not examine the information actually used in the loan contracts. While the
REAL ASSET LIQUIDITY AND ASSET IMPAIRMENTS 42
current study focuses on the relation between real asset liquidity and the information role of
accounting, research into the contracting role of accounting may yield additional insights into
how the properties of accounting information differ across firms with more vs. less liquid assets.
REAL ASSET LIQUIDITY AND ASSET IMPAIRMENTS 43
References
Aboody, David, Mary E Barth, and Ron Kasznik. 1999. “Revaluations of Fixed Assets and
Future Firm Performance: Evidence from the UK.” Journal of Accounting & Economics
26 (1-3): 149.
Alderson, Michael J., and Brian L. Betker. 1995. “Liquidation Costs and Capital Structure.”
Journal of Financial Economics 39 (1): 45–69.
Almeida, Heitor, and Murillo Campello. 2007. “Financial Constraints, Asset Tangibility, and
Corporate Investment.” Review of Financial Studies 20 (5): 1429–60.
Almeida, Heitor, Murillo Campello, and Dirk Hackbarth. 2011. “Liquidity Mergers.” Journal of
Financial Economics 102 (3): 526–58.
Altamuro, Jennifer, and Haiwen Zhang. 2012. “The Financial Reporting of Fair Value Based on
Managerial Inputs versus Market Inputs: Evidence from Mortgage Servicing Rights.”
Review of Accounting Studies, 1–26.
Altman, Edward I. 1968. “Financial Ratios, Discriminant Analysis and the Prediction of
Corporate Bankruptcy.” The Journal of Finance 23 (4): 589–609. doi:10.2307/2978933.
———. 2000. “Predicting Financial Distress of Companies: Revisiting the Z-Score and ZETA
Models.” Stern School of Business, New York University, 9–12.
Barron, Orie E., Oliver Kim, Steve C. Lim, and Douglas E. Stevens. 1998. “Using Analysts’
Forecasts to Measure Properties of Analysts’ Information Environment.” The Accounting
Review 73 (4): 421–33.
Barron, Orie E., Mary Harris Stanford, and Yong Yu. 2009. “Further Evidence on the Relation
between Analysts’ Forecast Dispersion and Stock Returns.” Contemporary Accounting
Research/Recherche Comptable Contemporaine 26 (2): 329–57.
REAL ASSET LIQUIDITY AND ASSET IMPAIRMENTS 44
Barth, Mary E., William H. Beaver, and Wayne R. Landsman. 1998. “Relative Valuation Roles
of Equity Book Value and Net Income as a Function of Financial Health.” Journal of
Accounting and Economics 25 (1): 1–34.
Barth, Mary E., and Greg Clinch. 1998. “Revalued Financial, Tangible, and Intangible Assets:
Associations with Share Prices and Non-Market-Based Value Estimates.” Journal of
Accounting Research 36: 199–233.
Bartov, Eli. 1993. “The Timing of Asset Sales and Earnings Manipulation.” The Accounting
Review 68 (4): 840–55.
Basu, Sudipta. 1997. “The Conservatism Principle and the Asymmetric Timeliness of Earnings.”
Journal of Accounting & Economics 24 (1): 3.
Beatty, Anne, and Joseph Weber. 2006. “Accounting Discretion in Fair Value Estimates: An
Examination of SFAS 142 Goodwill Impairments.” Journal of Accounting Research 44
(2): 257–88.
Benmelech, Efraim, and Nittai K. Bergman. 2008. “Liquidation Values and the Credibility of
Financial Contract Renegotiation: Evidence from US Airlines.” The Quarterly Journal of
Economics 123 (4): 1635–77.
Berger, Philip G., Eli Ofek, and Itzhak Swary. 1996. “Investor Valuation of the Abandonment
Option.” Journal of Financial Economics 42 (2): 257–87.
Bhojraj, Sanjeev, Charles Lee, and Derek K. Oler. 2003. “What’s My Line? A Comparison of
Industry Classification Schemes for Capital Market Research.” Journal of Accounting
Research 41 (5): 745–74.
REAL ASSET LIQUIDITY AND ASSET IMPAIRMENTS 45
Bini, Mauro, and Stephen Penman. 2013. “Companies with Market Value below Book Value
Are More Common in Europe than in the US: Evidence, Explanations, and Implications.”
KPMG’s Global Valuation Institute.
Christensen, Hans, and Valeri Nikolaev. 2013. “Does Fair Value Accounting for Non-Financial
Assets Pass the Market Test?” Review of Accounting Studies 18 (3): 734–75.
Collins, Daniel W., Edward L. Maydew, and Ira S. Weiss. 1997. “Changes in the Value-
Relevance of Earnings and Book Values over the Past Forty Years.” Journal of
Accounting and Economics 24 (1): 39–67.
Dietrich, J. Richard, Mary S. Harris, and Karl A. Muller III. 2000. “The Reliability of Investment
Property Fair Value Estimates.” Journal of Accounting and Economics 30 (2): 125–58.
Easton, Peter D., Peter H. Eddey, and Trevor S. Harris. 1993. “An Investigation of Revaluations
of Tangible Long-Lived Assets.” Journal of Accounting Research 31: 1–38.
Elliott, John A., and Wayne H. Shaw. 1988. “Write-Offs As Accounting Procedures to Manage
Perceptions.” Journal of Accounting Research 26 (January): 91–119.
Fama, Eugene F, and Kenneth R French. 1993. “Common Risk Factors in the Returns on Stocks
and Bonds.” Journal of Financial Economics 33 (1): 3.
Fama, Eugene F., and Kenneth R. French. 1992. “The Cross-Section of Expected Stock
Returns.” The Journal of Finance 47 (2): 427.
Financial Accounting Standards Board (FASB). 1995. “Accounting for the Impairment of Long-
Lived Assets and for Long-Lived Assets to Be Disposed Of.” Statement of Financial
Accounting Standards No. 121. Norwalk, CT.
———. 2001a. “Accounting for the Impairment or Disposal of Long-Lived Assets.” Statement
of Financial Accounting Standards No. 144. Norwalk, CT.
REAL ASSET LIQUIDITY AND ASSET IMPAIRMENTS 46
———. 2001b. “Goodwill and Other Intangible Assets”. Statement of Financial Accounting
Standards No. 142. Norwalk, CT.
———. 2006. “Fair Value Measurements”. Statement of Financial Accounting Standards No.
157. Norwalk, CT.
Francis, Jennifer, J. Douglas Hanna, and Linda Vincent. 1996. “Causes and Effects of
Discretionary Asset Write-Offs.” Journal of Accounting Research 34: 117–34.
Francis, Jennifer, and Katherine Schipper. 1999. “Have Financial Statements Lost Their
Relevance?” Journal of Accounting Research, 319–52.
Gavazza, Alessandro. 2011. “The Role of Trading Frictions in Real Asset Markets.” The
American Economic Review 101 (4): 1106–43.
Gopalan, Radhakrishnan, Ohad Kadan, and Mikhail Pevzner. 2012. “Asset Liquidity and Stock
Liquidity.” Journal of Financial and Quantitative Analysis 1 (1): 1–70.
Hainmueller, Jens. 2012. “Entropy Balancing for Causal Effects: A Multivariate Reweighting
Method to Produce Balanced Samples in Observational Studies.” Political Analysis 20
(1): 25–46.
Hann, R., F. Heflin, and K. Subramanayam. 2007. “Fair-Value Pension Accounting*.” Journal
of Accounting & Economics 44 (3): 328.
Hayn, Carla, and Patricia J. Hughes. 2006. “Leading Indicators of Goodwill Impairment.”
Journal of Accounting, Auditing & Finance 21 (3): 223–65.
Hoberg, G., and G. Phillips. 2010. “Real and Financial Industry Booms and Busts.” The Journal
of Finance 65 (1): 45.
Karuna, Christo. 2007. “Industry Product Market Competition and Managerial Incentives*.”
Journal of Accounting & Economics 43 (2/3): 275.
REAL ASSET LIQUIDITY AND ASSET IMPAIRMENTS 47
Khan, M., and R. Watts. 2009. “Estimation and Empirical Properties of a Firm-Year Measure of
Accounting Conservatism.” Journal of Accounting & Economics 48 (2/3): 132.
Kothari, S.P., and Jerold L. Zimmerman. 1995. “Price and Return Models.” Journal of
Accounting and Economics 20 (2): 155–92.
Lawrence, Alastair, Subprasiri Siriviriyakul, and Richard G. Sloan. 2013. “Who’s the Fairest of
Them All? Evidence from Closed-End Funds”. SSRN Scholarly Paper ID 2330490.
Rochester, NY: Social Science Research Network.
Lawrence, Alastair, Richard G. Sloan, and Yuan Sun. 2013. “Non-Discretionary Conservatism:
Evidence and Implications.” Journal of Accounting & Economics 56 (2-3): 112–33.
Lindenberg, Eric B., and Stephen A. Ross. 1981. “Tobin’s Q Ratio and Industrial Organization.”
Journal of Business, 1–32.
McMullin, J. 2013. “Can I Borrow Your Footnotes? Footnote Similarity and Accounting
Comparability.” Working Paper, Indiana University.
Ohlson, James A. 1995. “Earnings, Book Values, and Dividends in Equity Valuation.”
Contemporary Accounting Research 11 (2): 661.
Ortiz-Molina, Hernán, and Gordon M. Phillips. 2013. “Asset Liquidity and the Cost of Capital.”
Journal of Financial and Quantitative Analysis, Forthcoming.
http://www.nber.org/papers/w15992.
Pulvino, Todd C. 1998. “Do Asset Fire Sales Exist? An Empirical Investigation of Commercial
Aircraft Transactions.” The Journal of Finance 53 (3): 939–78.
Ramanna, Karthik, and Ross L. Watts. 2012. “Evidence on the Use of Unverifiable Estimates in
Required Goodwill Impairment.” Review of Accounting Studies 17 (4): 749–80.
REAL ASSET LIQUIDITY AND ASSET IMPAIRMENTS 48
Ramey, Valerie A., and Matthew D. Shapiro. 2001. “Displaced Capital: A Study of Aerospace
Plant Closings.” Journal of Political Economy 109 (5): 958–92.
Riedl, E., and G. Serafeim. 2011. “Information Risk and Fair Values: An Examination of Equity
Betas.” Journal of Accounting Research 49 (4): 1083.
Riedl, Edward J. 2004. “An Examination of Long-Lived Asset Impairments.” The Accounting
Review 79 (3): 823–52.
Roychowdhury, Sugata, and Xiumin Martin. 2013. “Understanding Discretion in Conservatism:
An Alternative Viewpoint.” Journal of Accounting and Economics 56 (2): 134–46.
Roychowdhury, Sugata, and Ross L Watts. 2007. “Asymmetric Timeliness of Earnings, Market-
to-Book and Conservatism in Financial Reporting.” Journal of Accounting & Economics
44 (1/2): 2.
Schlingemann, Frederik P., Rene M. Stulz, and Ralph A. Walkling. 2002. “Divestitures and the
Liquidity of the Market for Corporate Assets.” Journal of Financial Economics 64 (1):
117–44.
Shleifer, Andrei, and Robert W. Vishny. 1992. “Liquidation Values and Debt Capacity: A
Market Equilibrium Approach.” The Journal of Finance 47 (4): 1343–66.
