Why Do Firms Fail?
Managerial Acquisitiveness and Corporate Failure
Mohammad M. Rahaman∗
April 30, 2008
Abstract
Can managerial actions precipitate corporate failure? In this paper, we focus on an important but easily identifiable
managerial action, i.e. mergers and acquisitions (M&A), whose effect on firm value is arguably random as shown
by the empirical corporate finance literature. We show that for a sample of industrial firms that use the M&A
investment technology to pursue aggressive corporate growth strategies, excessive acquisitiveness relative to the
median industry counterpart can aggravate firms’ failure hazard. After removing the failure risk arising from
various industry and aggregate economic disturbances, a one standard deviation increase around the mean of the
excessive acquisitiveness measure augments the conditional failure risk by 61% (conditional on other exogenous
variables evaluated at the mean). We find that excessively acquisitive firms shrink in market value, sink in
operating performance, and dislodge the balance between firms’ debt and assets structure by taking on more short
term debt with less liquid assets at hand between the periods of their intense M&A activities. This mismatch
between debt maturity and asset liquidity also explains why excessive acquisitiveness can pave the way to corporate
default: a one standard deviation increase around the mean of the excessive acquisitiveness measure increases the
conditional default risk by 34% (conditional on other exogenous variables evaluated at the mean) after controlling
for various determinants of financial distress that are widely used in the bankruptcy prediction literature. Using
a mediating instrument methodology, we argue that the causality from the excessive use of M&A to the firmfailure is channeled through amplified business risk along with managerial cognitive bias and limited attention
span. The mediation process seems to be stronger through the behavioral channel than the risk channel. Finally,
we document capital market myopia in disciplining excessively acquisitive managers - although the market, on
average, punishes aggressive acquirers at the time of the bid announcement, it does not do so at all quantiles of the
conditional distribution of acquirers’ cumulative abnormal return from announcement events. However, despite
this seeming myopia, the external corporate control market eventually reins in the excessive acquirers by turning
them into future targets of takeover.
JEL Classification: G33, G34
Key Words: Corporate Failure, Mergers and Acquisitions, Corporate Default
∗ Department
of Economics and Rotman School of Management, University of Toronto, Email: m.rahaman@utoronto.ca
M.M. Rahaman
1
Corporate Failure
Introduction
Can managerial actions precipitate corporate failure? As the business climate deteriorates or the guidance
of the firm falls into the hands of people with less energy and less creative genius, the firm starts sinking
deep into troubled water and there comes a time when continuing the money losing operation becomes
too painful to bear and failure becomes imminent. Managers may buy some time to save the sinking ship
by liquidating assets to finance their excessive continuation, but as the liquidity runs out the inevitable
reckoning with failure strikes hard and equity holders are faced with the ultimate decision of being acquired
or going bankrupt. Unfortunately, this scenario is all too common in the modern corporate landscape. Yet,
our understanding of the causes of failure is very limited even though this issue bears tremendous importance
for investors, managers, and policy-makers alike.
In this paper, we focus on an important but easily identifiable managerial action, i.e. mergers and acquisitions (M&A) bids, whose effect on firm value is arguably random as shown by the empirical corporate finance
literature. However, the M&A actions do entail real and financial consequences on firms. We investigate (i)
whether the excessive use of M&A investment technology relative to an industry benchmark can precipitate
corporate failure, and if it does, (ii) what the possible channels are through which it catalyzes the eventual
failure of firms.1 Mergers and acquisitions are widely-used investment technologies at the disposal of managers pursuing aggressive corporate growth strategies. In recent years, M&A deals have been ballooning both
in terms of value and volume2 , although empirical evidence in corporate finance shows that three-quarters of
mergers and acquisitions never pay off - the acquiring firm’s shareholders lose more than the acquired firm’s
shareholders gain [Lovallo and Kahneman (2003)]. In a recent article, Moeller, Schlingemann and Stulz
(2005) document that during the recent merger wave in the U.S. shareholders’ values have been destroyed
on a massive scale squandering more value in absolute dollar term than the value destruction due to M&A
during all of the 1980s.3 It is thus puzzling to see these flurries of M&A deals when we know that potential
value creation for the shareholders from this investment technology is at best random. When a number of
firms create value through M&A while an equal or greater number of firms destroy value using the same
investment technology, on average, we may not see any identifiable effect of M&A investment technology
on firm value. On the other hand, comparing a treatment sample of acquirers with a control sample of
non-acquirers confounds the identification through selectivity and due to the arguably random effect of the
treatment (in this case M&A) on firm value. To meaningfully relate the hazard of corporate failure with
the managerial M&A actions, our identification strategy focuses on a particular sample of firms that use the
M&A investment technology to pursue their corporate growth strategies and investigate whether firms that
use this technology more aggressively than the typical firm in the industry fail more often than firms that
1 We use ‘excessive use of M&A’ and ‘aggressive use of M&A’ interchangeably in this paper. We define the precise measure
of ‘excessive acquisitiveness’ in the data section of the paper. Succinctly, ‘excessive acquisitiveness’ is defined to be the degree
of managerial acquisitiveness that is greater than the median firm in the industry.
2 According to recent statistics by Thomson Financial, deals involving U.S. targets totaled $845 billion during the first five
months of 2007, 53% of the total for 2006, and 10% more than deal value in the entire first half of 2006. At the same time,
the value of M&A deals in Canada almost doubled from $89 billion to $173.6 billion (expressed in US dollars) by January 2007
and the number of deals increased by 26%.
3 Moeller, Schlingemann and Stulz (2005) show that acquiring-firm shareholders lost 12 cents around acquisition announcements per dollar spent on acquisitions for a total loss of 240billionf rom1998through2001, whereastheylost7 billion in all of the
1980s, or 1.6 cents per dollar spent. The 1998 to 2001 aggregate dollar loss of acquiring-firm shareholders is so large because of
a small number of acquisitions with negative synergy gains by firms with extremely high valuations. Without these acquisitions,
the wealth of acquiring-firm shareholders would have increased. Firms that make these acquisitions with large dollar losses
perform poorly afterwards.
2
M.M. Rahaman
Corporate Failure
utilize this technology conservatively relative to the same industry benchmark.4 This research contributes
to the existing literature in two ways. First, it addresses the yet unresolved question in corporate finance
of whether using the M&A investment technology necessarily creates value for the corporate stakeholders in
the long run and the mechanism through which the excessive use of M&A investment technology creates or
destroys firm value. Second, by linking managerial excessive use of M&A investment technology with firm
failure, it tries to shed light on the age old debate in finance of whether managers of the failed businesses
are villains or scapegoats.
Although conventional wisdom in corporate finance suggests that M&A necessitate an alteration in a firm’s
investment and financial policies, debates on what shake and shape the firm level resource reallocation
through this mechanism remain vibrant to this day. Managers may acquire new capital through M&A
instead of building up internally if the external economic disturbances alter the underlying economic fundamentals within the industry and render the existing asset structure suboptimal. In that sense, managerial
acquisitiveness is a rational response to changes in broad economic fundamentals.5 On the other hand, bull
markets may lead groups of bidders with overvalued stock to use the stock to buy real assets of undervalued
targets through mergers and acquisitions. In that sense, managerial acquisitiveness is essentially the managerial acumen to time the market to create windfall value for the shareholders.6 Irrespective of what drives
managerial acquisitiveness, in a world without frictions and agency problems, access to M&A investment
technology helps managers to achieve the optimal asset structure faster in response to deregulation and
changes in economic fundamentals in turn creating value for the shareholders. However, in the presence
of frictions and agency problems within the firm, it is not obvious whether having access to M&A investment technology will always create value for the shareholders, let alone the aggressive use of this investment
technology. In fact, one can argue from the empirical literature in corporate finance that value creation
through M&A in the short run and long run is at best random - some firms create value while an equal
or greater number of firms also destroy value. Although value creation and destruction through M&A is
a vast research question, in this paper we focus only on one tail of the value distribution and investigate
whether aggressive use of this particular investment technology can in fact destroy value for the firm more
often than the firms that use this technology relatively conservatively. And more narrowly so, we focus on a
particular set of industrial firms that use the M&A investment technology to pursue their corporate growth
strategies, and investigate the effect of excessive use of this investment technology on an extreme measure
of firm value destruction, i.e. firm failure. We hypothesize that aggressive use of an investment technology
with an uncertain value implications for the firm may lead to pitfalls in a firm’s assets and financial structure
creating structural imbalances and eventually paving the way to failure.
4 We select our sample based on whether the firm uses the M&A investment technology to pursue corporate growth strategy.
However, our identification of causality from the managerial M&A actions to the firm failure arises from the extent to which a
firm in the acquiring sample uses the M&A investment technology more aggressively than its industry peers. Since our sample
selection is not based on the degree of acquisitiveness of the acquiring firms, selection at the level of using M&A versus not using
M&A at all should not seriously confound our causality. In fact, we show later on in the paper that focusing on the acquiring
sample biases against our identification of causality between managerial M&A actions and firm failure because acquiring sample,
on average, has lower failure risk profile than the non-acquiring sample.
5 Coase (1937) is one of the earliest to argue that technological changes drive acquisitiveness. Building on the new classical
premise, Jovanovic and Rousseau (2001, 2002) provide a Q-theory of merger where technological change and the subsequent
dispersion of Q-ratio lead to high-Q firms taking over the low-Q firms. More recently, Harford (2005) reinforces the earlier
evidences by Mitchell and Mulherin (1996) and Andrade, Mitchell, and Stafford (2001) that much of the takeover activities of
the 1980s and 1990s were driven by broad fundamental factors.
6 To explain the recent merger wave, Shleifer and Vishny (2003) stress the role of stock market misvaluations. Recent
empirical works by Rhodes-Kropd et al (2004), Ang and Cheng (2003), Dong et al. (2003) and Verter (2002) find evidences
that dispersion of market valuations are correlated with aggregate merger activities.
3
M.M. Rahaman
Corporate Failure
Although firms may fail due to competing but mutually exclusive causes such as takeover, bankruptcy, and
liquidation, for the purpose of this paper, we treat all types of exit other than bankruptcy/liquidation as
failure if the ‘Buy-and-Hold’ return (capital gain plus cash dividend and share repurchase) to the equity
holders from the first trading month in CRSP until delisting is less than 0. Whenever the firm exits through
bankruptcy/liquidation, we assume that the ‘Buy-and-Hold’ return is always -100% and thus connotes failure.