Sibilkov, Valeriy. 2009. “Asset Liquidity and Capital Structure.” Journal of Financial and
Quantitative Analysis 44 (05): 1173–96.
Song, C., W. Thomas, and H. Yi. 2010. “Value Relevance of FAS No. 157 Fair Value Hierarchy
Information and the Impact of Corporate Governance Mechanisms.” The Accounting
Review 85 (4): 1375.
Williamson, Oliver E. 1988. “Corporate Finance And Corporate Governance.” The Journal of
Finance 43 (3): 567.
REAL ASSET LIQUIDITY AND ASSET IMPAIRMENTS 49
Appendix A
Variable Definitions
Variable Definition Data source
Real asset liquidity measures
AT_SaleSW
(ATS)
Asset sales on the investing section of the cash flow statement for the firm's
3-digit SIC industry scaled by the book value of total industry non-cash
assets. Measures are averaged over years t-2 through t to capture stable
information on resale activity available at time t and are weighted by the
share of firm i’s identifiable segment assets for firms with multiple
Compustat segments. Measures are similar in spirit to the measure of
activity in asset resale markets developed using hand-collected data from the
Bureau of Census’ Economic Census in Almeida and Campello (2007) and
in Almeida, Campello, and Hackbarth (2011).
Compustat
DOC_CountSW
(DOC)
Natural log of the number of firms with a gain/loss on discontinued
operations on the cash flow statement for the firm's 3-digit SIC industry.
Measures are averaged over years t-2 through t to capture stable information
on resale activity available at time t and are weighted by the share of firm i’s
identifiable segment assets for firms with multiple Compustat segments.
Measures are similar in spirit to the measure of activity in asset resale
markets developed using hand-collected data from the Bureau of Census’
Economic Census in Almeida and Campello (2007) and in Almeida,
Campello, and Hackbarth (2011).
Compustat
LIQDum
Indicator tracking high vs. low real asset liquidity firms calculated as -0.5
for firms in the lowest quartile, 0 for firms in the middle quartiles, and +0.5
for firms in the highest quartile of the specified real asset liquidity
distribution each year.
Compustat /
SDC
Platinum
MA_CountSW
(MAC)
Natural log of the number of merger and acquisition transactions in the
firm's 3-digit SIC industry. Measures are averaged over years t-2 through t
to capture stable information on resale activity available at time t and are
weighted by the share of firm i’s identifiable segment assets for firms with
multiple Compustat segments. Corporate transactions include all disclosed
and completed leveraged buyouts, tender offers, exchange offers, stake
purchases, privatizations, and spinoffs. Buybacks (repurchases and self-
tenders) and recapitalizations are excluded. These measures use transactions
involving buyers both from inside and outside the 3-digit SIC industry, and
thus do not rely on assumptions about the transferability of assets across
industries. Measures are similar in spirit to measures of the dollar value of
M&A activity used by Ortiz-Molina and Phillips (2013) and by
Schlingemann, Stulz, and Walkling (2002).
SDC
Platinum
RAL_Factor
First principal component from a principal components analysis over the
three measures of real asset liquidity: AT_SaleSW, DOC_CountSW, and
MA_CountSW. By construction, the mean of the RAL_Factor component is
zero and the standard deviation is one.
Compustat /
SDC
Platinum
REAL ASSET LIQUIDITY AND ASSET IMPAIRMENTS 50
Appendix A, continued
Variable Definition Data source
Dependent variables
dAssym
Change in information asymmetry based on the decomposition in Barron
et al. (1998) of analyst forecast dispersion into uncertainty and
information asymmetry components. I follow Barron, Stanford, and Yu
(2009) in calculating the information asymmetry component as 1 – (SE –
D/n) / [(1-1/n)D + SE] where SE is squared error in the mean forecast
measured as the difference between earnings per share and the mean
forecast (EPS – Mean forecast)
2
, D is forecast dispersion measured as the
variance of the individual forecasts in the consensus around the mean
forecast, and n is the number of individual forecasts. I take the difference
in asymmetry for forecasts of year t+1 earnings for the first consensus
analyst forecast following the year t earnings announcement and the last
consensus analyst forecast prior to the year t earnings announcement.
I/B/E/S
dDisp
Change in analyst earnings forecast dispersion measured as the change in
the standard deviation of the individual forecasts in the consensus around
the mean forecast. I take the difference in dispersion for forecasts of year
t+1 earnings for the first consensus analyst forecast following the year t
earnings announcement and the last consensus analyst forecast prior to
the year t earnings announcement.
I/B/E/S
BM (t)
Book-to-market ratio calculated as (Book value of common equity +
Book value of debt) / (Total market-value of the firm’s outstanding
equity securities + Total book value of debt) at fiscal year-end for year t.
Compustat
BVPS
Book value of equity per share calculated as Book value of equity /
Common equity shares outstanding 3 months after fiscal year-end.
Compustat/
CRSP
EPS
Earnings per share calculated as Income before extraordinary items /
Common equity shares outstanding 3 months after fiscal year-end.
Compustat/
CRSP
FWO
Negative fixed asset write-offs in earnings calculated as (Fixed asset
write-offs
t
< 0) / (Total market-value of the firm’s outstanding equity
securitiest-1 + Total book value of debt
t-1
). Note that fixed asset write-
offs are only separately tracked in Compustat beginning in 2000.
Compustat
FWO_Dummy
Indicator set to 1 where fixed asset write-offs in year t are less than zero,
and equal to 0 otherwise.
Compustat
GWO
Negative goodwill write-offs in earnings calculated as (Goodwill write-
offs
t
< 0) / (Total market-value of the firm’s outstanding equity
securities
t-1
+ Total book value of debt
t-1
). Note that goodwill write-offs
are only separately tracked in Compustat beginning in 2000.
Compustat
GWO_Dummy
Indicator set to 1 where goodwill write-offs in year t are less than zero,
and equal to 0 otherwise.
Compustat
PRC_3M
Equity price per share 3 months after fiscal year-end. See Hann, Heflin,
and Subramanayam (2007) for details on the use of this measure in value
relevance tests.
CRSP
WO
Negative asset write-offs in earnings calculated as (Fixed asset write-offs
t
< 0 + Goodwill write-offs
t
< 0) / (Total market-value of the firm’s
outstanding equity securities
t-1
+ Total book value of debt
t-1
). Note that
fixed asset and goodwill write-offs are only separately tracked in
Compustat beginning in 2000.
Compustat
WO_Dummy
Indicator set to 1 where total asset write-offs in year t are less than zero,
and equal to 0 otherwise.
Compustat
WOPS
After-tax total asset write-offs (fixed asset + goodwill) per share
calculated as After-tax write-offs
t
/ Common equity shares outstanding 3
months after fiscal t year-end.
Compustat/
CRSP
REAL ASSET LIQUIDITY AND ASSET IMPAIRMENTS 51
Appendix A, continued
Variable Definition Source
Controls
Cash (R)
Cash holdings calculated as Cash and cash equivalents
t
/ Book value of total
assets
t-1
. Ranked values (CashR) control for positive skewness in the raw measure.
Compustat
Comp1
Market size calculated as log( Total industry sales
t
) for each 3-digit SIC industry,
consistent with Karuna's (2007) definition.
Compustat
Comp2
Product substitutability calculated as Total industry sales
t
/ (Total industry sales
t
–
Total industry operating income after depreciation
t
) for each 3-digit SIC industry,
consistent with Karuna’s (2007) definition.
Compustat
Comp3
Product market entry costs calculated as log( Sales dollar weighted gross PPE
t
) for
each 3-digit SIC industry, consistent with Karuna’s (2007) definition.
Compustat
CR_Dum
Indicator set to 1 if a firm has debt publicly rated by Standard & Poor’s, and 0
otherwise.
Compustat
D
Indicator set to 1 if equity returns over the 12 months ending 3 months after fiscal
year-end are less than zero, and 0 otherwise. See Basu (1997) for further details.
CRSP
d_PreIB
Change in pre-write-off earnings from period t-1 to t calculated as Δ[(Income
before extraordinary items
t+n
– Fixed asset write-offs
t+n
– Goodwill write-offs
t+n
)] /
Book value of total assets
t-1.
Compustat
DepRate
Depreciation rate for tangible fixed assets calculated as Gross property, plant, and
equipment
t
/ Depreciation and amortization expense
t
. This measures tangibility of
the firm's assets using the average number of years before PPE is fully depreciated.
Compustat
EMI_up
(down)
Earnings smoothing (big-bath) incentive measured following Riedl (2004) as
Δ[(Income before extraordinary items
t+n
– Fixed asset write-offs
t+n
)] / Book value
of total assets
t-1
when this change is above (below) the median of nonzero positive
(negative) values of this variable, and 0 otherwise.
Compustat
HHI_Sale
Industry concentration ratio calculated using a sales Herfindahl Index where Σ[
(Sales
it
/ Total industry sales
t
)
2
].
Compustat
HiTech
Indicator set to 1 if a firm operates in a high-technology industry based on the list
of SIC industries in Srivastava (2011) Appendix B. This list is an expanded form
of the list of high-technology industries in Francis and Schipper (1999).
Compustat
lag_GDWL
Recorded book value of goodwill calculated as Goodwill recorded on balance
sheet
t-1
/ Book value of total assets
t-1
.
Compustat
Leverage
Financial leverage calculated as (Current portion of long-term debt
t
+ Non-current
portion of long-term debt
t
) / Book value of total assets
t.
Compustat
log_MCAP
Natural log of market capitalization calculated as log( Total market-value of the
firm’s outstanding equity securities at fiscal year-end
t
).
Compustat
log_VOL
Natural log of equity trading volume calculated as log( Total CRSP trading volume
for the 6-month period ending 3 months after fiscal year-end ).
CRSP
Numest
Analyst following measured as the number of analysts included in the final
consensus forecast of annual earnings prior to the earnings announcement.
I/B/E/S
PreROA
Return on assets before write-offs calculated as (Income before extraordinary
items
t
- Fixed asset write-offs
t
- Goodwill write-offst) / Book value of total assets
t-1.
Compustat
RET (t+n)
Buy and hold equity return for the 12 months ending 3 months after fiscal t+n
year-end calculated as Π (1 + Return
it
) for returns in month t for firm i.
CRSP
Rev_Growth Percentage change in total sales calculated as (Total revenue
t
/Total revenue
t-1
) – 1. Compustat
sd_RET
Standard deviation of monthly equity returns for the 12 month period ending 3
months after fiscal t year-end.
CRSP
Z_Score
Altman's (1968; 2000) credit risk score calculated as: Z’ = 0.717(X1) + 0.847(X2)
+ 3.107(X3) + 0.420(X4) + 0.998(X5) where X1 is working capital/total assets, X2
is retained earnings/total assets, X3 is earnings before interest and taxes/total
assets, X4 is book value of equity/book value of total liabilities, and X5 is
sales/total assets.
Compustat
REAL ASSET LIQUIDITY AND ASSET IMPAIRMENTS 52
Appendix B
Impairment Testing and Discussion: 10-K Report Examples
Firms using discounted cash flow valuation techniques to estimate impairment:
9/30/2003 10-K report for Skyworks Solutions, Inc. (CIK 0000004127):
During the fourth quarter of fiscal 2003, we recorded a $26.0 million charge for the impairment
of assets related to certain infrastructure products manufactured in our Woburn, Massachusetts
and Adamstown, Maryland facilities. The Woburn facility primarily manufactures
semiconductor products based on both silicon wafer technology and gallium arsenide
technology. Our Adamstown, Maryland facility primarily manufactures ceramics components.
We experienced a significant decline in factory utilization resulting from a downturn in the
market for products manufactured at these two facilities and a decision to discontinue certain
products. The impairment charge was based on a recoverability analysis prepared by
management based on these factors and the related impact on our current and projected outlook.
We projected lower revenues and new order volume for these products and management
believed these factors indicated that the carrying value of the related assets (machinery,
equipment and intangible assets) may have been impaired and that an impairment analysis
should be performed.
In performing the analysis for recoverability, management estimated the future cash flows
expected to result from these products over a five-year period. Since the estimated undiscounted
cash flows were less than the carrying value of the related assets, it was concluded that an
impairment loss should be recognized. In accordance with SFAS No. 144, Accounting for the
Impairment or Disposal of Long-Lived Assets, the impairment charge was determined by
comparing the estimated fair value of the related assets to their carrying value. The fair value of
the assets was determined by computing the present value of the estimated future cash flows
using a discount rate of 16%, which management believed was commensurate with the
underlying risks associated with the projected future cash flows. Management believes the
assumptions used in the discounted cash flow model represented a reasonable estimate of the fair
value of the assets.
12/31/2008 10-K report for L-1 Identity Solutions, Inc. (CIK 0001018332):
The asset impairments consist of goodwill of $430.0 million and long-lived assets of $98.6
million, principally intangible assets recorded in connection with acquisitions, and relate to the
Company’s biometrics businesses included in the Identity Solutions segment. The impairment
charges result from the deteriorating economic conditions that manifested themselves in the
fourth quarter of 2008, particularly as they impacted the biometrics businesses, as well as capital
market conditions that adversely impacted valuation of businesses the Company acquired and the
Company’s stock price and market capitalization.
The Company utilized a valuation advisor to assist in performing the impairment analyses and
valuations. Estimates of fair values were primarily based on the discounted cash flows based on
the Company’s latest plans and projections. The use of the discounted cash flow method requires
significant judgments and assumptions of future events many of which are outside the control of
the Company, including estimates of future growth rates, income tax rates, and discount rates,
among others.
REAL ASSET LIQUIDITY AND ASSET IMPAIRMENTS 53
Appendix B, continued
Firms using observable market data to estimate impairment:
12/31/2010 10-K report for Retractable Technologies, Inc. (CIK 0000946563):
During 2010, the Company recognized impairment charges of $365,295 on equipment designed
in connection with research and development activities. The Company will likely outsource the
majority of this production through overseas manufacturers. Minimal cash flows, if any, are
expected to be generated by this equipment. Accordingly, the Company reduced the carrying
value of this equipment to an estimated fair value of zero.
The Company’s management estimated the fair value of the equipment based on guidance
established by the Fair Value Measurements and Disclosures Topic of the FASB ASC. In this
instance, the Company’s management determined the impairment charge by utilizing observable
market data, a Level 2 input under the FASB ASC. A Level 1 input would require quoted prices,
which were not available in this matter.
2/28/2007 10-K report for Steelcase, Inc. (CIK 0001050825):
During Q4 2007, we performed our annual impairment assessment of goodwill in our reporting
units. In the first step to test for potential impairment, we measured the estimated fair value of
our reporting units using a combination of two methods based upon a discounted cash flow
valuation (“DCF”) and a market value approach (“MVA”). The first method used a 100%
weighting factor based on DCF while the second valuation was based upon 50% of DCF and
50% of MVA. The MVA was only calculated for International and North America excluding
consolidated dealers because these are the two reporting units where we could obtain comparable
market data.
The DCF analysis was based on the present value of projected cash flows and a residual value
and used the following assumptions:
a business is worth today what it can generate in future cash to its owners,
cash received today is worth more than an equal amount of cash received in the future, and
future cash flows can be reasonably estimated.
The MVA used a set of four comparable companies to derive a range of market multiples for the
last twelve months’ revenue and earnings before interest, taxes, depreciation and amortization.
REAL ASSET LIQUIDITY AND ASSET IMPAIRMENTS 54
Appendix C
Compustat Fixed Asset Write-Offs: Random Sample Analysis
Table C1
Panel A: Categorization of assets underlying fixed asset write-offs based on 10-K discussion
Asset category FWO Frequency
FWO Frequency - Single asset
category written off
PPE 101 63
Investment 54 39
Intangible 70 42
Inventory 15 7
Goodwill 3 2
A/R 2 0
Other 2 2
Not found 9 9
Total 209 164
Panel B: Categorization of reasons for fixed asset write-offs based on 10-K discussion
Disclosed reason for write-off Frequency
Discontinued Operations 16
Restructuring 29
Asset Sale 12
Fraud 1
Other 142
Total 200
Note: The above panels provide results of the subjective coding of 200 randomly selected fixed asset write-offs from
Compustat from the final sample of 33,641 firm-years from Dec 2000 – Dec 2012 outlined in Table 1 with
underlying data available in 10-K reports available in SEC’s EDGAR database. Note that 9 write-offs selected were
unable to be matched to underlying 10-K reports, and are included in Panel A as “Not Found.” Compustat separately
tracks fixed asset write-offs included in special items in earnings separately from remaining special items beginning
in 2000.
REAL ASSET LIQUIDITY AND ASSET IMPAIRMENTS 55
Page 55
Table 1
Sample Selection Criteria
Sample criteria N
Compustat sample firms with valid 3-digit SIC codes from December 31, 2000 - December 31, 2012 110,528
Exclude regulated utilities (SIC 4900-4999), financial firms (SIC 6000-6999 and SIC 872), and ADR/foreign
firms -26,225
Exclude 3-digit SIC industry-years with fewer than 5 observations with data necessary to calculate industry
measures of real asset liquidity, competition, and high-tech industries (requires book value of total assets
and total sales) -10,938
Require data available to calculate balance sheet liquidity (cash holdings) measures at time t
-3,470
Require positive book value of equity, total assets in excess of $1 million, CRSP share price 3 months after year-
end > $1, and CRSP data on shares outstanding 3-months after year-end
-29,951
Require data available to calculate key dependent variables: asset write-offs, book value per share, earnings per
share, and B/M
t
-351
Require data available to calculate key control variables: return on assets, B/M
t-1
, market capitalization, standard
deviation of equity returns, leverage, Z-score, revenue growth, book value of goodwill
t-1
, equity returns
over year t, and trading volume for the 6 months around year-end
-3,452
Final firm-year observations remaining 36,141
Subsample of firm-years with I/B/E/S earnings forecast data available (annual consensus earnings surprise and
forecast dispersion for firms with at least 3 analysts)
21,745
REAL ASSET LIQUIDITY AND ASSET IMPAIRMENTS 56
Page 56
Table 2
Descriptive Statistics
Panel A: Descriptive statistics for full sample (N = 36,141 firm-years covering 5,603 unique firms in
winsorized sample)
Variable Mean St. deviation Min Q1 Median Q3 Max N
RAL_Factor -0.001 1.001 -1.655 -0.810 -0.216 0.650 2.520 36,141
AT_SaleSW 0.084 0.102 0.000 0.009 0.034 0.131 0.530 36,141
DOC_CountSW 2.464 1.108 0.000 1.674 2.398 3.178 4.913 36,141
MA_CountSW 1.728 1.253 0.000 0.693 1.386 2.631 4.641 36,141
WO_dummy 0.214 0.410 0 0 0 0 1 36,141
WO -0.007 0.027 -0.186 0.000 0.000 0.000 0.000 36,141
FWO_dummy 0.165 0.371 0 0 0 0 1 36,141
FWO -0.002 0.008 -0.057 0.000 0.000 0.000 0.000 36,141
GWO_dummy 0.082 0.274 0 0 0 0 1 36,141
GWO -0.004 0.020 -0.143 0.000 0.000 0.000 0.000 36,141
BM 0.691 0.324 0.084 0.446 0.668 0.898 1.665 36,141
NOAS 0.886 1.418 -0.899 0.295 0.533 0.938 12.037 36,141
PRC 21.005 39.786 1.00 5.10 13.22 28.16 2,799.99 36,141
BVPS 9.130 8.774 0.150 2.880 6.568 12.337 44.390 36,141
EPS 0.610 1.836 -5.399 -0.237 0.422 1.411 6.789 36,141
WOPS -0.094 0.384 -2.716 0.000 0.000 0.000 0.000 36,141
RET(t) 0.178 0.723 -0.842 -0.253 0.049 0.400 3.931 36,141
dAssym -0.004 0.313 -1.010 -0.069 -0.001 0.053 1.042 21,745
dDisp -0.002 0.009 -0.046 -0.002 -0.001 0.000 0.033 21,745
PreROA -0.006 0.214 -1.816 -0.026 0.037 0.087 0.413 36,141
Cash 0.255 0.350 0.000 0.040 0.139 0.353 4.111 36,141
BM(t-1) 0.676 0.335 0.076 0.422 0.651 0.891 1.676 36,141
log_MCAP 6.444 2.001 0.052 4.994 6.368 7.747 13.436 36,141
sd_RET 0.149 0.093 0.037 0.087 0.124 0.183 0.592 36,141
HiTech 0.423 0.494 0 0 0 1 1 36,141
Lev 0.181 0.182 0.000 0.004 0.141 0.300 0.719 36,141
Comp1 11.181 1.345 5.645 10.124 11.293 12.380 14.280 36,141
Comp2 1.133 0.099 0.522 1.068 1.116 1.193 1.878 36,141
Comp3 8.317 1.481 2.223 7.145 8.600 9.405 12.149 36,141
HHI 0.157 0.124 0.037 0.078 0.102 0.196 0.970 36,141
Z_Score 2.036 2.330 -7.861 1.165 2.131 3.177 10.099 36,141
D 0.456 0.498 0 0 0 1 1 36,141
Rev_Growth 0.180 0.621 -0.711 -0.027 0.080 0.221 5.864 36,141
d_PreIB 0.009 0.142 -0.758 -0.028 0.007 0.039 0.692 36,141
EMI_up 0.036 0.100 0.000 0.000 0.000 0.000 0.692 36,141
EMI_down -0.028 0.089 -0.758 0.000 0.000 0.000 0.000 36,141
CR_Dum 0.269 0.443 0 0 0 1 1 36,141
lag_GDWL 0.099 0.138 0.000 0.000 0.028 0.158 0.560 36,141
log_VOL 12.251 2.073 4.754 10.881 12.427 13.668 18.566 36,141
Numest 3.105 5.052 0 0 1 4 52 36,141
This table provides descriptive statistics for firm-year observations between Dec 2000 and Dec 2012 with
underlying data available. See Table 1 for details of sample selection procedures. Definitions for all variables
used in analysis are included in Appendix A.