We use the cumulative number of acquisition bids by the firm since the time it first appears in our data
set until the time of exit or until the end of the sample period divided by the total number of calendar
quarters the firm survives in the sample as the measure of the degree of managerial acquisitiveness.7 We
then construct an indicator variable that returns 1 if in a given calendar year the degree of acquisitiveness
of the firm exceeds that of the median firm in the industry otherwise the indicator variable returns 0. By
multiplying the degree of managerial acquisitiveness with the indicator variable we construct the measure of
excessive acquisitiveness, which is (by definition) excessive only relative to the median firm in the industry.
This construction is motivated by the industry equilibrium models where positioning with the typical firm
within the industry serves as a natural hedge for a firm in formulating its real and financial policies given the
uncertainty associated with a particular investment decision. When constructing the corresponding degree of
acquisitiveness of the industry median for a particular firm, we exclude the firm itself so that the benchmark
remains exogenous to that firm. We also normalize the excessive acquisitiveness measure by the range of
acquisitiveness across all industries in our sample so that the measure is bounded between 0 and 1 and thus
comparable across all firms and industries. Furthermore, in order to gauge the unanticipated changes in
economic fundamentals, we construct various economic disturbance measures which we describe in details
in the data section.
We find evidence that firm-level resource reallocation induced by M&A in our sample is driven by broad
fundamental factors related to firm’s size, operating performance, growth opportunity, and external economic
disturbances that alter the underlying industry fundamentals. Firms that are excessively acquisitive in our
sample grow at a stupendous rate relative to their conservative counterparts.8 Figure 1 shows that by the
9th acquisition bid, the median excessively acquisitive firm has grown by almost 1000% of its size (book
value of total assets) when it made the first acquisition bid while the conservative counterpart grew by a
modest 300%.
Using a discrete-time hazard model, we show that the excessive use of M&A investment technology does
indeed aggravate firm’s failure risk. After removing the failure risk arising from idiosyncratic firm characteristics, industry and aggregate economic disturbances beyond the realm of managerial control, a one standard
deviation increase around the mean of the excessive acquisitiveness measure can augment the conditional failure risk by 61% (conditional on other exogenous variables evaluated at the mean). Firms that eventually fail
in our sample shrink in market value, sink in operating performance, and decouple the balance between their
debt and assets structure by taking on more short term debt but with less liquid assets at hand compared
to the non-failed sample between the periods of their intense M&A activities. The excessively acquisitive
sample portrays a strikingly similar evolution of assets and debt structure to those of the failed sample. This
classic mismatch between debt maturity and asset liquidity manifests itself through an increased amount of
7 This construction design assigns higher weight to the most recent bids and lower weight to the earlier bids. For example, if
a firm survives 3 periods in our sample and in each period it makes an M&A bid then the degree of managerial acquisitiveness
in period 1 would be 1/3, in period 2 would be 2/3 and in period 3 would be 3/3.
8 We define a firm to be conservatively acquisitive if the degree of acquisitiveness of that firm is below the degree of acquisitiveness of the median firm in its industry
4
M.M. Rahaman
Corporate Failure
default risk for the excessively acquisitive firms in our sample. A one standard deviation increase around
the mean of the excessive acquisitiveness measure can increase the conditional default risk by almost 34%
(conditional on other exogenous variables evaluated at the mean) after controlling for other determinants of
financial distress that are widely used in the bankruptcy prediction literature. These findings are statistically
robust to alternative specifications.
We hypothesize that the causality from the excessive use of M&A investment technology to the firm failure
could be channeled by three mediating instruments and correspondingly we develop three hypotheses in
the spirit of the three predominant paradigms that try to explain the corporate failure phenomena in the
modern corporate landscape. Hypothesis 1, along the line of the standard rational economic theory, argues
that frequency of poor outcomes is an unavoidable result of managers taking rational risks in uncertain
situations. Given the hard-to-predict stochastic external environment, firm failure is a phenomenon beyond
the realm of managerial control. Hypothesis 2, along the spirit of the behavioral theory, argues that when
forecasting the outcomes of risky projects executives all too easily fall victims to what psychologists call
the planning fallacy. In its grip, managers make decisions based on behavioral optimism or conservatism
rather than on rational balance of gains, losses and probabilities thus paving the way for failure. And finally,
hypothesis 3, along the vein of the bounded rationality theory, argues that managers have limited capacity to
process information and excessively acquisitive managers suffer from this limitation more severely than their
conservative counterparts because excessive acquisitiveness, demanding greater attention allocation, may
divert managerial attention away from the relevant economic functions of the firm thus worsening operating
performance and eventually leading to failure. Using a mediating instrument methodology following Baron
and Kenny (1986) and Judd and Kenny (1981), we find strong evidence of mediation through aggravated
business risk and managerial cognitive bias. We also find weak evidence of mediation through managerial
attention distortion arising from the increased number of lawsuits filed against the acquirers as a result
of their M&A activities. From these findings we argue that the causality from the excessive use of M&A
investment technology to the firm failure is channeled through amplified business risk coupled with managerial
cognitive bias and attention distortion. However, the mediation process seems to be stronger through the
behavioral channel than the business risk channel. Finally, we find evidence that capital market reaction
to the M&A announcement events and to the various mediating instruments are broadly inconsistent with
the ultimate effects of these measures on the failure hazard of firms. In our sample, the market, on average,
punishes aggressive acquirers at the time of the bid announcement but it does not do so at all quantiles of the
conditional distribution of acquirers’ cumulative abnormal returns from bid announcements revealing a sense
of myopia in the capital market reaction.9 However, despite this seeming myopia, the external corporate
control market eventually reins in the excessive acquirers by turning them into future targets of acquisition.
The remainder of the paper is organized as follows. Section II illustrates the contemporary literature involving
the debate of value creation and destruction through M&A and the causes of corporate failure. Section III
discusses the data and variable construction. Section IV presents the regression analysis involving the
determinants of firm-level acquisition propensity in the business sector and section V estimates the effects of
managerial excessive acquisitiveness on firm failure hazard with various robustness tests. Section VI develops
the empirically testable hypotheses delineating the channels through which excessive acquisitiveness catalyzes
failure and empirically tests those hypotheses. Finally, section VII discusses the role of the capital market
9 We calculate the cumulative abnormal return around a three day event window - the day of bid announcement, one trading
day before announcement and one trading day after announcement.
5
M.M. Rahaman
Corporate Failure
in disciplining managerial acquisitiveness with section VIII presenting the concluding remarks of the paper.
2
Related Literature
The contribution of this research is related to two broad questions in the corporate finance literature. First,
why do firms fail? And second, does having access to M&A investment technology necessarily create value
for the firm? In the subsequent parts of this section we highlight the contemporary debates on these two
broad themes and discuss how these are related to the current research question, i.e. can the excessive use of
M&A investment technology by managers pursuing aggressive corporate growth strategies aggravate firms’
failure risk?
2.1
Corporate Failure: The Debate
The fiery debate on why firms go bust remains flamboyant ever since Alfred Marshall (1890) argued that
collapse may be the consequence of the firm’s own success. Schumpeter (1942), on the other hand, argues that
the stability of any economic equilibrium is constantly perturbed by the forces of creative destruction. As
new innovations arrive, the competitive positions of existing technologies deteriorate and eventually succumb
to the creative forces of destruction of new innovations. During the punctuated flux of creative destruction,
resources move from lower to higher value users and remain with the state-of-the-art users until the process
repeats itself. Self-interested firms do not internalize the destruction of rents generated by their innovation
and hence introduce a business-stealing effect that forces others to leave the industry [Aghion and Howitt
(1992)]. These models generate business failure as the denouement of endogenous growth dynamics while
abstracting away from the firm and managerial idiosyncrasies.
Theoretical models incorporating ‘passive-learning’ [Jovanovic (1982), Hopenhayn (1992) and Cabral (1993)]
depict firms as entering uncertain of their growth opportunities and then receiving noisy signals of their
capabilities which in turn induce them to expand, contract or exit. These models predict exit hazard as a
function of firm’s age because low capability firms learn of their poor fitness only from their experiences.
Empirical evidence in favor of these models include Evans (1987) and Dunne, Roberts and Samuelson (1989).
In contrast to the ‘passive-learning’ models, ‘active-learning’ models formulation [Nelson and Winter (1978)
and Ericson and Pakes (1998)] allows firms to invest in uncertain but expectedly profitable ventures and grow
if successful, shrink or exit if unsuccessful. More recently, Cooley and Quadrini (2001) introduce financial
frictions in a basic model of industry dynamics with persistent shocks and show how financial factors affect
firm survival through the internal finance channels. These standard economic models of firm life-cycle assume
that entrepreneurs and managers know and accept the odds because the rewards of success are sufficiently
enticing. Corporate debacles are the result of rational choices of the executive that have adverse effects due
to the external business environment beyond the realm of managerial control. By empirically assessing the
role of external macroeconomic conditions on business failure, Bhattacharjee, Higson, Holly and Kattuman
(2004) find more bankruptcies and fewer acquisitions in periods of high economic instability in a sample of
publicly quoted firms in the U.S. and the U.K. as previously argued by Sheilfer and Vishny (1992). The
impact of macroeconomic instability on exit through bankruptcy in the U.S. is much smaller compared to
6
M.M. Rahaman
Corporate Failure
the U.K. due to the Chapter 11 bankruptcy code which insulates defaulted firms from being taken over by
their creditors.
In stark contrast to the standard economic theory, behavioral models depict economic agents as irrational or
at best bounded rational. Behavioral models [Conlisk (1996)] argue that economic agents make systematic
errors by using decision heuristics or rules of thumb. In application to corporate finance, these behavioral
models [Lovallo and Kahneman (2003)] portray executives as suffering from delusional optimism and in
its grip they all too often fall victims to what psychologists call the planning fallacy. They overestimate
benefits and underestimate costs, spin scenarios of success while overlooking the potential for miscalculation and mistake. Delusional optimist executives do not easily evolve into rational decision makers since
important corporate decisions are rather infrequent and involve noisy feedback [Heaton (2002)]. One viable
way through which the managerial irrationality can be arbitraged away is corporate takeover although it involves high transaction costs and is difficult to implement. The resounding implication of behavioral models
is that corporate debacles are not best explained by rational choices with adverse effects, but rather as a
consequence of flawed decision making. Empirically, Malmendier and Tate (2005) deem CEOs who persistently fail to reduce their personal exposure to company specific risk as overconfident. They show that CEO
overconfidence can account for corporate investment distortions by overestimating the return to investment
projects and perceiving external funds unduly costly. In another paper, Malmendier and Tate (2003) argue
that overconfident CEOs overestimate their abilities to generate returns, both in their current firms and in
potential takeover targets. Thus, on the margin, they undertake mergers that destroy value. Besides the
behavioral trait of optimism, Hirshlifer and Thakor (1992) show that when managers are concerned about
reputation building this may lead to excessive conservatism relative to shareholders’ optimum in investment
policy in favor of relatively safe projects, thereby aligning managers’ interests with those of the bondholders
even though managers are hired and fired by the shareholders. They also argue that conservatism induced
by managerial reputation building may ex-ante make shareholders better off by enhancing the debt capacity
of the firm.
On the empirical side of untangling the forces that lead to corporate debacles, one of the earliest attempts was
taken by Asquith, Gertner and Scharfstein (1994) who argue that economic distress is the most significant
cause of financial distress in their sample of junk bond issuers. Denis and Denis (1995) analyze a sample
of levered recapitalized firms and argue that poor operating performance is largely due to industry wide
problems such as surprisingly low proceeds from asset sale and negative stock price reactions to the economic
and regulatory events associated with the demise of the highly levered transaction market. Lang and Stulz
(1992) find evidence that industry rather than firm specific factors matter for firm bankruptcy. Opler and
Titamn (1994) show that highly leveraged firms in a poorly performing industry are more likely to lose
substantial market share than less levered firms within the same industry. In the spirit of the current paper,
Khanna and Poulsen (1995) show that managers of financially distressed firms make similar decisions to
their financially healthy counterparts prior to their fall from the grace. They argue that firms plunge into
financial distress due to factors outside the domain of managerial control.
Delving deeper into the state of research on corporate debacle reveals the following theoretical and empirical
regularities. New classical theory of corporate failure puts the blame squarely on the underlying forces of
creative destruction in the economy. Managers tirelessly trying to understand the industry dynamics learn
7
M.M. Rahaman
Corporate Failure
and improve their capabilities. When a firm fails, it is because of stochastic hard-to-predict external shocks
beyond the realm of managerial control. On the contrary, the behavioral theory argues that managers,
just like any other economic agents, suffer from cognitive biases and make systematic errors in judgment
by sometimes overestimating the odds of success due to excessive optimism and other times overestimating
the odds of failure due to excessive conservatism. However, if these behavioral traits are random then, on
average, cognitive biases may not have any identifiable effects on firm failure. Thus, the behavioral theory
makes another critical assumption that these traits are persistent and eventually renders the firm inefficient,
pushing it to the brink of debacle. Empirically, we know that exogenous shocks lead to resource reallocation
in the industry and in the process some firms exit making room for the more efficient ones. Managers are
helpless spectators who do their parts and let the natural course of creative destruction take its toll. But
we know very little about how exogenous shocks and managerial actions co-determine the failure hazard of
a firm. This paper attempts to fill that gap by investigating whether excessively acquisitive firms fail more
often than others, and if they do whether it is because the rational decisions of managers have unpredicted
adverse effects or simply because of biased decision making.
2.2
M&A Investment Technology and Value Destruction: The Debate
What effect does M&A have on firm failure? More importantly for this paper, can the excessive use of M&A
investment technology by managers harbinger the inevitable failure of firms? This question is inherently
linked with the broader question in corporate finance - does access to M&A investment technology necessarily
create value for the shareholders? The effect of M&A on shareholders’ wealth has been extensively studied
in the literature. The related research in this area is primarily focused on short term and long term effects
of M&A on firms’ equity prices and operating performances. Financial performance of mergers surrounding
the announcement date is almost unanimously positive in terms of cumulative abnormal return for the
target firms while the performance of the bidding firms is arguably random. While some papers have
reported significantly positive performance for bidding firms, quite a few others have found either zero
performance or even negative performance. In an often cited review article, Roll (1986) concludes that the
null hypothesis of zero abnormal performance of acquirers should not be rejected. While there have been
many subsequent articles, the results appear to be mixed enough that Roll’s conclusion appears to hold
[Agrawal and Jaffe (1999)]. In a recent article, Lovallo and Kahneman (2003) argues that three-quarters
of mergers and acquisitions never pay off - the acquiring firm’s shareholders lose more than the acquired
firm’s shareholders gain. However, Moeller, Schlingemann and Stulz (2005) show that losses occur because
of a small number of acquisitions with negative synergy gains done by firms with extremely high valuations.
Without these acquisitions, the wealth of the acquiring-firm shareholders would have increased. Firms that
make these acquisitions with large dollar losses perform poorly afterward.
Studies on the long term effect of M&A on shareholders’ wealth gained momentum after Franks, Harris and
Titman (1991). These authors and subsequent papers find some evidence of statistically significant negative
abnormal returns. However, some studies have found evidence of significant underperformance only for
subsets of bidders. Rau and Vermaelen (1998) find that low book-to-market “glamour” firms underperform
following acquisitions and Loughran and Vijh (1997) find that firms that use stock as the method of payment
experience long-run underperformance. A recent paper by Mitchell and Stafford (2000), which reviews
the long-run return literature, questions the common methodology of calculating buy-and-hold returns and
8
M.M. Rahaman
Corporate Failure
forming event-time portfolios. They show that positive cross-correlations for event firms, especially in dealing
with events that cluster in time and industry, such as M&A, invalidate the bootstrapping approach used for
statistical inference in this methodology. Instead, they implement a calendar portfolio approach advocated
by Fama (1998). This approach does not suffer from the above problems. Using the methodology proposed
by Mitchell and Stafford (2000), Harford (2005) find that evidence of long-run underperformance of M&A is
mixed, consistent with the findings of Moeller, Schlingemann, and Stulz (2005) that large acquirers destroy
billions in value while small acquirers actually create value in mergers. While it is fair to conclude from the
existing literature that the long and short term effects of M&A on a firm’s performance are at best random,
the literature has not addressed so far the issue of whether too much use of the investment technology can
in fact precipitate failure. By focusing on managerial excessive use of M&A investment technology, whose
effect on firm value is arguably random, we wish to address the broader question of whether managerial
actions can indeed precipitate corporate failure.
3
3.1
Data
Sample Selection
We use the Thomson Financial SDC Platinum Merger and Acquisition data set to identify the corporate
M&A decisions. SDC details all public and private corporate transactions involving at least 5% of the
ownership of a company where the transaction was valued at $1 million or more, but after 1992, deals of any
value (including undislcosed values) are covered. Sample transactions in SDC include mergers, acquisitions,
leveraged buyouts, stake purchases, tender offers, stock swaps, privatizations, reverse acquisitions, spin offs
and split offs, asset sales and divestitures, and bankruptcy liquidations. We focus on the U.S. industrial firms
and collect all SDC documented deals involving U.S. acquirers and targets from 1979 until 2006 totaling
208105 deals. We then match the SDC deals with the merged COMPUSTAT-CRSP dataset using the 6-digit
cusip, ticker symbol and company name. Through this process we could trace 76797 transactions involving
13333 acquirers and 22437 transactions involving 9577 targets in the merged COMPUSTAT-CRSP dataset.
We then apply another filter and keep only the deals for which we have CRSP daily stock price data on the
transaction date, one day after the transaction date and at least two months of daily stock price data prior
to the transaction date. This filter ensures that we have sufficient history of daily stock price data prior
and after the transaction date to calculate cumulative abnormal return to the equity holders as a result of
the transaction. The final data set contains 63613 transactions involving 10779 distinct acquiring firms and
3582 deals involving 2124 distinct target firms. We use Fama and French (1997) industry classifications to
categorize the deals into one of the 49 industries based on the reported four digit SIC in SDC.
We also collect firm level financial data from the quarterly COMPUSTAT industrial file. To identify the
final status of firms in our data set, particularly in cases when firms drop out of COMPUSTAT, we use the
yearly COMPUSTAT data footnotes AFTNT33, AFTNT34 and AFTNT35 which code, respectively, the
month, year and reason of deletion from the COMPUSTAT data file. We also verify these footnotes with the
CRSP delisting codes to accurately identify the reason as well as the precise time of exit. We also collect all
default and subsequent bankruptcy and reorganization events from the Moody’s Default Risk Services (DRS)
9
M.M. Rahaman
Corporate Failure
database, SDC Corporate restructuring database, and LoPuki’s Bankruptcy Research Database (BRD) for
the period of 1980 to 2006. We then manually combine the default and bankruptcy data with the merged
COMPUSTAT-CRSP data set taking into account historical name changes, cusip and ticker symbol changes.
Our final acquiring sample consists of 10779 firms and out of those 10779 bidding firms, 6144 (57%) firms
eventually drop out of COMPUSTAT-CRSP while the rest 4635 (43%) firms remain active until the end
of our sample period. Of the firms that eventually exit the industry, 445 (7.24%) are either bankrupt or
liquidated, 4338 (70.61%) are acquired, and the rest 1361 (22.15%) drop out due to other reasons such as
leverage buy out, management buy out, dropping off the exchange.
3.2
Sample Description
Table 1 reports the characteristics of the sample firms used in the subsequent sections to put forward the
main findings of the paper. It presents two panels of statistics, one for the bidding firms and the other for
the targets involving each M&A deal. Bidding firms in our sample are significantly larger than the targets
both in terms of book and market value. Bidders also have significantly higher operating performance
(Net Income/Total Assets, EBITDA/Total Assets) than their target counterparts. Although bidding firms
have lower leverage and lower total liabilities to total assets ratio relative to target firms, decomposing the
liabilities into shorter and longer term reveals that the favorable liability position of the bidders stems from
the structure of their debt tilting towards the longer term as opposed to the targets whose liabilities seem
to be more of shorter term although the difference is not statistically significant. In terms of cash and
immediate liquidity positions, targets fare better than bidders at the time of the deal announcement but
fare worse in terms of asset structure since bidders have a relatively more liquid assets structure (Fixed
Assets/Total Assets). Bidding firms survive longer in our data set and also make more bids compared to
target firms, and in doing so, an average bidder pays around 13% control premium to an average target
reflected in the difference of the cumulative abnormal return at the time of the bid announcement. In the
spirit of the Q-theory of Jovanovic and Rousseau (2001, 2002), table 1 also shows that indeed bidding firms
have significantly higher growth opportunities measured by market-to-book ratio than their corresponding
targets. Succinctly, bidding firms are larger in size, better in operating performance, have higher growth
opportunities, more liquid assets structure, and fewer debt obligations in the shorter term. As a result, they
live longer in the data set, make more bids and also pay a control premium to the targets for that. On
the contrary, targets are smaller in size but rich in cash relative to their corresponding bidders. They exit
the data set early but also get a significantly higher premium from their bidders to do so. Henceforth, we
exclusively focus on the bidding firms’ sample to investigate the causes of business failure since firms in this
set are actively pursuing expansion by bidding for other firms’ assets, and hence, reveal their preferences
to remain going concern rather than leaving the industry. Moreover, this set of firms is financially and
economically healthier than the target sample, thus focusing on them makes good sense to bias our empirical
investigation against finding willing scapegoat who are more likely to go bust anyway.