REAL ASSET LIQUIDITY AND ASSET IMPAIRMENTS 57
Page 57
Table 2, continued
Panel B: Correlation matrix
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13)
WO_Dummy (1) 1.00
WO (2) -0.48 1.00
(0.00)
RAL_Factor (3) -0.01 -0.04 1.00
(0.28) (0.00)
Cashr (4) -0.06 0.05 0.43 1.00
(0.00) (0.00) (0.00)
log_MCAP (5) 0.07 0.08 -0.13 -0.17 1.00
(0.00) (0.00) (0.00) (0.00)
BM(t-1) (6) 0.11 -0.19 -0.25 -0.37 -0.22 1.00
(0.00) (0.00) (0.00) (0.00) (0.00)
sd_RET (7) 0.05 -0.16 0.21 0.17 -0.35 0.03 1.00
(0.00) (0.00) (0.00) (0.00) (0.00) (0.00)
Lev (8) 0.07 -0.03 -0.28 -0.49 0.25 0.20 -0.03 1.00
(0.00) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00)
PreROA (9) -0.05 0.12 -0.20 -0.20 0.28 0.05 -0.35 0.03 1.00
(0.00) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00)
HiTech (10) 0.02 -0.04 0.66 0.46 -0.15 -0.22 0.20 -0.27 -0.22 1.00
(0.00) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00)
HHI (11) 0.02 0.01 -0.47 -0.20 0.07 0.12 -0.10 0.10 0.11 -0.28 1.00
(0.00) (0.11) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00)
RET (t) (12) -0.07 0.07 -0.04 0.00 0.02 0.25 0.12 0.00 0.09 -0.05 0.01 1.00
(0.00) (0.00) (0.00) (0.94) (0.00) (0.00) (0.00) (0.69) (0.00) (0.00) (0.01)
d_Disp_BKLS (13) 0.00 0.00 -0.02 -0.01 -0.02 0.00 0.00 0.00 0.00 -0.02 0.00 -0.02 1.00
(0.81) (0.75) (0.00) (0.12) (0.02) (0.60) (0.51) (0.49) (0.91) (0.01) (0.84) (0.01)
Note: P-values are included in parentheses for the significance of two-way correlation coefficients. See Appendix A for variable definitions.
REAL ASSET LIQUIDITY AND ASSET IMPAIRMENTS 58
Page 58
Table 2, continued
Panel C: Most and least liquid industries for each measure of real asset liquidity across the full sample period
Top: AT_SaleSW
DOC_CountSW
MA_CountSW
Rank SIC Description
SIC Description
SIC Description
1 352 Farm and Garden Machinery
737
Computer and Data Processing
Services
737
Computer and Data Processing
Services
2 371 Motor Vehicles and Equipment
283 Drugs
283 Drugs
3 737
Computer and Data Processing
Services
366 Communications Equipment
384 Medical Instruments and Supplies
4 357 Computer and Office Equipment
131 Crude Petroleum and Natural Gas
481 Telephone Communications
5 760 Miscellaneous Repair Services
357 Computer and Office Equipment
131 Crude Petroleum and Natural Gas
6 751 Automotive Rentals, No Drivers
384 Medical Instruments and Supplies
367
Electronic Components and
Accessories
7 801 Offices and Clinics of Medical Doctors
382 Measuring and Controlling Devices
366 Communications Equipment
8 367
Electronic Components and
Accessories
367
Electronic Components and
Accessories
357 Computer and Office Equipment
9 573 Radio, Television, and Computer Stores
481 Telephone Communications
738 Miscellaneous Business Services
10 820 Educational Services
581 Eating and Drinking Places
382
Measuring and Controlling
Devices
Bottom: AT_SaleSW
DOC_CountSW
MA_CountSW
Rank SIC Description
SIC Description
SIC Description
191 539 Misc. General Merchandise Stores
540 Food Stores
020 Agricultural Production - Livestock
190 391 Jewelry, Silverware, and Plated Ware
539
Miscellaneous General
Merchandise Stores
278 Blankbooks and Bookbinding
189 540 Food Stores
482
Telegraph and Other
Communications
540 Food Stores
188 461 Pipelines, Except Natural Gas
440 Water Transportation
515 Farm-Product Raw Materials
187 784 Video Tape Rental
229 Miscellaneous Textile Goods
207 Fats and Oils
186 229 Miscellaneous Textile Goods
513 Apparel, Piece Goods, and Notions
410
Local and Interurban Passenger
Transit
185 222 Broadwoven Fabric Mills, Manmade
784 Video Tape Rental
539 Misc. General Merchandise Stores
184 285 Paints and Allied Products
206 Sugar and Confectionery Products
299 Misc. Petroleum and Coal Products
183 070 Agricultural Services
261 Pulp Mills
339 Misc. Primary Metal Products
182 422 Public Warehousing and Storage 302 Rubber and Plastics Footwear 249 Miscellaneous Wood Products
Industries in bold are ranked in the top/bottom 10 industries across two or more measures of real asset liquidity for the sample period covering Dec 2000 – Dec
2012. See Appendix A for descriptions of real asset liquidity measures.
REAL ASSET LIQUIDITY AND ASSET IMPAIRMENTS 59
Page 59
Table 3
Likelihood of Impairment
Panel A: Impairment likelihood and real asset liquidity (Logit model)
Dep variable WO_Dum (t) WO_Dum (t) WO_Dum (t)
(1) (2) (3)
RAL_Factor
Constant -3.577*** -3.887*** -4.045***
(-30.876) (-9.247) (-8.013)
Liquidity Measure 0.050** 0.130*** 0.125***
(2.296) (2.901) (2.588)
BM(t-1) 0.802*** 1.123*** 1.071***
(12.238) (21.276) (13.774)
log_MCAP
0.214*** 0.225***
(12.579) (11.402)
Cash Rank
-0.094 -0.174
(-0.698) (-1.131)
sd_RET
2.914*** 2.705***
(12.338) (9.707)
HiTech
0.227** 0.156
(2.320) (1.539)
Comp1
-0.109** -0.085
(-2.226) (-1.593)
Comp2
-0.873*** -1.081***
(-3.276) (-3.459)
Comp3
0.015 0.012
(0.328) (0.257)
HHI
0.173 0.444
(0.514) (1.276)
Leverage
0.283** 0.180
(2.326) (0.966)
Rev_Growth
-0.214*** -0.158***
(-3.071) (-3.364)
Δpre_IB
-5.298*** -4.555***
(-7.282) (-4.517)
RET(t)
-0.494*** -0.513***
(-17.115) (-8.903)
RET(t-1)
-0.284*** -0.280***
(-9.304) (-6.538)
EMI_Down
3.832*** 3.310***
(5.113) (3.764)
EMI_Up
5.440*** 4.590***
(7.060) (4.488)
CR_Dummy
0.046 -0.082
(0.760) (-0.951)
Sample Full Full Sales removed
Year dummies? Yes Yes Yes
ROC Area 0.634 0.701 0.703
Observations 36,141 36,141 14,412
This table provides Logit model tests of impairment frequency for firm-year observations from Dec
1995 – Dec 2012. Robust t-statistics in parentheses and corresponding p-values are calculated using 1-
way clustered standard errors by 3-digit SIC industry. See Appendix A for variable definitions.
*** p<0.01, ** p<0.05, * p<0.1
REAL ASSET LIQUIDITY AND ASSET IMPAIRMENTS 60
Page 60
Table 3, continued
Panel B: Impairment likelihood and real asset liquidity (Logit model)
Dep variable FWO_Dum FWO_Dum FWO_Dum FWO_Dum FWO_Dum FWO_Dum GWO_Dum GWO_Dum GWO_Dum
(1) (2) (3) (1) (2) (3) (1) (2) (3)
AT_SaleSW DOC_CountSW MA_CountSW
Constant -4.090*** -4.789*** -5.050*** -4.218*** -4.641*** -5.037*** -4.626*** -5.900*** -6.011***
(-28.413) (-11.162) (-10.130) (-30.099) (-11.830) (-9.813) (-21.272) (-5.919) (-6.068)
Liquidity Measure 0.488* 0.288 0.684* 0.059** 0.084* 0.080* -0.029 -0.004 -0.018
(1.792) (1.179) (1.900) (2.492) (1.758) (1.679) (-0.971) (-0.083) (-0.379)
BM(t-1) 0.497*** 0.799*** 0.747*** 0.518*** 0.811*** 0.756*** 1.875*** 1.892*** 1.876***
(8.241) (11.997) (7.963) (8.388) (12.508) (8.045) (17.803) (17.520) (16.385)
log_MCAP
0.188*** 0.217*** 0.190*** 0.218*** 0.124*** 0.119***
(9.760) (10.812) (9.747) (10.611) (5.156) (4.849)
Cash Rank
0.042 -0.086 0.048 -0.071 -0.091 -0.095
(0.322) (-0.539) (0.352) (-0.435) (-0.584) (-0.598)
sd_RET
2.135*** 2.242*** 2.102*** 2.205*** 4.560*** 4.454***
(9.345) (8.613) (9.044) (8.376) (12.320) (11.765)
HiTech
0.313*** 0.234** 0.278*** 0.228** 0.066 0.091
(3.740) (2.550) (2.704) (2.227) (0.714) (0.964)
Comp1
-0.116** -0.088* -0.142** -0.095 -0.030 -0.023
(-2.397) (-1.656) (-2.547) (-1.563) (-0.555) (-0.424)
Comp2
-0.531* -0.721** -0.626** -0.814*** -0.354 -0.269
(-1.784) (-2.297) (-2.338) (-2.644) (-0.552) (-0.430)
Comp3
0.075* 0.053 0.077 0.049 -0.020 -0.024
(1.650) (1.168) (1.638) (1.057) (-0.467) (-0.551)
HHI
-0.323 -0.073 -0.149 0.130 0.424 0.479
(-1.037) (-0.209) (-0.514) (0.388) (1.115) (1.249)
Leverage
0.298** 0.209 0.316*** 0.213 -0.015 0.021
(2.367) (1.078) (2.609) (1.108) (-0.075) (0.097)
Rev_Growth
-0.200*** -0.189*** -0.201*** -0.192*** -0.506*** -0.500***
(-2.920) (-3.363) (-2.918) (-3.396) (-3.420) (-3.262)
Δpre_IB
-3.914*** -3.661*** -3.923*** -3.646*** -7.583*** -7.556***
(-4.957) (-3.016) (-4.893) (-2.976) (-6.782) (-6.490)
RET(t)
-0.361*** -0.379*** -0.362*** -0.378*** -0.832*** -0.831***
(-12.056) (-6.754) (-12.190) (-6.745) (-12.421) (-12.147)
REAL ASSET LIQUIDITY AND ASSET IMPAIRMENTS 61
Page 61
RET(t-1)
-0.208*** -0.208*** -0.206*** -0.206*** -0.512*** -0.518***
(-6.632) (-4.724) (-6.632) (-4.686) (-7.905) (-7.758)
EMI_Down
2.877*** 2.754** 2.902*** 2.752** 3.844*** 3.755***
(3.580) (2.413) (3.591) (2.395) (2.899) (2.750)
EMI_Up
4.073*** 3.808*** 4.052*** 3.776*** 7.818*** 7.910***
(5.022) (3.084) (4.960) (3.033) (6.531) (6.411)
CR_Dummy
0.003 -0.121 0.004 -0.116 0.149* 0.162**
(0.044) (-1.532) (0.053) (-1.480) (1.916) (2.089)
Sample Full Full
Sales
removed Full Full
Sales
removed GW
t-1
GW
t-1
GW
t-1
+ M&A
removed
Year dummies? Yes Yes Yes Yes Yes Yes Yes Yes Yes
ROC Area 0.619 0.678 0.684 0.619 0.678 0.684 0.696 0.766 0.766
Observations 36,141 36,141 14,412 36,141 36,141 14,412 21,858 21,858 20,911
This table provides Logit model tests of impairment frequency for firm-year observations from Dec 1995 – Dec 2012. Robust t-statistics in parentheses and
corresponding p-values are calculated using 1-way clustered standard errors by 3-digit SIC industry. See Appendix A for variable definitions.