[Table 1 is about here]
10
M.M. Rahaman
3.3
3.3.1
Corporate Failure
Variable Construction
Firm Failure
The primary dependant variable of interest in our investigation is firm failure. Failure is inherently linked
with value destruction and consequently we define failure when we believe that firms exit after destroying
either debt holders’ value or equity holders’ value. Whenever a firm exits through liquidation, both the
equity holders’ and the debt holders’ value get curtailed while in the case of exit through bankruptcy,
typically equity holders’ wealth evaporates. In both cases, i.e. exit through bankruptcy and liquidation, the
firm fails to preserve value for at least one of its stakeholders and thus also fail according to our criterion.
Whenever a firm exits through means other than bankruptcy/liquidation, we calculate the ‘Buy-and-Hold’
return from the monthly CRSP return (including dividend) from the first trading month until the firm gets
delisted from CRSP in the following way:
BHRiT =
T
Y
1 + rit − 1
(1)
t=1
where BHRiT is the ‘Buy-and-Hold’ return at the time of exit, t = 1 is the first trading month, t = T
is the last trading month in which the firm gets delisted from CRSP, and rit is the monthly CRSP return
(including dividend) for firm i in our sample. If BHRiT < 0 it means that if an investor puts $1 in the
stock of that company in the beginning, at exit he/she gets back less than $1 that is equity’s value has
been destroyed. In other words, the firm fails according to our criterion. If the firm is still active while
BHRit < 0, we do not classify it as a failed firm simply because we do not want to ignore the potential of
the firm in creating value in light of the future resolution of economic uncertainties. With this definition of
firm failure we classify 2789 (25.87%) of the firms in the acquiring sample as failed firms and out of those
failed firms, 445 (15.96%) firms exit the sample through bankruptcy/liquidation, 1268 (45.46%) firms exit
the sample through acquisition, and the remaining 1076 (38.58%) firms exit the sample through other means
such as leverage buy out, management buy out, dropping off the exchange.
3.3.2
Managerial Excessive Acquisitiveness
The primary explanatory variable of interest in our investigation is the extent to which managers aggressively
use M&A investment technology relative to an industry benchmark. We define this degree of managerial acquisitiveness over and above the industry benchmark as our measure of managerial excessive acquisitiveness.
In order to construct the industry benchmark we focus on the importance of industry equilibrium forces to
firm’s real and financial structure. Maksimovic and Zechner (1991), Williams (1995) and, Fries, Miller and
Perraudin (1997) show that industries can play a subtle role in the determination of within-industry financial
and real structure. Put simply, these models emphasize the simultaneity of financial structure, technology,
and risk, and endogenize the distribution of firm characteristics within industries. Maksimovic and Zechner
(1991) show that in industry equilibrium, a firm’s financial structure is irrelevant because a technology’s
risk and profitability depend not only on ex-ante characteristics but also on how many firms adopt that
technology. Thus, adoption of a technology with uncertain payoff is very risky for the first mover but when
11
M.M. Rahaman
Corporate Failure
more and more firms start to adopt the technology risk dissipates and in industry equilibrium positioning
with the average firm in the industry serves as a natural hedge for the firm. Mackay and Philips (2006)
empirically find that positioning with the median firm in the industry indeed serves as a natural hedge for
firms simultaneously making investments, financing and business risk decisions. Motivated by this argument,
we use the M&A bids of the sample median firm in the industry as benchmark assuming that the median
firm behaves as a typical firm in industry equilibrium and acquiring decision of the median firm is driven by
some underlying economic fundamentals that necessitate restructuring of corporate assets. The distance to
natural hedge DIST. N Hijt of firm i in industry j at time t is given by:
DIST. N Hijt
Xijt − M edian(X−ijT )
=
Range Xijt − M edian(X−ijT ) ∀ i ∈ ψ(j, T )
(2)
where Xijt is cumulative number of M&A bids of firm i in industry j until calendar quarter t divided by the
total number of calendar quarters the firm survives in our sample and ψ(j, T ) is the set of all firms in industry
j and calender year T . We normalize the cumulative number of bids of a firm by the total number of calendar
quarters the firm survives in our sample to attenuate the survivorship bias in the managerial acquisitiveness
measure, i.e. the longer the firm remains active in the industry, the more likely it is to undertake a greater
number of acquisitions. This construction design also assigns more importance to the most recent bids while
giving less weight to the earlier bids. We calculate the corresponding industry median for firm i in industry
j for each calendar year T . When calculating the median for a particular firm i we include all firms in
calender year T in firm i’s industry
but exclude firm i itself so that the benchmark remains exogenous to
10
the firm. Moreover, we divide Xijt − M edian(X−ijT ) by its range across all firms and industries at time
T to make the distance to natural hedge comparable for all firms in all industries in a given period. This
distance to natural hedge proxy (i) reflects a firm’s acquisitiveness; (ii) measures the distance between a
firm’s acquisitiveness and the typical firm in the firm’s industry; and (iii) it is comparable across industries
since it is unit free and bounded between 0 and 1. From the distance to natural hedge proxy we define our
measure of the degree of managerial excessive acquisitiveness in the following way:
EXCESSIV E ACQijt = DIST. N Hijt × I(Xijt −M edian(X−ijT )>0)
(3)
where I is an indicator function that returns 1 if Xijt is above the industry median and returns 0 if Xijt is
below the industry median.11 Table 2 reports the differential firm characteristics at the time bid announcement for the excessively acquisitive bidders vis-a-vis their relatively conservative counterparts. It quite
vividly shows that excessively acquisitive bidders are larger in size and better in operating performance but
fare worse in growth opportunities compared to their relatively conservative counterparts at the time of bid
announcement. To finance excessive acquisitiveness, bidders take on more leverage while their liquid assets at
hand shrink. Moreover, the average and median stock price performance surrounding the bid announcement
is worse for the excessively acquisitive bidders relative to their conservative counterparts - they, on average,
10 We
impose the restriction of at least 5 or more firms to calculate the median in a given year.
example, lets assume that there are only two firms in our data set and both of them are in the same industry and
survive exactly 4 quarters or 1 year. Firm 1 makes 4 bids in total, one in each period, and firm 2 makes 2 bids in total 1 in each
of the first two periods and no bid in the last two periods. Then the degree of acquisitiveness of firm 1 and firm 2 from period
1 to period 4 would be (1/4, 2/4, 3/4, 4/4) and (1/4, 2/4, 2/4, 2/4), respectively. The corresponding industry median for firm
1 and firm 2 would be 0.5 and 0.625, respectively. The excessive acquisitiveness for firm 1 and firm 2 before adjustment would
be (0, 0, 0.25, 0.5) and (0, 0, 0, 0), respectively. After adjusting with the range of excessive acquisitiveness across both firms in
the industry the excessive acquisitiveness measure becomes (0, 0, 0.5, 1) for firm 1 and ( 0, 0, 0, 0) for firm 2.
11 For
12
M.M. Rahaman
Corporate Failure
lose 1% in value surrounding the announcement event due to their aggressive acquisitiveness after correcting
for broad market return on that day.
[Table 2 is about here]
3.3.3
Other Exogenous Variables
i. Idiosyncratic productivity shocks: To estimate the idiosyncratic productivity shocks of each firm, we
assume that all firms have access to the following production technology:
α
Yijt = Aijt × Kijt
L1−α
ijt
(4)
where Yijt is the sales revenues, Kijt is the capital stocks, Lijt is the number of employees, and Aijt is the
idiosyncratic total factor productivity of firm i in industry j and at time t. By taking natural logarithm, we
get:
yijt = aijt + α.kijt + (1 − α).lijt
(5)
We then use the methodology developed by Ollay and Pakes (1996) to estimate the productivity shocks of
firms from the above trans-log production function.
ii. Industry demand and supply shocks: For each of the Fama-French (1997) industries we calculate the
total industry net sales from the quarterly COMPUSTAT data using item 2 as a proxy for industry demand.
We also calculate the total industry costs of goods sold from the quarterly COMPUSTAT data using item
30 as a proxy for industry supply. We then decompose these series into trend and irregular components
using the Hodrick-Prescott (H-P) filter. The H-P filter calculates the trend component by minimizing the
following loss function:
T X
t=1
2
et
Xt − X
2
T X
e
e
Xt − Xt−1 − Xt−1 − Xt−2
+λ
(6)
t=3
et is the trend component of the series. The first term punishes the
where Xt is the actual series and X
(squared) deviations of the actual series from the trend; the second term punishes the (squared) acceleration
(change of change) of the trend level. The method thus involves a trade-off between tracking the original
series and the smoothness of the trend level: λ = ∞ generates a linear trend, while λ = 0 generates a trend
that matches the original series. Ravn and Uhlig (2002) have shown that the smoothing parameter should
vary by the fourth power of the frequency observation ratios, so that for annual data a smoothing parameter
of 6.25 and for monthly data a smoothing parameter of 129,600 is recommended, while for quarterly data a
smoothing parameter 1600 is commonly used. After decomposing the actual series into trend and irregular
components, we calculate the series instability by estimating the acceleration (change of change) of the
irregular component. Thus, the instabilities or shocks in the industry demand and the industry supply series
are given by:
et − Xt−1 − X
et−1
et−1 − Xt−2 − X
et−2
Xt − X
−
Xt−1 − X
(7)
13
M.M. Rahaman
Corporate Failure
iii. Industry technology shocks: We collect information about all patents for the period of 1963-2002
from the NBER patent database and convert the assigned technology class of each of these patents into the
international patent class using the methodology developed by Silverman (2002). From the international
patent class we convert them back into 1987 Standard Industry Classifications (SIC) and assign the patents
by grant year to each of our 49 Fama and French (1997) industries. We then apply the H-P filter on the
total number of patents granted each year in each of the Fama-French industries to calculate our industry
level technology shocks variable using equation 6 and equation 7.
iv. Industry regulatory shocks: We use major deregulatory initiatives during the sample period as
proxies for industry regulatory shocks. Deregulatory events and dates for our sample industries are collected
from Harford (2005) for the period of 1981-1996 and from the Wikipedia for the rest of the sample period.
v. Aggregate demand and supply shocks: We use the quarterly real GDP data from the Federal Reserve
Bank of St. Louis as a proxy for aggregate demand and the real price of crude petroleum in the U.S. from
the U.S. Energy Information Administration as a proxy for aggregate supply. Utilizing the H-P filter, we
calculate the aggregate demand and supply shocks series.
vi. Capital market instability and stock market momentum: To construct measures of capital
market instability we apply H-P filter on the Dow Jones Industrial average and the bank prime lending rate.