*** p<0.01, ** p<0.05, * p<0.1
REAL ASSET LIQUIDITY AND ASSET IMPAIRMENTS 62
Page 62
Table 4
Airline Industry Tests
Panel A: Airline sample descriptives
Variable Mean St. deviation Min Q1 Median Q3 Max N
PlaneTrans 0.079 0.045 0.042 0.057 0.067 0.084 0.316 69
PlaneTrans (Owned) 0.199 0.070 0.094 0.152 0.184 0.232 0.407 67
Imp_Amount 0.015 0.038 0.000 0.000 0.000 0.002 0.209 69
Imp_Dummy 0.261 0.442 0 0 0 1 1 69
B/M (t) 0.755 0.235 0.287 0.584 0.798 0.952 1.292 69
B/M (t-1) 0.721 0.221 0.226 0.555 0.719 0.880 1.292 69
PPESALE 0.572 0.317 0.173 0.359 0.512 0.686 1.781 69
log_LDGS 9.840 1.112 4.138 9.691 9.999 10.431 11.085 69
log_MTOW 16.750 1.702 11.809 15.513 17.199 18.097 18.795 69
Pct_Own 0.611 0.236 0.000 0.492 0.653 0.739 1.000 69
Ind_2001 0.116 0.323 0.000 0.000 0.000 0.000 1.000 69
PreROA 0.044 0.085 -0.232 0.005 0.054 0.079 0.234 69
RET (t) 0.129 0.734 -0.929 -0.257 0.033 0.311 4.067 69
sd_RET (t) 0.147 0.063 0.043 0.097 0.136 0.179 0.331 69
Rev_Growth 0.090 0.168 -0.403 0.021 0.088 0.136 1.078 69
Firm-year observations underlying airline tests must meet the following sample selection criteria: firms must operate
in one of three SIC industries (4512 - scheduled air transportation, 4513 - air couriers, or 4522 - non-scheduled air
transportation), fiscal years must end during 1995-2002, firms must have non-missing 10-K reports on SEC's
EDGAR database, firms must operate aircraft tracked in the ACAS database, and firms must have positive book
value of equity. A subsample of firm-years with available data on returns and market capitalization on the CRSP
database is used for valuation tests in Table 6, Panel C and Table 7, Panel D below. Aircraft tagged as unrepairable
or dismantled/retired are removed from calculations of airline fleet liquidity, number of landings, and maximum
take-off weight. Computation of financial statement variables is described in Appendix A. Airline variables
calculated from ACAS data are as follows:
Imp_Amount = Aircraft impairment amounts identified via hand-collecting data from 10-K reports available via
SEC’s EDGAR site for airline firms;
Imp_Dummy = Indicator variable set to 1 where aircraft impairments are identified via hand-collecting data from 10-
K reports available via SEC’s EDGAR site for airline firms, and equal to 0 otherwise;
Ind_2001 = Indicator variable set to 1 for 2001 fiscal year observations to capture the one-time effect of September
11
th
on write-downs for airlines, and equal to 0 otherwise;
log_LDGS = Natural log of the number of total landings made by all aircraft operated in an airline’s fleet as of fiscal
year-end. Number of landings for each aircraft is weighted by the maximum towing capacity for that
aircraft to derive a weighted average total number of landings for the airline fleet;
log_MTOW = Natural log of the maximum towing capacity for the entire fleet of planes operated by each airline as
of fiscal year-end. This measures size of the aircraft fleet;
PlaneTrans (Owned) = Ratio of the number of times each make and model of aircraft changes operators on the
secondary market for used aircraft during the 12-month period ending at the fiscal year-end scaled by the
average number of aircraft in operation during the year as Transaction Count
t-1,t
/ (Plane count
t-1
+ Plane
count
t
)/2 matched to full aircraft fleet information for each airline-year. For the “Owned” version of this
variable, resale transaction counts are scaled by the average number of aircraft that are owned (not leased)
to capture resale transactions relative to the stock of aircraft owned and operated by other carriers. This
variable is then matched to fleet information on owned (not leased) aircraft for each airline-year. Airline-
years with no owned aircraft are not defined for the “Owned” version of this variable;
Pct_Own = Percentage of aircraft fleet owned (not leased) by the airline firm.
REAL ASSET LIQUIDITY AND ASSET IMPAIRMENTS 63
Page 63
Table 4, continued
Panel B: Aircraft impairment likelihood for 1995-2002 SEC EDGAR observations for airline industry firms (Logit model)
Dep variable Imp_Dummy Imp_Dummy Imp_Dummy Imp_Dummy Imp_Dummy Imp_Dummy
(1) (2) (3) (4) (5) (6)
Constant -1.647 -30.580 -94.951** -0.641 -18.092*** -41.872*
(-1.293) (-1.559) (-2.202) (-1.049) (-2.856) (-1.893)
PlaneTrans 3.257 9.120 15.227*
(1.024) (1.348) (1.898)
Pct_Own 0.557 -2.556 -4.646
(0.397) (-1.251) (-0.990)
PlaneTrans (Owned)
-1.829 10.932** 14.472**
(-0.813) (2.021) (2.016)
B/M (t-1)
8.102*** 9.558***
8.705*** 9.331***
(4.967) (2.610)
(3.699) (3.606)
log_MTOW
0.469 2.247**
0.159 1.062*
(1.142) (2.324)
(0.671) (1.787)
log_LDGS
1.568 4.551
0.495 0.945
(1.021) (1.630)
(1.165) (0.727)
Ind_2001
3.031*** 2.784*
3.948** 3.627
(2.766) (1.765)
(2.274) (1.550)
PreROA
-2.625
-1.341
(-0.595)
(-0.446)
RET (t)
-3.035***
-2.771***
(-3.291)
(-4.308)
sd_RET (t)
21.443***
17.781***
(3.497)
(2.879)
ROC Area 0.513 0.875 0.973 0.539 0.883 0.964
Observations 69 69 69 67 67 67
This table provides results for tests of aircraft impairment frequency using hand-collected data on aircraft impairments from the SEC EDGAR
database from Dec 1995 – Dec 2002 for firms operating in the airline industry based on 4-digit SIC code (4512 - scheduled air transportation,
4513 - air couriers, or 4522 - non-scheduled air transportation). Robust t-statistics in parentheses and corresponding p-values are calculated
using robust standard errors clustered by firm. See the notes to Panel A of this table for definitions of airline fleet liquidity and airline-specific
control variables and see Appendix A for remaining variable definitions.
*** p<0.01, ** p<0.05, * p<0.1
REAL ASSET LIQUIDITY AND ASSET IMPAIRMENTS 64
Page 64
Table 5
Magnitude of Impairment
Dep variable WO WO FWO FWO FWO FWO GWO GWO
(1) (2) (1) (2) (1) (2) (1) (2)
RAL_Factor AT_SaleSW DOC_CountSW MA_CountSW
Constant 0.055*** 0.053*** -0.007 -0.008* -0.010** -0.013*** 0.090*** 0.080***
(3.007) (3.079) (-1.607) (-1.675) (-2.399) (-2.722) (4.277) (3.732)
Liquidity Measure -0.004*** 0.005*** 0.006*** 0.016*** -0.001** 0.000 -0.003*** 0.001
(-3.728) (2.772) (8.924) (5.199) (-2.490) (0.608) (-2.660) (0.658)
BM(t-1) -0.056*** -0.055*** -0.015*** -0.014*** -0.015*** -0.011*** -0.054*** -0.043***
(-10.915) (-13.405) (-23.819) (-16.534) (-24.577) (-6.651) (-12.596) (-8.063)
LIQ*BM
-0.011*** -0.013*** -0.002*** -0.005***
(-4.598) (-3.307) (-3.347) (-3.631)
log_MCAP 0.001* 0.001* 0.002*** 0.002*** 0.002*** 0.002*** 0.001* 0.001*
(1.922) (1.888) (9.485) (9.447) (9.490) (9.273) (1.776) (1.748)
Cash Rank 0.016*** 0.015*** 0.001 0.001 0.001 0.001 0.018*** 0.018***
(4.591) (4.532) (1.297) (1.245) (1.608) (1.493) (8.121) (8.118)
sd_RET -0.103*** -0.105*** -0.020*** -0.020*** -0.020*** -0.020*** -0.086*** -0.088***
(-12.413) (-13.807) (-11.329) (-11.276) (-12.447) (-11.951) (-7.743) (-7.829)
HiTech -0.004** -0.004** -0.002*** -0.002*** -0.001 -0.001 -0.007** -0.007**
(-2.097) (-2.177) (-3.851) (-3.802) (-1.602) (-1.621) (-2.428) (-2.452)
Comp1 -0.002 -0.002 0.001*** 0.001*** 0.002*** 0.002*** -0.004* -0.004**
(-0.939) (-1.015) (2.849) (2.848) (5.222) (5.250) (-1.935) (-2.005)
Comp2 -0.036*** -0.035*** -0.002 -0.002 -0.002 -0.002 -0.064*** -0.062***
(-3.408) (-3.517) (-0.745) (-0.700) (-0.819) (-0.793) (-4.804) (-4.650)
Comp3 0.002* 0.002* -0.001*** -0.001*** -0.002*** -0.002*** 0.005** 0.005**
(1.734) (1.841) (-6.559) (-6.609) (-8.602) (-8.896) (2.478) (2.541)
HHI -0.008 -0.009 0.005** 0.005** 0.004 0.004 -0.011 -0.013
(-1.327) (-1.524) (2.328) (2.331) (1.452) (1.459) (-1.257) (-1.424)
Leverage 0.005 0.004 0.002 0.002 0.001 0.001 -0.003 -0.003
(1.156) (0.947) (1.433) (1.398) (1.190) (1.050) (-0.380) (-0.453)
Rev_Growth -0.002 -0.002 0.000 0.000 0.000 0.000 -0.008*** -0.008***
(-0.985) (-1.122) (0.574) (0.566) (0.470) (0.426) (-2.614) (-2.684)
Δpre_IB 0.103*** 0.109*** 0.022* 0.023* 0.023* 0.023* 0.181*** 0.187***
(6.450) (7.990) (1.884) (1.894) (1.925) (1.936) (8.061) (9.062)
REAL ASSET LIQUIDITY AND ASSET IMPAIRMENTS 65
Page 65
RET(t) 0.013*** 0.013*** 0.002*** 0.002*** 0.002*** 0.002*** 0.012*** 0.012***
(5.358) (5.329) (2.990) (3.021) (3.073) (3.154) (4.102) (4.004)
RET(t-1) 0.008*** 0.008*** 0.002*** 0.002*** 0.002*** 0.002*** 0.010*** 0.010***
(7.357) (7.071) (8.834) (8.419) (8.263) (8.006) (5.450) (5.319)
EMI_Down -0.016 -0.019 -0.001 -0.001 -0.001 -0.001 -0.088*** -0.091***
(-0.887) (-1.137) (-0.039) (-0.057) (-0.060) (-0.080) (-3.438) (-3.619)
EMI_Up -0.123*** -0.127*** -0.027** -0.027** -0.027** -0.027** -0.200*** -0.205***
(-6.994) (-8.238) (-2.210) (-2.208) (-2.186) (-2.190) (-7.922) (-8.871)
CR_Dummy 0.001 0.001 -0.000 -0.000 -0.000 -0.000 0.002 0.003
(0.495) (0.611) (-0.527) (-0.492) (-0.440) (-0.363) (1.231) (1.273)
Sample WO req'd WO req'd FWO req'd FWO req'd FWO req'd FWO req'd GWO req'd GWO req'd
Year dummies? Yes Yes Yes Yes Yes Yes Yes Yes
Adj R-squared 0.286 0.292 0.247 0.247 0.247 0.248 0.315 0.316
Observations 7,735 7,735 5,970 5,970 5,970 5,970 2,948 2,948
This table provides results of OLS regressions of differences in predicted impairment magnitude conditional on the firm recording a
fixed asset or goodwill impairment during the year. Robust t-statistics calculated using 2-way clustered standard errors by 3-digit SIC
industry and fiscal year in parentheses. See Appendix A for variable definitions.
*** p<0.01, ** p<0.05, * p<0.1
REAL ASSET LIQUIDITY AND ASSET IMPAIRMENTS 66
Page 66
Table 6
Conditional Conservatism of Write-Offs
Dep variable WO (t) FWO (t) FWO (t) GWO (t)
(1) (2) (3) (4)
RAL_Factor DOC ATS MAC
Intercept 0.004*** -0.000 -0.000 0.002***
(3.645) (-0.393) (-0.259) (4.143)
D 0.001 -0.001 0.000 0.006*
(0.605) (-0.594) (0.193) (1.895)
Ret -0.001 0.000 -0.000 0.002
(-0.883) (0.123) (-0.166) (0.978)
LIQDum*Ret -0.001 -0.000 -0.000 -0.002*
(-0.797) (-0.085) (-0.669) (-1.766)
B/M*Ret 0.001 -0.000 -0.000 -0.001
(0.646) (-0.496) (-0.635) (-0.442)
Size*Ret 0.000* 0.000 0.000 -0.000
(1.939) (1.022) (1.119) (-0.743)
Lev*Ret -0.005*** -0.001 -0.001 -0.005**
(-3.516) (-1.038) (-1.626) (-2.134)
Ret*D -0.017*** -0.003 -0.001 -0.005
(-3.803) (-1.132) (-0.327) (-0.801)
LIQDum*Ret*D 0.010 0.002 0.001 0.009**
(1.594) (1.626) (0.933) (2.543)
B/M*Ret*D 0.058*** 0.011*** 0.008*** 0.065***
(3.307) (4.386) (4.708) (3.455)
Size*Ret*D 0.002* 0.000 0.000 -0.000
(1.900) (0.944) (0.554) (-0.246)
Lev*Ret*D -0.002 -0.003 -0.004 -0.008
(-0.181) (-0.669) (-1.222) (-0.458)
Sample Full Full Full GW
Main effects? Yes Yes Yes Yes
Adjusted R-squared 0.099 0.039 0.039 0.126
Observations 36,141 36,141 36,141 21,858
This table provides results of conditional conservatism tests using a Basu (1997) regression model with
impairments as the dependent variable using firm-year observations from Dec 2000 – Dec 2012. Controls for
the main effects of non-discretionary determinants of conservatism advanced by Khan and Watts (2007) are
included along with interactions. The regression model used is:
WO = α0 + α1*D + α2*LIQDum + γ3*D*LIQDum + γ4*ARET + γ5*ARET*D + γ6*ARET*LIQDum +
γ7*ARET*D*LIQDum + Σλ*CScore_Controls + ε
Robust t-statistics in parentheses and corresponding p-values are calculated using 2-way clustered standard
errors by firm and fiscal year. See Appendix A for variable definitions.
*** p<0.01, ** p<0.05, * p<0.1
REAL ASSET LIQUIDITY AND ASSET IMPAIRMENTS 67
Page 67
Table 7
Asset Valuation Tests
Panel A: Multivariate analysis of B/M ratios for full sample
Dep variable B/M (t) B/M (t) B/M (t) B/M (t) B/M (t) B/M (t) B/M (t) B/M (t)
(1) (2) (1) (2) (1) (2) (1) (2)
RAL_Factor AT_SaleSW DOC_CountSW MA_CountSW
Constant 1.316*** 1.677*** 1.333*** 1.806*** 1.381*** 1.732*** 1.371*** 1.676***
(43.918) (8.726) (42.555) (10.736) (45.778) (10.221) (49.372) (9.340)
LIQ Measure -0.028*** -0.027** -0.084 -0.026 -0.027*** -0.028*** -0.028*** -0.029***
(-3.397) (-2.289) (-1.361) (-0.401) (-3.778) (-2.951) (-3.572) (-2.994)
log_MCAP -0.065*** -0.072*** -0.065*** -0.071*** -0.065*** -0.072*** -0.065*** -0.072***
(-11.322) (-11.532) (-11.303) (-11.547) (-11.250) (-11.463) (-11.239) (-11.545)
Cash Rank -0.374*** -0.273*** -0.405*** -0.278*** -0.375*** -0.276*** -0.366*** -0.273***
(-11.183) (-13.218) (-9.631) (-12.914) (-11.913) (-13.344) (-12.501) (-13.606)
sd_RET
0.134 0.125 0.139 0.141
(1.245) (1.172) (1.299) (1.327)
HiTech
-0.018 -0.035 -0.019 -0.014
(-0.745) (-1.505) (-0.836) (-0.603)
Comp1
0.008 -0.003 0.008 0.010
(0.733) (-0.256) (0.859) (1.028)
Comp2
-0.464*** -0.488*** -0.459*** -0.429***
(-3.608) (-4.041) (-3.656) (-3.325)
Comp3
0.010 0.012 0.011 0.010
(1.310) (1.396) (1.390) (1.366)
HHI_Sale
-0.027 0.000 -0.055 -0.048
(-0.609) (0.004) (-1.279) (-1.085)
Leverage
0.105*** 0.112*** 0.105*** 0.104***
(3.612) (3.749) (3.594) (3.619)
Rev_Growth
-0.025*** -0.025*** -0.025*** -0.025***
(-3.016) (-2.944) (-3.012) (-2.998)
Δpre_IB
-0.926*** -0.925*** -0.926*** -0.926***
(-11.013) (-10.928) (-10.902) (-11.039)
RET(t)
-0.077*** -0.077*** -0.077*** -0.078***
(-5.358) (-5.372) (-5.328) (-5.408)
REAL ASSET LIQUIDITY AND ASSET IMPAIRMENTS 68
Page 68
RET(t-1)
-0.063*** -0.063*** -0.063*** -0.063***
(-5.718) (-5.714) (-5.737) (-5.826)
EMI_Down
1.138*** 1.140*** 1.136*** 1.133***
(10.956) (11.049) (10.775) (10.873)
EMI_Up
0.679*** 0.672*** 0.681*** 0.682***
(10.569) (10.250) (10.462) (10.564)
CR_Dummy
0.131*** 0.132*** 0.131*** 0.130***
(12.854) (12.750) (12.881) (12.911)
Year dummies? Yes Yes Yes Yes Yes Yes Yes Yes
Adj R-squared 0.347 0.396 0.342 0.394 0.348 0.397 0.350 0.398
Observations 36,141 36,141 36,141 36,141 36,141 36,141 36,141 36,141
This table provides multivariate results of book value ratio tests using firm-year observations from Dec 2000 – Dec 2012.
Robust t-statistics in parentheses and corresponding p-values are calculated using 2-way clustered standard errors by 3-digit SIC
industry and fiscal year. See Appendix A for variable definitions.
*** p<0.01, ** p<0.05, * p<0.1
REAL ASSET LIQUIDITY AND ASSET IMPAIRMENTS 69
Page 69
Table 7, continued
Panel B: Multivariate analysis of B/M ratios for 1995-2002 SEC EDGAR observations
for airline industry firms
Dep variable B/M (t) B/M (t) B/M (t) B/M (t)
(1) (2) (3) (4)
Constant 0.954*** 0.803*** 0.971*** 0.508*
(7.842) (2.836) (11.204) (1.924)
PlaneTrans -1.422*** -0.934***
(-3.300) (-3.000)
Pct_Own -0.141 0.235
(-0.795) (1.096)
PlaneTrans (Owned)
-1.120** -0.597**
(-2.647) (-2.025)
log_MTOW
0.047**
0.053***
(2.555)
(3.160)
log_LDGS
-0.116***
-0.071***
(-4.067)
(-4.726)
Rev_Growth
-0.173
-0.143
(-1.466)
(-1.120)
RET (t)
-0.042*
-0.045
(-1.769)
(-1.598)
sd_RET (t)
1.741***
1.597***
(5.159)
(4.705)
Year dummies No No No No
Adjusted R-squared 0.054 0.453 0.099 0.426
Observations 69 69 67 67
This table provides results for multivariate tests of book value ratios for firms with underlying
data on SEC EDGAR’s database operating in the airline industry based on 4-digit SIC code
(4512 - scheduled air transportation, 4513 - air couriers, or 4522 - non-scheduled air
transportation). Robust t-statistics in parentheses and corresponding p-values are calculated
using robust standard errors clustered by firm and fiscal year. See the notes to Panel A of
Table 4 for definitions of airline fleet liquidity and airline-specific control variables and see
Appendix A for remaining variable definitions.
*** p<0.01, ** p<0.05, * p<0.1
REAL ASSET LIQUIDITY AND ASSET IMPAIRMENTS 70
Page 70
Table 8
Value Relevance of Accounting Information Tests
Panel A: Firm price level value relevance regressions - Coefficient differences
Dep variable PRC (t+3 mos) PRC (t+3 mos) PRC (t+3 mos) PRC (t+3 mos) PRC (t+3 mos)
(1) (2) (4) (3) (5)
Full sample RAL_Factor ATS DOC MAC
Constant 2.365* 1.690 1.938 1.616 1.582
(1.928) (1.370) (1.491) (1.196) (1.186)
BVPS 1.306*** 1.343*** 1.386*** 1.329*** 1.345***
(8.984) (10.343) (9.194) (10.722) (10.437)
EPS 5.451*** 5.327*** 5.422*** 5.371*** 5.375***
(8.123) (8.400) (8.197) (8.742) (8.525)
LIQDum
4.309 -1.172 5.061 4.796
(1.311) (-0.623) (1.457) (1.410)
BVPS*LIQDum
0.155 0.643*** 0.114 0.155
(0.387) (2.669) (0.303) (0.385)
EPS*LIQDum
-3.361*** -0.817 -3.957*** -3.213***
(-2.607) (-0.995) (-3.059) (-2.633)
Year dummies? Yes Yes Yes Yes Yes
Adj R-squared 0.235 0.238 0.238 0.240 0.239
Observations 36,141 36,141 36,141 36,141 36,141
This table provides results of price-level value relevance regressions using firm-year observations from Dec
2000 – Dec 2012. Robust t-statistics in parentheses and corresponding p-values are calculated using 2-way
clustered standard errors by 3-digit SIC industry and fiscal year. See Appendix A for variable definitions.