Using equation 6 and equation 7 we then construct measures of equity and debt market instability series,
respectively. To capture the momentum in the aggregate equity market, we apply the H-P filter on the S&P
500 index and use the smoothed trend portion of the series as our proxy for momentum in the aggregate
equity market.
vii. Industry merger momentum: A plethora of evidence in corporate finance shows that mergers and
takeovers come in waves. Identification of restructuring waves, however, has been a difficult one although
it has been widely recognized in the literature that there have been three distinct waves respectively in the
1980s, 1990s and 2000s [Harford (2005) and Andrade, Mitchell and Stafford (2001)]. Following Mitchell and
Mulherin (1996), Harford (2005) defines a wave as the highest clustering of M&A bids in any of the adjacent
24 months in each of the distinct merger wave decades that conforms to a simulated empirical distribution.
The 24 months length of a wave is rather arbitrary. We develop a distinct method of wave identification
where the wave length is data driven rather than arbitrary. For each of the Fama-French industries we
decompose the monthly M&A bids series into trend, seasonal and idiosyncratic components using X-12ARIMA, a seasonal adjustment software produced and maintained by the U.S. Census Bureau. It is used for
all official seasonal adjustments at the U.S. Census Bureau. We use X-12-ARIMA instead of the H-P filter
because there is evidence that the H-P filter is less accurate in higher frequency data. After extracting the
idiosyncratic and seasonal components from the monthly M&A bids series, we calculate the potential merger
ejdt − X
ejdt−1 > 0, where X
ejdt is the X-12-ARIMA smoothed
momentum as the period with successive X
component of the monthly bids series in industry j and wave decade d and calender month t. Out of the
ejdt − X
ejdt−1 > 0 period as a
potential waves in industry j and wave decade d, we classify the adjacent X
wave if it has the maximum clustering of bids among all potential waves in the industry j and wave decade d
and the maximum bids clustering must also have to be unique. For robustness we also do all our estimations
using Harford (2005) and Mitchell and Mulherin (1996) definition of wave.
14
M.M. Rahaman
Corporate Failure
Armed with the necessary measures, we now turn to regression analysis to gauge the information contents
of our constructed variables and to understand how these relate to the managerial propensity to be more or
less acquisitive.
4
Managerial Acquisitiveness: What Shakes It, What Shapes It?
In an ideal world, M&A reallocate scarce industry resources from the lower to the higher value users of these
resources. However, with agency problems and behavioral biases, the level of managerial acquisitiveness may
not always reflect the objective wealth-creation motive of the executives and thus may lead to a distorted
allocation of resources within the industry which in turn may create conditions for failure for otherwise
healthy firms. Thus, before dissecting the causal mechanism between managerial acquisitiveness and firm
failure, we need to understand what shakes and shapes the managerial inclination to be acquisitive in the
first place. Gort (1969) was one of the earliest to argue that economic disturbances alter the structure of
expectations among the market participants and generate discrepancies in valuations of income-producing
assets. A non-owner with a higher valuation of firm’s assets than the owner places bid for firm’s asset
in pursuit of economies of scale, monopoly power or yet, other sources of gain. More recently, Jovanovic
and Rousseau (2002) along the vein of Coase (1937) argue that technological change alters the available
profitable capital reallocation opportunities at the disposal of firms and leads to restructuring. Empirical
evidence by Mitchell and Murhelin (1996), Andrade, Mitchell and Stafford (2002), and Harford (2005) show
that economic disturbances lead to clustering of takeover activities within industries and across time. Shleifer
and Vishny (2003), on the other hand, posit that bull markets lead groups of bidders with overvalued stock
to use the stock to buy real assets of undervalued targets through mergers. Rhodes-Kropd et al (2004),
Ang and Cheng (2003), Dong et al. (2003) and Verter (2002) find evidence that the dispersion of market
valuations is correlated with aggregate merger activities. From these recent empirical endeavors we have
a better understanding of why mergers and acquisitions cluster within industries and across time but our
understanding of how industry and aggregate disturbances propel the firm level M&A propensity is very
limited. Using our various measures of industry and aggregate economic disturbances along with idiosyncratic
firm characteristics, we provide a clear and elaborate understanding of what moves the tectonics of firm level
resource reallocation through M&A in the corporate sector.
4.1
What Drives Firm-level M&A Propensity?
Table 3 presents the regression results from a multi-period logit model using three sets of explanatory variables. The first set of variables comprises the firm characteristics, the second set captures the industry
economic disturbances, and the third set consists of aggregate economic disturbance variables. The dependant variable in the regression is a dichotomous variable which equals 1 if the firm announces an M&A
bid during the current fiscal quarter, otherwise it is 0. All explanatory variables are lagged by one period.
In each regression model, we control for the industry fixed effects, correct for clustering of bids and observations by firms, and use robust standard errors of estimates to test their statistical significance. Among
the firm characteristics, results show that size (logarithm of Total Assets) and business performance (Net
Income/Total Assets) increase the odds of making a bid to acquire assets of other firms supporting earlier
15
M.M. Rahaman
Corporate Failure
evidence [Meek (1977), Levine and Aaronovich (1986)] that the main discriminators between acquirers and
their targets are size and performance. Although higher debt obligations (Total Liabilities/Total Assets)
decrease the propensity of making bids, if most of the debts are longer term the firm is more likely to bid
for others’ assets. Shleifer and Vishny (1992) posit that asset liquidity is crucial in determining whether the
assets of the target firms will be used in their first-best use. Table 3 shows that indeed the asset illiquidity
(Fixed Assets/Total Assets) of the bidders reduces the likelihood of M&A bids. Harford (1999) finds that
firms that have built up large cash reserves are more active in the corporate control market. We, however,
find that cash holdings (Cash/Total Assets) do not increase the likelihood of M&A bids. Firms with higher
future growth opportunities, proxied by the Market-to-Book ratio and changes in firm-level total factor
productivity (TFP), are more likely to be acquisitive than others, supporting the Q-theory of merger of
Jovanovic and Rousseau (2001, 2002).
At the industry level, we find that deregulatory and technology shocks increase the M&A propensity while
industry demand and supply shocks decrease it. Bidding is more intense when the industry is experiencing
a merger wave. However, once we decompose the relevant industry economic disturbances into positive
and negative components in table 4, we find that positive industry demand and supply shocks significantly
increase the odds of M&A bids at the firm-level while negative industry demand shocks decrease the odds of
bids. These findings reaffirm the earlier evidence by Mitchell and Mulherin (1996), Andrade, Mitchell and
Stafford (2001) and Harford (2005) that fundamental economic and regulatory changes drive M&A activities.
Moreover, Gort (1969), Coase (1937), and Jovanovic and Rousseau (2002) rightly argued that technology
shocks alter the balance of growth opportunities among the industry participants and thus lead to greater
reallocation of resources through mergers and acquisitions.
At the aggregate level, both the demand and supply shocks increase the likelihood of M&A bids. While
instability in the equity market dampens the bid propensity, instability in the costs of debt actually augments
this propensity. Decomposing the aggregate disturbances into positive and negative components reveals that
only the positive aggregate demand shock matters for raising up the likelihood of M&A bids. Negative
aggregate supply shocks (by construction) decrease the costs of production in the economy and as a consequence increase the propensity of M&A bids. Instability in equity market, no matter positive or negative,
always lessens the likelihood of M&A bids but instability in the costs of debt raises the likelihood of M&A
bids if the shocks are in the positive range and lessens the likelihood if the shocks are in the negative range.
Finally, equity market momentum leads to an increased M&A activities as argued by Shleifer and Vishny
(2003) and empirically shown by Rhodes-Kropd et al (2004), Ang and Cheng (2003), Dong et al. (2003) and
Verter (2002).
[Table 3 and Table 4 are about here]
Succinctly, the findings here conform to the postulation of Jensen (1993) who relates the restructuring
activities of the 1980s to changes in technologies, input prices, and regulations. We show that disturbances
in economic fundamentals do lead to reorganization of corporate resources at the firm level. But do firmmanagers always react to the altered business environment judiciously or do they react too much or too
little?
16
M.M. Rahaman
4.2
Corporate Failure
Why Some Firms are More Acquisitive than Others?
Although the M&A propensity in our sample is, in general, driven by broad fundamental factors, some firms
seem to be more acquisitive relative to their natural hedge counterpart within the industry. In order to understand why some firms are more acquisitive than others, we estimate the idiosyncratic productivity shocks
of the sample firms in each year following Olly and Pakes (1996). We also construct two dichotomous variables to characterize the nature of M&A bids firms make. The first dichotomous variable equals 1 if the firm
receives a negative productivity shock in period ‘t’ but still announces an acquisition bid which we denote
as optimism driven M&A bid. The second dichotomous variable equals 1 if the firm has a market-to-book
ratio greater than 1 in period ‘t’ and announces an acquisition bid which we denote as growth driven M&A
bid. Table 5 reports the correlation structure of these variables with our managerial acquisitiveness measure. It shows that excessive acquisitiveness is significantly positively correlated with positive productivity
shocks firms receive in the year in which they announce M&A bids. Furthermore, both optimism driven bids
and growth driven bids are significantly positively correlated with the excessive acquisitiveness measure and
also significantly negatively correlated with the conservative counterparts. Excessively acquisitive firms also
spend significantly more in capital and have higher acquisition expenses than their conservative counterparts.
Moreover, firms with higher anti-takeover provisions, proxied by the Gompers, Ishii, and Metrick (2003) G
index, tend to be more acquisitive than their conservative counterparts. From the correlation structure of
these variables one may deduce that internal suboptimal corporate assets structure, future growth opportunities, corporate governance, and managerial behavioral biases drive excessive acquisitiveness in our sample.
Thus, to estimate the effect of excessive acquisitiveness on firm failure hazard, we use instrumental variable
estimation and also control for firm characteristics, growth opportunities, industry and year fixed effects
and a set of exogenous economic disturbances beyond the realm of managerial control that characterize the
changes in the underlying economic fundamentals and may render the current assets structure suboptimal.
[Table 5 is about here]
5
5.1
Managerial Acquisitiveness and Corporate Failure
Estimation Methodology
We use a discrete-time hazard model to estimate the failure risk of the sample acquirers. We treat each
firm-manager as a decision unit and assume that each decision unit is always at the risk of failure and the
risk process is governed by a simple form of proportional hazard function [Cox (1972)]:
λ τ, X = λ0 τ expXβ
(8)
where λ0 is the baseline hazard of failure over time τ under the condition expXβ = 1, i.e. no heterogeneity
among firm-managers. Heterogeneity among firm-managers reflected, for example, by differences in information set (X), might change the actual hazard. Here the multiplicative effect of the covariates (X) has a
clear and intuitive meaning. If expXβ > 1, the risk of failure would increase over the whole sample period,
17
M.M. Rahaman
Corporate Failure
whereas the failure risk would decrease if expXβ < 1. Without any restriction on λ0 , however, this model
postulates no direct relationship between X and τ . Cox (1972) proposed an extension of this proportional
hazard model to discrete time by working with the conditional odds of failure at each time τ given no failure
up to that point (conditional on the covariates X). Specifically, Cox (1972) proposed the model:
λ0 τ
λ τ /X
=
expXβ
1 − λ τ /X
1 − λ0 τ
(9)
Taking logs, we obtain a model
of failure at τ given no
on the logit
of the hazard or conditional probability
failure up to that time, Logit λ τ /X
= α + Xβ, where α = Logit λ0 τ
is the logit of the baseline
hazard and Xβ is the effect of the covariates on the logit of the actual hazard. Note that the model
essentially treats time as a discrete factor by introducing one parameter, α, for each possible failure time
τ . Interpretation of the parameters β associated with the other covariates follows along the same lines as
in logistic regression. Shumway (2001) argues that hazard models are more suited to analyze the failure
intensity of corporate events and shows that a multi-period logit model is equivalent to the discrete-time
hazard model with the inclusion of log of firm age among the covariates as a proxy for the baseline hazard.
In this discrete-time hazard setting, covariates X affect the hazard rate of failure and the direction of the
covariate specific effects are given by the associated β parameters. Moreover, we argue that the design
considerations of our experiment also weaken the plausibility of reverse causation. Our primary dependent
variable, i.e. firm failure, is an absorbing state in the sense that once failure occurs firms never recover and we
do not observe any of the explanatory variables for the failed firms anymore. That is, a causal effect from the
outcome variable to any of the explanatory variables does not make sense since all the explanatory variables
are measured temporally before the outcome variable. This of course assumes that managers cannot predict
failure some period ahead. If managers can predict failure ahead of the actual failure time then the reverse
causality is still a concern. To alleviate this concern we estimate the discrete-time hazard regression with
up to three lags of all explanatory variables. Since the results do not vary with higher lags we report the
results where all explanatory variables are lagged by one period.
5.2
Estimation Results
Table 6 reports the regression results from the discrete-time hazard model. The dependent variable is a
dichotomous variable which equals 1 for the last fiscal quarter in which a firm fails and 0 otherwise. All
explanatory variables are lagged by one period. We also include industry fixed effects, year fixed effects,
correct for clustering of observations by firm, and use robust standard errors to test the significance of
the estimated coefficients in each regression model. We present all coefficients in the form of logarithm of
odds ratio in the table. It shows that the most important firm characteristics that cushion against failure
are firm size, age (baseline hazard), and growth opportunity (Market Value/Book Value). Firms fail more
often during the times of industry and aggregate demand instability while stock market instability reduces
failure risk in all cases except in one specification where we use the instrumental variable (IV) estimation.
After removing the failure risk arising from the idiosyncratic firm characteristics, industry and year fixed
effects, and industry as well as aggregate economic disturbances, we find that the excessive use of M&A
relative to the industry median does indeed aggravate firm’s failure hazard. The results also show that
the further the firm is away from its natural hedge the more likely it is for the firm to fail. However, the
18
M.M. Rahaman
Corporate Failure
failure augmenting effect of DIST. N Hijt is primarily due to the excessive acquisitiveness rather than the
conservative acquisitiveness since the coefficient of Excess Acq. is always higher in magnitude than that of
the DIST. HNijt . Furthermore, inclusion of the excessive acquisitiveness measure in the hazard regression
improves the model fit, measured by McFadden’s Pseudo-R2 , by up to 36%. We can correctly identify the
failure events for our sample firms 72% of the time using model 3 in table 6 and 75% of the time using model
9, and in both cases the inclusion of the excessive acquisitiveness measure increases the likelihood of correct
identification by 6%.12
However, the causal effect of excessive acquisitiveness on a firm’s failure hazard may be corrupted by endogeneity, omitted covariates, or errors in the excessive acquisitiveness measure. These problems can be
addressed using instrumental variable estimation in linear setting but in non-linear setting instruments cannot in general be used to produce a consistent estimator of the desired causal effects. To this end, we use a
methodology developed by Hardin, Schmeidiche, and Carroll (2003) to consistently estimate the causal effect
of the excessive acquisitiveness on firm failure using instrumental variable estimation in our discrete-time
hazard model setting. A valid instrument must be highly correlated with the firm-level excessive acquisitiveness while having no clear effect on the dependent variable, i.e. firm failure, so that the correlation
between the instrument and the error term is not significantly different from zero. We instrument the degree
of excessive acquisitiveness with a measure of industry merger momentum. The M&A literature has long
recognized that intense mergers and acquisitions activities come in waves and tend to cluster within industries and across time although there are considerable debates about what drives those M&A waves. But
it is well understood that firms are more active in M&A transactions during industry merger waves than
in any other periods and the effects of greater activism during merger waves on firm failure is not obvious
from the existing literature. Harford (2005) argues that mergers before the optimal stopping point within
a wave are value creating whereas mergers after the optimal stopping point are value destroying compared
to non-wave mergers and acquisitions without any reference to firm failure. Thus, it is fair to conclude
that firm-level acquisitiveness is related with industry merger waves but industry merger waves, as far as we
know, do not have any clear-cut effects on firm failure. Using industry merger wave dummy as an instrument
for the firm-level excessive acquisitiveness we find a statistically significant causal effect of the excessive use
of M&A on firm failure. For diagnostic purpose, we do a two stage least square (2SLS) estimation and
our instrument satisfies the non-exludability criterion in the first stage with a very high F-statistics. The
instrument also statistically significantly effect firm failure in the second stage of our 2SLS estimation. For
robustness purposes, we do a false instrument experiment in which we instrument the period t − 1 excessive
acquisitiveness with the period t + 1, t + 2, t + 3, and t + 4 industry merger wave and in all cases the false
instrument do not have any statistically significant effect on firm failure, buttressing the causal as well as
temporal validity of our instrument.
[Table 6 is about here]
One could very well argue from what we have discussed so far that bad firms are more active in M&A and
firms fail not because of their relatively excessive use of M&A but because they are essentially bad firms
12 Our primary dependent variable, i.e. firm failure, is centered around .01. We consider a failure event as correctly identified
if the predicted probability from the hazard model during the fiscal quarter in which firm fails is higher than the centered value
of the dependent variable.
19
M.M. Rahaman
Corporate Failure
to begin with. In other words, if we could find a variable that influences both the excessive acquisitiveness
and the firm failure measures, it would suffice to cast serious doubt in the regression results that we have
presented above. One possible candidate for such a variable is the Gompers, Ishii, and Metrick (2003)
governance score of firms, generally known as the G index. The G index is derived from the incidence of
24 unique governance rules that proxy for the level of shareholder rights in a firm. They show that an
investment strategy of buying firms in the lowest decile of the index (strongest rights) and selling firms in
the highest decile of the index (weakest rights) would have earned abnormal returns of 8.5% per year during
their sample period. They also find that firms with lower G index values (stronger shareholder rights) had
higher firm values, higher profits, higher sales growths, lower capital expenditures, and made fewer corporate
acquisitions. We use the average value of the G index as a measure of firm quality in the sense that firms with
higher average governance scores (G index), i.e. bad corporate-governance firms, will be more acquisitive
than firms with lower governance scores, i.e. good corporate-governance firms, as shown by Gompers, Ishii,
and Metrick (2003). We find that the inclusion of the governance score as a measure of firm quality does not
alter the result that we discussed before. The governance score enters the hazard regression with or without
the excessive acquisitiveness measure and in both cases, irrespective of specifications, the governance score
does not have any statistically significant causal effect on firm failure risk while the excessive acquisitiveness
measure retains its significance although the logarithm of odds ratio declines. We are thus confident that
our estimated causal effect of the excessive use of M&A on firm failure hazard is robust.
5.3
A Quasi Experiment and Some Robustness Tests
One valid concern with our instrument and firm quality proxy is that bad firms may hide in the crowd
in a merger wave and do lots of acquisitions. Thus, when firms fail it may not be due to their aggressive
acquisitiveness during the merger waves rather it may be the case that it is easy for the bad firms to hide in
the crowd and be aggressively acquisitive during merger waves which in turn may increase the failure risk
of the excessively acquisitive sample. To address this concern and to clearly identify the causality from the
aggressive acquisitiveness to the firm failure we do a quasi experiment where we compare the failure risk
profile of the acquiring sample with the failure risk profile of the non-acquiring sample. We collect the nonacquiring sample (firms that do not appear in the acquiring sample) from the merged CRSP-COMPUSTAT
universe. We estimate the failure risk profile (hazard function) of the acquiring and the non-acquiring sample
using various baseline hazard specifications conditional on firms’ age since incorporation, a dummy variable
indicating whether the firm is in the acquiring sample, and the aggressive acquisitiveness of firms.13 Figure 2
shows the risk profile of the acquiring and the non-acquiring sample for various hazard model specifications.
It clearly delineates that failure risk profile of the acquiring sample is always below the failure risk profile of
the non-acquiring sample meaning that acquisitiveness actually lowers failure risk. However, all else equal,
when the the acquiring sample starts becoming aggressive in their use of M&A, figure 3 shows that their
failure risk profile shifts up and as they become more and more aggressive in their use of M&A it becomes
increasingly likely that they are going to fail more often than their non-acquiring counterparts. This pattern
of shifting failure risk profile is even stronger if we use a matching sample of non-acquiring firms instead of
the universe of all non-acquiring firms.14 Our experiment shows that acquisitiveness, on average, lowers the
13 For
the non-acquiring sample, excessive acquisitiveness is always 0.
use the propensity score matching using age of the firm since incorporation as the common support for both the acquiring
and the non-acquiring firms so that both the acquiring and the non-acquiring sample has similar risk profile to begin with. We
then vary their aggressive use of M&A and find even stronger shift in the pattern of the risk profile of the acquiring sample.
14 We
20
M.M. Rahaman
Corporate Failure
failure risk of firms relative to the non-acquiring sample, but excessive acquisitiveness causes the firms to
fail more often not only relative to the conservatively acquisitive firms but also relative to the non-acquiring
firms.