*** p<0.01, ** p<0.05, * p<0.1
REAL ASSET LIQUIDITY AND ASSET IMPAIRMENTS 71
Page 71
Table 8, continued
Panel B: Firm price level value relevance regressions
Coefficients
Vuong Test (a) - (b)
Constant BVPS EPS R-squared Statistic P-value
Group 1: Illiquid industry firms using RAL_Factor (9,031 observations)
Full model -2.946 1.343 7.173 0.143
( -0.78 ) (4.47) (6.05)
Book value only -4.974 2.103
0.110 (a)
( -1.20 ) (5.12)
Earnings only 8.348
10.804 0.114 (b) -0.938 0.348
(5.67)
(5.53)
Group 2: Liquid industry firms using RAL_Factor (8,979 observations)
Full model 4.866 1.730 4.148 0.397
(4.69) (8.39) (5.01)
Book value only 0.866 2.206
0.342 (a)
(0.63) (8.92)
Earnings only 15.397
7.346 0.243 (b) 8.797 0.000
(12.66)
(7.40)
Group 3: Illiquid firms using AT_SaleSW (9,047 observations)
Full model 1.501 0.984 5.767 0.197
(0.68) (5.62) (6.92)
Book value only 0.023 1.571
0.148 (a)
(0.01) (6.30)
Earnings only 9.314
8.357 0.160 (b) -1.430 0.153
(9.68)
(6.39)
Group 4: Liquid firms using AT_SaleSW (8,916 observations)
Full model 3.691 1.461 5.203 0.404
(6.37) (7.30)
Book value only -1.214 2.063
0.326 (a)
( -0.97 ) (6.98)
Earnings only 13.502
8.303 0.292 (b) 3.023 0.003
(22.30)
(6.69)
Group 5: Illiquid industry firms using DOC_CountSW (8,996 observations)
Full model -3.904 1.352 7.440 0.144
( -0.94 ) (4.30) (6.02)
Book value only -6.100 2.138
0.110 (a)
( -1.33 ) (4.89)
Earnings only 7.826
11.124 0.115 (b) -1.283 0.199
(4.79)
(5.39)
Group 6: Liquid industry firms using Count_DOCSW (9,013 observations)
Full model 5.994 1.618 3.776 0.440
(4.73) (12.02) (4.59)
Book value only 2.340 2.028
0.389 (a)
(1.67) (10.67)
Earnings only 15.390
7.001 0.266 (b) 7.994 0.000
(13.40)
(7.92)
REAL ASSET LIQUIDITY AND ASSET IMPAIRMENTS 72
Page 72
Group 7: Illiquid firms using MA_CountSW (8,979 observations)
Full model -3.807 1.363 7.109 0.141
( -0.92 ) (4.34) (6.01)
Book value only -5.904 2.122
0.109 (a)
( -1.32 ) (4.91)
Earnings only 7.892
10.758 0.112 (b) -0.666 0.506
(4.94)
(5.33)
Group 8: Liquid firms using MA_CountSW (9,181 observations)
Full model 6.320 1.751 4.308 0.430
(4.68) (9.35) (5.42)
Book value only 1.610 2.252
0.373 (a)
(1.04) (9.26)
Earnings only 16.850
7.813 0.269 (b) 8.936 0.000
(16.72) (6.44)
This table provides explanatory power and results of Vuong tests for price-level value relevance regressions using
firm-year observations from Dec 2000 – Dec 2012. Fiscal year indicator variables are included in all regression
models. Robust t-statistics in parentheses and corresponding p-values are calculated using 2-way clustered
standard errors by firm and fiscal year. See Appendix A for variable definitions.
REAL ASSET LIQUIDITY AND ASSET IMPAIRMENTS 73
Page 73
Table 8, continued
Panel C: Firm price level value relevance regressions with after-tax fixed asset write-offs removed
Coefficients
Vuong Test (a) - (b)
Constant Pre-BVPS Pre-EPS WOPS R-squared Statistic P-value
Group 1: Illiquid industry firms using RAL_Factor (9,031 observations)
Full model -2.713 1.129 8.959 -4.083 0.154
( -0.72 ) (4.41) (5.39) ( -0.81 )
(BV + WO) only -4.968 2.089
0.113 (a)
( -1.16 ) (4.97)
BV only -4.974 2.103
0.110 (b) 4.457 0.000
( -1.20 ) 5.120
Group 2: Liquid industry firms using RAL_Factor (8,979 observations)
Full model 5.682 1.578 4.940 2.133 0.401
(9.35) (7.30) (3.58) (1.09)
(BV + WO) only 1.245 2.108
0.329 (a)
(1.00) (9.10)
BV only 0.866 2.206
0.342 (b) -3.520 0.000
0.630 8.920
Group 3: Illiquid firms using AT_SaleSW (9,047 observations)
Full model 1.997 0.798 7.150 -1.033 0.206
(0.94) (6.03) (6.03) ( -0.29 )
(BV + WO) only 0.407 1.522
0.145 (a)
(0.16) (5.94)
BV only 0.023 1.571
0.148 (b) -1.135 0.257
(0.01) (6.30)
Group 4: Liquid firms using AT_SaleSW (8,916 observations)
Full model 4.931 1.238 6.595 -0.891 0.418
(3.59) (6.31) (6.82) ( -0.76 )
(BV + WO) only -0.867 1.976
0.316 (a)
( -0.64 ) (6.92)
BV only -1.214 2.063
0.326 (b) -3.380 0.001
( -0.97 ) (6.98)
Group 5: Illiquid industry firms using DOC_CountSW (8,996 observations)
Full model -3.707 1.144 9.206 -4.083 0.155
( -0.89 ) (4.19) (5.44) ( -0.77 )
(BV + WO) only -6.127 2.126
0.113 (a)
( -1.29 ) (4.75)
BV only -6.100 2.138
0.110 (b) 4.962 0.000
( -1.33 ) (4.89)
Group 6: Liquid industry firms using Count_DOCSW (9,013 observations)
Full model 6.524 1.508 4.309 2.691 0.442
(6.93) (11.84) (3.50) (1.62)
(BV + WO) only 2.545 1.951
0.378 (a)
(2.09) (11.72)
BV only 2.340 2.028
0.389 (b) -3.296 0.001
(1.67) (10.67)
REAL ASSET LIQUIDITY AND ASSET IMPAIRMENTS 74
Page 74
Group 7: Illiquid firms using MA_CountSW (8,979 observations)
Full model -3.508 1.154 8.835 -3.992 0.151
( -0.85 ) (4.26) (5.41) ( -0.79 )
(BV + WO) only -5.878 2.108
0.112 (a)
( -1.27 ) (4.76)
BV only -5.904 2.122
0.109 (b) 4.322 0.000
( -1.32 ) (4.91)
Group 8: Liquid firms using MA_CountSW (9,181 observations)
Full model 7.059 1.620 5.000 2.676 0.435
(7.72) (8.45) (3.89) (1.37)
(BV + WO) only 1.874 2.169
0.363 (a)
(1.35) (9.29)
BV only 1.610 2.252
0.373 (b) -3.129 0.002
(1.04) (9.26)
This table provides explanatory power and results of Vuong tests for price-level value relevance regressions with
fixed asset write-offs considered separately for firm-year observations from Dec 2000 – Dec 2012. Fiscal year
indicator variables are included in all regression models. Robust t-statistics in parentheses and corresponding p-
values are calculated using 2-way clustered standard errors by firm and fiscal year. See Appendix A for variable
definitions.
REAL ASSET LIQUIDITY AND ASSET IMPAIRMENTS 75
Page 75
Table 8, continued
Panel D: Firm price level value relevance regressions for 1995-2002 SEC
EDGAR observations with underlying data available on CRSP for airline
industry firms - Coefficient differences
Dependent variable PRC (t+3 mos) PRC (t+3 mos) PRC (t+3 mos)
(1) (2) (2)
PlaneTrans PlaneTrans (Owned)
Constant 8.517*** 9.562*** 4.779
(5.605) (4.141) (0.850)
BVPS 1.280*** 0.543* 1.216*
(12.780) (2.226) (2.283)
EPS 1.770*** 3.050 1.749
(5.128) (1.230) (0.749)
LIQ Measure
-11.052 21.182
(-0.683) (0.622)
BVPS*LIQ
10.526** 0.315
(2.724) (0.106)
EPS*LIQ
-20.635 -0.057
(-0.486) (-0.004)
Adjusted R-squared 0.770 0.772 0.776
Observations 69 69 67
Robust t-statistics in parentheses and corresponding p-values calculated using
standard errors clustered by fiscal year. See the notes to Panel A of Table 4 for
definitions of airline fleet liquidity and airline-specific control variables and see
Appendix A for remaining variable definitions.