We report various robustness tests of the causal effect in table 7. The first robustness test shows that there
is non-linearity in the causal effect of excessive acquisitiveness in the sense that the causal effect is not
monotonically increasing in the acquisitiveness of the firm. Instead, excessive use of the M&A investment
technology drives the failure risk while conservative use of M&A actually reduces failure risk relative to
their excessively acquisitive counterparts. In the second robustness test we estimate a linear probability
model (LPM) of failure with firm fixed effects which we cannot do in the discrete-time hazard model due to
non-convergence. Inclusion of firm fixed effects removes any firm specific effects on failure, such as inherently
bad firm effect, that is constant across time and we find that excessive use of M&A increases failure risk
in this case as well. In the third robustness test, we focus on the acquiring firms for which we can observe
their complete bidding history in SDC data set since the time the firm went public that is after the year
1980 (almost 20% of the sample firms went public before 1980 for which we do not observe complete bidding
history). We find evidence of causal effect from the excessive use of M&A to the firm failure for the complete
bidding history sample as well. One potential explanation of failure could be that aggressively acquisitive
firms suffer from winners’ curse in the sense that they end winning their bids but they also end up with
bad target more often. We use the cumulative number of completed contested bids normalized by the total
number of bids by firms to construct a measure of winners’ curse and find that it does indeed increases
failure risk but winners’ curse does not have enough explanatory power to soak up the explanatory power of
our excessive acquisitiveness measure. Finally, we estimate the causal discrete-time hazard model with two
dimensional clustering (cluster the observations by firms and also by size) and find robust causal effect of
excessive acquisitiveness on firm failure.15
[Figure 2, 3 and table 7 are about here]
5.4
Economic Significance of the Causal Effect
The statistical significance of the causal effect that we discuss in the previous section does not necessarily
imply economic significance. To this end, we estimate the marginal effects of the relevant variables from
the hazard regression. We estimate marginal effects at the mean, 1/2 standard deviation below the mean,
1/2 standard deviation above the mean, and 1 standard deviation around the mean. Table 8 reports the
marginal effect estimates from the hazard regression. The results are consistent with what we have found
in table 6, where we report the logarithm of the odds ratio of the coefficients. It shows that marginal
effects are rising, as we move from 1/2 standard deviation below the mean to 1/2 standard deviation above
the mean of the excessive acquisitiveness measure, by 77% in one specification and by 82.70% in the other
specification, where we also include economic disturbances in the hazard regression. At the mean, a 1%
increase in the excessive acquisitiveness measure increases the conditional failure risk by .33% (conditional
15 If the market capitalization of the firm is in 25th percentile, we classify the firm as small cap, if the market capitalization
is between the 25th and 75th percentile we classify the firm as medium cap and if the market capitalization of the firm is more
than the 75th percentile we classify the firm as large cap.
21
M.M. Rahaman
Corporate Failure
∂Y X̄
∂Y
on other exogenous variables evaluated at the mean) calculated using ∂X
. Ȳ , where ∂X
is the marginal effect
at the mean, and Ȳ and X̄ are the means of the predicted conditional failure probability and the excessive
acquisitiveness measure, respectively. This translates into a 61% increase in conditional failure risk with
a one standard deviation increase around the mean16 of our excessive acquisitiveness measure (conditional
on other exogenous variables evaluated at the mean). We have shown that in the discrete-time hazard
framework the excessive use of M&A has economically as well as statistically significant causal effects in
increasing the failure risk. But we are yet to untangle why excessively acquisitive firms end up failing more
often than non-excessively acquisitive firms.
[Table 8 is about here]
5.5
Can the Deal Characteristics Discriminate Between the Failed and NonFailed Sample?
The focal point of the bidders and the targets firms’ interactions revolve around the specificity of the
transaction at hand. Thus, one possible explanation why excessively acquisitive firms end up failing more
often than others could be that excessively acquisitive firms make deals that are inherently inferior along
some characteristics relative to their conservative counterparts. Table 9 presents the characteristics of deals
involving the set of bidding firms researched in this paper. It presents two classes of statistics for failed and
non-failed firms in our sample. Panel-A presents the class of statistics involving deal size and execution for
which we can test the statistical significance of the estimate whereas panel-B presents descriptive statistics
generated from dummy variables involving various specificities of the deal for which no test of significance is
available. In panel-A, average deal size is US$ 41.37 million and the median is about US$ 7.93 million for the
failed sample while average and median are US$ 225.99 million and US$ 24.42 million, respectively, for the
non-failed sample, which quite evidently reveals the positive skewness of the deal size distribution. Bidders
who do not fail in our sample take on significantly larger deals than the firms that eventually fail and exit
the sample through various routes. However, once we normalize the deal value with the book and market
value of assets as well as the market value of equity, the regularity is not quite straightforward; in fact, it
reverses in all cases meaning that relative to their size the failed sample ends up making larger deals than the
non-failed sample. Panel-A also shows that average execution delay after the announcement is 46.47 days
and the median delay is 0 days for the failed sample whereas these are 66.23 days and 12 days, respectively,
for the non-failed sample. Failed firms in our sample take significantly less time to complete the deal than
their non-failed counterparts.
Panel-B of table 9 details some salient features of the transactions involving the bidding firms in our sample.
In the table, we do not observe any significant difference in the likelihood of completing a bid between the
failed and non-failed sample firms. It shows that 70.79% of total bids were eventfully completed by the
failed sample while 70.50% of total bids were eventually completed by the non-failed sample. The failed
sample, however, is 3.17% less likely to make M&A bids in a related industry than the non-failed sample.
Furthermore, failed firms are more likely to finance the deal purely with stock whereas non-failed firms
16 One standard deviation around the mean is calculated from 1/2 standard deviation below the mean to 1/2 standard
deviation above the mean.
22
M.M. Rahaman
Corporate Failure
are more likely to finance the deal with pure cash. Moreover, the failed sample is less likely to do block
purchases and bid for divested assets or divisions of target firms relative to their non-failed counterparts.
The propensity to finance the deal through internal funds is lower for failed firms while the propensity to
finance the deal through stock swap is lower for non-failed firms. Two caveats are in order. First, examining
the testable statistics in panel-A does seem to reveal some regularities although not universal about the
acquisitiveness and failure hazard of bidding firms in the sense that failed firms take on larger bids relative
to their size and also complete bids at a faster rate compared to their non-failed counterparts. Second, there
also seem to be some regularities in deal specificities of the non-failed and failed sample that might shed light
on the failure hazard of the sample firms. The failed sample acquires less in similar industries, is more likely
to finance deals with pure stock but less likely to finance deals with pure cash and internal funds, is less
likely to do block purchases and acquire divested parts or divisions of targets. But these are not statistically
testable statistics. Thus, we need to delve beyond the deal characteristics into the evolution of firms’ debt
and assets structure to fathom the deeper question of why the use of a particular investment technology
precipitates corporate debacle.
[Table 9 is about here]
5.6
Evolution of Firm’s Assets and Debt Structure
In order to delineate the evolution of debt and assets structure, we divide the firms that make exactly 3 bids
(which is also the median number of bids by firms in our sample) into (i) failed and non-failed sample and,
(ii) excessively acquisitive and non-excessively acquisitive sample.17 The left panel of table 9 presents the
evolution of differential assets and debt structures between the failed (F) and non-failed (NF) samples at
the fiscal quarter right before the first acquisition bid and the fiscal quarter right after the last acquisition
bid (in this case third acquisition bid). The right panel of table 10 presents the evolution of differential
assets and debt structures between the excessively acquisitive (X) and the non-excessively acquisitive (NX)
samples at the fiscal quarter right before the first acquisition bid and the fiscal quarter right after the last
acquisition bid (in this case third acquisition bid). In each panel, column (1) reports the differences in
e1,N F , where Z
e1,F is the median
sample median before the firm becomes active in M&A denoted as Ze1,F − Z
assets and debt characteristics of the failed sample in the fiscal quarter right before the first bid and Ze1,N F
is the median assets and debt characteristics of the non-failed sample in the fiscal quarter right before the
first bid. Column (2) reports the differences in sample median after the firm becomes inactive in M&A
e3,F − Z
e3,N F , where Z
e3,F is the median assets and debt characteristics of the failed sample
denoted as Z
e3,N F is the median assets and debt characteristics of the
in the fiscal quarter right after the last bid and Z
non-failed sample in the fiscal quarter right after the last bid. Column (3) reports the difference-in-difference
e3,N F − Ze1,F − Ze1,N F , which can
estimates between column (2) and column (1) denoted as Ze3,F − Z
e3,F − Ze1,F − Ze3,N F − Ze1,N F . It portrays the relative changes in assets and debt
also be expressed as Z
structure during the periods when the firms were active in M&A. And finally, column
(5) report
the relative
changes in percentage from column (1) to column (2) calculated as
17 Results
are even stronger if we use firms with more than 3 bids.
23
e3,F −Z
e3,N F − Z
e1,F −Z
e1,N F
Z
Ze1,F −Ze1,N F × 100.
M.M. Rahaman
Corporate Failure
Column (3), in the left panel of table 10, shows that between the periods of first and last bids (inclusive),
all performance measures decline for the failed sample relative to the non-failed sample with logarithm of
market value falling by almost 33%, net profit margin (Net Income/Total Assets) falling by 175%, and
growth opportunity (Market-to-Book) falling by 127%. At the same time, both market and book leverage of
the failed sample sky rocket with immediate debt obligations (Short term debt/Total Liabilities) increasing
by 158.33% while immediate asset liquidity (Cash/Total Assets) falling by 125% compared to the non-failed
sample. Furthermore, cash flow volatility of the failed sample increases by 48.57% compared to the non-failed
sample between these periods. Succinctly, the failed sample fares worse in operating performance, takes on
higher leverage with increased amount of debt maturing in the immediate future but with decreased liquid
assets at hand. This portrays a classic picture of debt maturity and asset liquidity mismatch for the failed
sample compared to the non-failed sample. Column (3), in the right panel of table 9 shows a similar picture
for the excessively acquisitive firms compared to their relatively non-excessive counterparts. It shows that
between the periods of first and last bids (inclusive), logarithm of market value falls by 29%, net profit
margin (Net Income/Total Assets) falls by 150%, gross profit margin (EBITDA/Total Assets) falls by 86%,
and growth opportunity (Market-to-Book) falls by 120% for the excessively acquisitive sample relative to the
non-excessively acquisitive sample. At the same time, both market and book leverage shoot-up by 177% and
288%, respectively, with bulk of the increase due to higher short term debt, which increases by 76%. But
to finance the higher leverage, relative asset liquidity (Current Assets/Current Liabilities) actually shrinks
by 259%. Quite evidently, this looks similar to the assets and debt structure of the failed sample relative
to the non-failed sample. The set of statistics presented here clearly illustrates the fact that the excessively
acquisitive sample, similar to the failed sample, during the periods of M&A activities gathered certain
asset characteristics that decimate the healthy balance between operating performance, debt maturity, asset
liquidity, and cash flow volatility. When operating performance declines, short term debt shoots up while
liquid assets at hand to finance the immediate debt obligations dry out and it becomes a deadly recipe for
failure since the firm suffers from both economic and financial distress.