*** p<0.01, ** p<0.05, * p<0.1
REAL ASSET LIQUIDITY AND ASSET IMPAIRMENTS 76
Page 76
Table 9
Changes in Information Asymmetry Around Earnings Announcements
Dep variable ΔAssym ΔAssym ΔAssym ΔAssym ΔAssym ΔAssym ΔAssym ΔAssym ΔAssym
Baseline (1) (2) (1) (2) (1) (2) (1) (2)
RAL_Factor AT_SaleSW DOC_CountSW MA_CountSW
Constant 0.030 0.006 0.007 0.031 0.027 0.018 0.014 -0.000 -0.001
(0.585) (0.097) (0.114) (0.591) (0.501) (0.319) (0.243) (-0.002) (-0.022)
Liquidity Measure
-0.005* -0.002 0.003 0.028 -0.005 -0.002 -0.008*** -0.006**
(-1.843) (-0.775) (0.135) (1.022) (-1.520) (-0.640) (-2.807) (-2.178)
WO_Dummy -0.001 -0.001 0.010 0.029*** 0.013**
(-0.162) (-0.212) (1.401) (3.053) (2.274)
LIQ*WO_Dummy
-0.013** -0.129*** -0.012*** -0.008**
(-2.495) (-3.218) (-2.662) (-2.056)
BM(t-1) -0.002 -0.003 -0.003 -0.003 -0.002 -0.003 -0.003 -0.004 -0.004
(-0.189) (-0.260) (-0.211) (-0.207) (-0.170) (-0.275) (-0.222) (-0.351) (-0.312)
log_MCAP -0.002 -0.002 -0.002 -0.002 -0.002 -0.002 -0.002 -0.002 -0.002
(-0.459) (-0.559) (-0.593) (-0.458) (-0.499) (-0.537) (-0.571) (-0.620) (-0.633)
Cash Rank -0.014** -0.013** -0.014** -0.014** -0.014** -0.014** -0.015** -0.014** -0.014**
(-2.131) (-2.015) (-2.093) (-2.045) (-2.096) (-2.176) (-2.230) (-2.182) (-2.231)
sd_RET -0.001 -0.000 -0.001 -0.001 -0.002 0.001 -0.001 0.002 0.001
(-0.016) (-0.001) (-0.023) (-0.022) (-0.029) (0.016) (-0.012) (0.033) (0.013)
HiTech -0.011** -0.007 -0.007 -0.011** -0.011** -0.008 -0.007 -0.005 -0.004
(-2.069) (-1.162) (-1.140) (-1.987) (-2.012) (-1.292) (-1.262) (-0.796) (-0.772)
Comp1 -0.005 -0.003 -0.003 -0.006 -0.005 -0.003 -0.003 -0.002 -0.002
(-1.404) (-0.753) (-0.753) (-1.398) (-1.371) (-0.736) (-0.739) (-0.496) (-0.509)
Comp2 0.016 0.022 0.020 0.017 0.016 0.022 0.020 0.033 0.030
(0.619) (0.740) (0.652) (0.641) (0.621) (0.781) (0.675) (1.053) (0.955)
Comp3 0.001 0.000 0.000 0.001 0.001 0.000 0.000 0.000 0.000
(0.245) (0.113) (0.098) (0.256) (0.233) (0.131) (0.108) (0.095) (0.093)
HHI -0.017 -0.021 -0.022 -0.017 -0.018 -0.025* -0.026 -0.029* -0.029*
(-1.035) (-1.363) (-1.371) (-1.052) (-1.095) (-1.669) (-1.630) (-1.882) (-1.859)
Leverage 0.008 0.007 0.006 0.009 0.008 0.007 0.007 0.006 0.006
(0.498) (0.396) (0.368) (0.496) (0.484) (0.404) (0.382) (0.355) (0.323)
Rev_Growth 0.007* 0.007* 0.006* 0.007* 0.007* 0.007* 0.007* 0.007* 0.007*
(1.931) (1.848) (1.780) (1.907) (1.885) (1.885) (1.806) (1.858) (1.819)
Δpre_IB -0.184*** -0.184*** -0.181*** -0.183*** -0.183*** -0.184*** -0.181*** -0.184*** -0.182***
(-28.949) (-9.190) (-25.140) (-8.400) (-25.015) (-9.495) (-22.044) (-10.783) (-23.192)
REAL ASSET LIQUIDITY AND ASSET IMPAIRMENTS 77
Page 77
RET(t) -0.010** -0.010** -0.010** -0.010** -0.010** -0.010** -0.010** -0.010** -0.010**
(-2.281) (-2.266) (-2.273) (-2.295) (-2.306) (-2.257) (-2.259) (-2.270) (-2.271)
EMI_Down 0.177*** 0.177*** 0.172*** 0.176*** 0.174*** 0.177*** 0.171*** 0.176*** 0.171***
(4.615) (4.081) (4.487) (3.994) (4.637) (4.157) (4.510) (4.230) (4.565)
EMI_Up 0.231*** 0.231*** 0.230*** 0.230*** 0.231*** 0.232*** 0.230*** 0.233*** 0.231***
(2.820) (2.769) (2.834) (2.799) (2.856) (2.783) (2.842) (2.811) (2.853)
CR_Dummy -0.011* -0.011* -0.011* -0.011* -0.011* -0.011* -0.011* -0.011* -0.012*
(-1.655) (-1.667) (-1.686) (-1.655) (-1.681) (-1.654) (-1.669) (-1.696) (-1.705)
log_VOL 0.003 0.003 0.003 0.003 0.003 0.003 0.003 0.003 0.003
(0.977) (0.968) (1.031) (0.968) (1.045) (0.958) (1.005) (0.958) (0.996)
Numest -0.001** -0.001** -0.001** -0.001** -0.001** -0.001** -0.001** -0.001** -0.001**
(-2.322) (-2.293) (-2.197) (-2.371) (-2.399) (-2.188) (-2.018) (-2.210) (-2.093)
Year dummies? Yes Yes Yes Yes Yes Yes Yes Yes Yes
Adj R-squared 0.002 0.002 0.002 0.002 0.002 0.002 0.002 0.002 0.002
Observations 21,745 21,745 21,745 21,745 21,745 21,745 21,745 21,745 21,745
This table provides results of information asymmetry tests using firm-year observations with underlying data on analyst forecast
dispersion available in I/B/E/S from Dec 2000 – Dec 2012. Robust t-statistics in parentheses and corresponding p-values are calculated
using 2-way clustered standard errors by 3-digit SIC industry and fiscal year. See Appendix A for variable definitions.
*** p<0.01, ** p<0.05, * p<0.1
REAL ASSET LIQUIDITY AND ASSET IMPAIRMENTS 78
Page 78
Table 10
Entropy Balancing Tests
Panel A: Distributions of key covariates for firms in the top (treated) vs. bottom (control) quartile of
RAL_Factor measure of real asset liquidity
Prior to entropy balancing: Treated (8,979 units) Control (9,031 units)
Covariate Mean Variance Skewness Mean Variance Skewness
HHI 0.097 0.001 3.474 0.259 0.024 1.653
HiTech 0.933 0.063 -3.448 0.075 0.069 3.240
BM(t-1) 0.536 0.109 0.998 0.765 0.100 0.400
ΔpreIB 0.011 0.042 0.069 0.006 0.007 0.313
sd_RET 0.176 0.011 1.658 0.128 0.006 2.246
log_MCAP 6.105 3.706 0.600 6.704 3.987 0.019
Following entropy balancing: Treated (8,979 units) Control (9,031 units)
Covariate Mean Variance Skewness Mean Variance Skewness
HHI 0.097 0.001 3.474 0.097 0.002 1.910
HiTech 0.933 0.063 -3.448 0.932 0.063 -3.439
BM(t-1) 0.536 0.109 0.998 0.536 0.109 1.561
ΔpreIB 0.011 0.042 0.069 0.011 0.042 -1.440
sd_RET 0.176 0.011 1.658 0.176 0.011 1.218
log_MCAP 6.105 3.706 0.600 6.105 3.705 0.495
See Hainmueller (2012) for details on the entropy balancing program used to weight observations in the
lowest quartile of the real asset liquidity distribution (control group observations) to achieve balance
relative to observations in the highest quartile of the real asset liquidity distribution (treated observations)
on the first (means) and second (variance) moments of the distribution for each covariate specified.
Covariate balance results for remaining measures of real asset liquidity are qualitatively similar to the
above and are not tabulated for brevity, as a result. See Appendix A for variable definitions.
REAL ASSET LIQUIDITY AND ASSET IMPAIRMENTS 79
Page 79
Table 10, continued
Panel B: Main analysis regression coefficients compared to entropy-balance weighted
regression coefficients
Dependent variable WO_Dummy FWO_Dummy GWO_Dummy B/M
Model Logit Logit Logit OLS
(1) (2) (3) (4)
Panel 1: RAL_Factor
Main analysis 0.130***
-0.027**
(2.901)
(-2.289)
Entropy-balanced analysis 2.203***
-0.027
(3.821)
(-0.843)
R-squared N/A
0.372
Observations 18,010 18,010
Panel 2: AT_SaleSW
Main analysis
0.288
-0.026
(1.179)
(-0.401)
Entropy-balanced analysis
0.629**
-0.007
(2.274)
(-0.201)
R-squared
N/A
0.265
Observations 17,963 17,963
Panel 3: DOC_CountSW
Main analysis
0.084*
-0.028***
(1.758)
(-2.951)
Entropy-balanced analysis
0.961
-0.025
(1.504)
(-0.836)
R-squared
N/A
0.374
Observations 18,009 18,009
Panel 4: MA_CountSW
Main analysis
-0.004 -0.029***
(-0.083) (-2.994)
Entropy-balanced analysis
1.067** -0.005
(2.021) (-0.279)
R-squared
N/A 0.363
Observations 10,683 18,160
Treatment indicator variables track the highest quartile of real asset liquidity for each liquidity
measure. Control firms are selected from the pool of the lowest quartile of real asset liquidity.
Control variables consistent with those in Table 3 and Table 7 are included in each regression
model, including year dummy variables, but are omitted here for space. Robust t-statistics
calculated using 1-way clustered standard errors by 3-digit SIC industry in parentheses. Weights
for weighted Logit and weighted ordinary least squares (WOLS) regression models are specified by
the entropy balancing program detailed in Hainmueller (2012). See Appendix A for variable
definitions.
*** p<0.01, ** p<0.05, * p<0.1
Abstract (if available)
Abstract
I examine how the presence of a more active (liquid) resale market for real assets influences the frequency, magnitude, and timeliness of asset impairments. Consistent with an available resale market providing a useful benchmark for evaluating recorded asset values, I find that firms with more liquid real assets recognize more frequent and timelier impairments, resulting in lower book‐to‐market ratios and more conditionally conservative earnings for firms with more liquid real assets. Impairments are more frequent in tests using both industry‐level measures of real asset liquidity and firm‐specific measures of aircraft fleet liquidity for firms in the airline industry. Real asset liquidity also improves the information content of accounting values, especially book values. Finally, impairments are associated with decreases in information asymmetry around earnings announcements for firms with more liquid real assets.
Linked assets
University of Southern California Dissertations and Theses
Conceptually similar
PDF
Can I borrow your footnotes? Learning and network benefits of footnote similarity
PDF
Who manages the firm matters: the incremental effect of individual managers on accounting quality
PDF
Slashing liquidity through asset purchases: evidence from collective bargaining
PDF
Essays in asset pricing
PDF
Accrual quality and expected returns: the importance of controlling for cash flow shocks
PDF
Does fair value accounting exacerbate the pro-cyclicality of bank lending?
PDF
Essays on real options
PDF
Share repurchases: how important is market timing?
PDF
Does differential sensitivity to aggregate earnings shocks drive post-earnings-announcement drift?
PDF
Workplace organization and asset pricing
PDF
Do shared auditors improve audit quality? Evidence from banking relationships
PDF
Private disclosure contracts within supply chains and managerial learning from stock prices
PDF
Gone with the big data: institutional lender demand for private information
PDF
Does mandatory adoption of international financial reporting standards decrease the voting premium for dual-class shares: theory and evidence
PDF
Essays in corporate finance
PDF
The balanced scorecard and long-term financial performance: evidence from publicly traded companies
PDF
How large are the coinsurance benefits of mergers?
PDF
The effects of accounting performance and professional relationships on promotion, dismissal, and transfer decisions in a conglomerate
PDF
Three essays in derivatives, trading and liquidity
PDF
Internal capital markets and competitive threats
Asset Metadata
Creator
Schonberger, Bryce A.
(author)
Core Title
Real asset liquidity and asset impairments
School
Marshall School of Business
Degree
Doctor of Philosophy
Degree Program
Business Administration
Degree Conferral Date
2014-08
Publication Date
06/16/2014
Defense Date
04/30/2014
Publisher
University of Southern California
(original),
University of Southern California. Libraries
(digital)
Tag
entropy balancing,impairment,information asymmetry,OAI-PMH Harvest,real asset liquidity
Format
application/pdf
(imt)
Language
English
Contributor
Electronically uploaded by the author
(provenance)
Advisor
Subramanyam, K.R. (
committee chair
), DeFond, Mark L. (
committee member
), Erkens, David H. (
committee member
), Matsusaka, John G. (
committee member
), Phillips, Gordon M. (
committee member
)
Creator Email
bryce.schonberger.2014@marshall.usc.edu,bryce.schonberger@gmail.com
Permanent Link (DOI)
https://doi.org/10.25549/usctheses-c3-420053
Unique identifier
UC11286022
Identifier
etd-Schonberge-2551.pdf (filename),usctheses-c3-420053 (legacy record id)
Legacy Identifier
etd-Schonberge-2551.pdf
Dmrecord
420053
Document Type
Dissertation
Format
application/pdf (imt)
Rights
Schonberger, Bryce A.
Type
texts
Source
University of Southern California
(contributing entity),
University of Southern California Dissertations and Theses
(collection)
Access Conditions
The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the a...
Repository Name
University of Southern California Digital Library
Repository Location
USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA
Tags
entropy balancing
impairment
information asymmetry
real asset liquidity