[Table 10 is about here]
5.7
Excessive Use of M&A Investment Technology and Corporate Default
The formidable combination of declining operating performance and imbalance in corporate assets and debt
structure, augured by the excessive use of M&A investment technology, may become the precursor of financial
distress for firms in our sample. We should thus observe some identifiable signs of distress even before firms
fail and exit or the sample period end. To test this proposition, we identify firms that defaulted on their
debt obligations before exiting the sample. From the Moody’s Default Risk Services (DRS) database, SDC
Corporate restructuring database, and LoPuki’s Bankruptcy Research Database (BRD) we could clearly
identify 603 default events involving 578 firms in our sample for the periods of 1980 to 2006. Of those
defaulted firms 420 (73%) firms eventually exit the sample while the rest 158 (27%) firms remain active. Of
the exited firms, 46% exit through bankruptcy/liquidation, 16% exit through acquisition, and the rest 38%
exit due to other reasons such as leverage buy out, management buy out, dropping off the exchange.
We use our discrete-time hazard model discussed earlier to estimate the default hazard under alternative
24
M.M. Rahaman
Corporate Failure
specifications incorporating Altman’s (1968), Zmijewski’s (1984), and Shumway’s (2001) independent variables in their respective bankruptcy prediction models. Altman’s variables are described extensively in
Altman (1968, 2000) and Mackie-Mason (1990). Using those variables, we construct Altman’s ZSCORE as:
ZSCORE =
3.3 × EBIT + Sales + 1.4 × Retained Earning + 1.2 × W orking Capital
T otal Assets
(10)
Zmijewski’s variables include the ratio of net income to total assets, the ratio of total liabilities to total
assets, and the ratio of current assets to current liabilities. Shumway (2001) criticizes Altman (1968) and
Zmijewski (1984) and offers market driven predictors of bankruptcy. Shumway’s variables include logarithm
of market value, firm’s past excess returns, and idiosyncratic standard deviation of each firm’s stock returns.
To measure firm’s past excess return, we take the value-weighted CRSP NYSE/AMEX index return as
benchmark and subtract the index return from the monthly stock return to calculate the firm’s excess return.
The final, perhaps the most important, market driven variable Shumway (2001) uses is the idiosyncratic
standard deviation of firm’s stock returns, denoted as sigma (σ) in this paper. Sumway (2001) argues that
sigma is strongly related to bankruptcy both statistically and logically. If a firm has more variable cash
flows (and hence more variable stock returns) then the firm ought to have a higher probability of bankruptcy.
Sigma may also measure something like operating leverage. To calculate sigma for each firm i in quarter t,
we regress each stock’s daily returns on the value-weighted NYSE/AMEX index returns for the same quarter.
We then calculate sigma as the standard deviation of the residuals of this regression. To avoid outliers, all
independent variables are truncated at the 99th and 1st percentile values in the same manner as all other
independent variables.
Table 11 reports the estimated coefficients from our discrete time hazard model of corporate default. The
dependent variable is a dichotomous variable which equals 1 for the quarter in which firm defaults or files for
bankruptcy and 0 otherwise. All explanatory variables are lagged by one period and in all regression models
we include industry fixed effects, year fixed effects, and correct for clustering of observations and distress
related events, i.e. default/bankruptcy, by firms. We also use robust standard errors to test the significance of
the estimated parameters. We report the estimates in logarithm of odds ratios for all explanatory variables
and also report the marginal effects for our two key explanatory variables. It shows that irrespective of
bankruptcy prediction models, the excessive use of M&A measure increases the default risk of firms in
our sample. The estimates from the hazard regression also show that Gompers, Ishii, and Metrck (2003)
governance score, as proxy for firm quality, does not have any statistical significance in predicting corporate
default irrespective of bankruptcy prediction model specifications. Results also show that Altman’s (2000)
ZSCORE decreases default risk, current ratio (Current Assets/Current Liabilities) attenuates default hazard
in Zmijewski’s (1984) model while idiosyncratic stock price volatility from Shumway’s (2001) model always
increases the default risk in our sample. These finding are consistent with the extant literature on default
and bankruptcy prediction. More importantly, we show that inclusion of excessive acquisitiveness measure
among the set of covariates, that are widely used in the default and bankruptcy prediction models, reduces
the forecast errors of the existing models and hence, improves model predictive power. To assess the economic
significance, we estimate the marginal effects of the excessive acquisitiveness measure and find that at the
mean, a 1% increase in excessive acquisitiveness increases the conditional default risk by .19% (conditional
∂Y
∂Y X̄
. Ȳ , where ∂X
is the marginal effect
on other exogenous variables evaluated at the mean) calculated using ∂X
at the mean, and Ȳ and X̄ are the mean of the predicted conditional default probability and the excessive
acquisitiveness measure, respectively. The estimated marginal effects are statistically significant for our
25
M.M. Rahaman
Corporate Failure
excessive acquisitiveness measure across all bankruptcy prediction models. This elasticity of conditional
default probability with respect to the excessive acquisitiveness measure translates into up to 34% increase
in conditional default risk (conditional on other exogenous variables evaluated at the mean) with a 1 standard
deviation increase around the mean of our excessive acquisitiveness measure.
[Table 11 are about here]
6
Managerial Acquisitiveness and Corporate Failure: Causal Mechanisms
So far we have shown that failure arrives at a faster rate when managers use an investment technology with
uncertain value implications for their firms in an excessive manner. However, casting the blame on managers
by simply looking at the causal link between managerial action and failure hazard is rather unfair because an
ex-post bad investment decision may very well be an ex-ante good investment decision when one factors in
the uncertainties surrounding the business environment with which managers have to interact continuously.
Without proper theoretical guidance, however, one would be at sea to fathom the deeper question of whether
managers of failed businesses are villains or scapegoats? Using the two predominant theoretical paradigms
that try to explain the failure phenomena in the modern corporate landscape, we develop three hypotheses
and use a mediating instrument methodology following Baron and Kenny (1986) and Judd and Kenny (1981)
to test those hypotheses. In the subsequent parts of this section we first discuss the mediating instrument
methodology and then develop our hypotheses for empirical testing.
6.1
Mediating Instrument Methodology
In an effort to avert confounding in observational studies, economists and social scientists have devised
“Instrumental Variable (IV)” method which is based on a basic principle that the instrument must be
correlated with the explanatory variable while being uncorrelated with the outcome variable (dependent
variable). A mediating instrumental variable, on the contrary, is an auxiliary variable which fulfills radically
different conditions than those demanded by the traditional instrumental variable. A mediating instrument
must be correlated with both the explanatory variable and the outcome variable so that it can mediate the
causation from the explanatory to the outcome variable. To explain the mediating instrument methodology,
consider a variable X that is assumed to affect another variable Y . The variable X is called the initial
variable and the variable that it causes or Y is called the outcome variable. The effect of X on Y may be
mediated by a process or mediating variable M , and the variable X may still affect Y . Complete mediation
is the case in which variable X no longer affects Y after M has been controlled for, whereas partial mediation
is the case in which the path from X to Y is reduced in absolute size but is still different from zero when
the mediator is controlled. Note that a mediational model is a causal model meaning that the mediator
is presumed to cause the outcome and not vice versa. If the presumed model is not correct, the results
from the mediational analysis are of little value. When the mediational model is correctly specified, Baron
and Kenny (1986) and Judd and Kenny (1981) outline four steps in establishing mediation: (i) the initial
26
M.M. Rahaman
Corporate Failure
variable must be correlated with the outcome in a regression model where Y is the criterion variable and
X is a predictor establishing the fact that there is an effect that may be mediated; (ii) the initial variable
X must be correlated with the mediator M in a regression model where M is the criterion variable and X
is a predictor; (iii) the mediator M must affect the outcome variable Y in a regression model where Y is
the criterion variable and X and M are predictors; (iv) to establish that M completely mediates the X − Y
relationship, the effect of X on Y controlling for M should be zero. The effects in both (iii) and (iv) are
estimated in the same equation. It is not sufficient just to correlate the mediator M with the outcome Y ;
the mediator and the outcome may be correlated because they are both caused by the initial variable X.
Thus, the initial variable X must be controlled in establishing the effect of the mediator M on the outcome
variable Y . To implement the mediation process we estimate the following regression models:
E Yit = 1 | X, Z = F α + βXit−1 + δZt−1 + εit
(11)
E Yit = 1 | X, M, Z = F α + β 0 Xit−1 + θMit−1 + δ 0 Zit−1 + εit
(12)
where Yit is the firm failure dichotomous variable, Xit−1 is our measure of managerial excessive acquisitiveness, Mit−1 is a mediating instrument, Zit−1 is other control variables. If F (.) is a linear function then with
appropriate distributional assumption on εit the regression models collapse into linear probability models
(LPM), whereas with F (.) as a logistic function then with appropriate distributional assumption on εit we
get back our discrete-time hazard model. Although mediation methodology is mostly applied to linear setting, it can easily be extended to non-linear setting, particularly in the case of F (.) as logistic function. We
estimate both cases, i.e. LPM and discrete-time hazard, but report the results only for discrete-time hazard
specification. In these models, β is called the ‘total effect’ of X on Y and β 0 is called the ‘indirect effect’ of
X on Y after M has been controlled for. From these regression models,
we calculate the percent reduction in
β−β 0
× 100 and bootstrap the percent reduction
the logarithm of odds ratio as a result of mediation using
β
parameter to come up with confidence intervals. The design considerations of our mediating instrument
methodology weaken the plausibility of reverse mediation. That is, mediation from the outcome variable to
any of the explanatory variables does not make sense since in all regressions the explanatory variables are
measured temporally before the outcome variable.
6.2
Risk Channel
In the spirit of the standard rational economic theory, which posits that frequency of poor outcomes is
an unavoidable result of managers taking rational risks in uncertain situations, we treat each M&A bid
like a random lottery given the hard-to-predict stochastic external environment with some probability of
success and some probability of failure. In that sense, excessively acquisitive managers accumulate more
lotteries by vying for more acquisitions and hence, with hindsight, may either increase or decrease the
underlying business risks (cash flow volatility) of their firms. Through this channel, excessive acquisitiveness
can lead to greater business risk and eventually can increase the failure hazard of firms if more acquisitions
amplify the cash flow volatility or it may lead to lower business risk and hence reduce firm failure hazard if
more acquisitions lessen cash flow volatility through diversification, synergies, and economies of scale. We
calculate two measures of cash flow volatility. The first measure use the real cash flows of firms to calculate
BRISKit = log abs EBIT DAit −EBIT DAit−1 , where BRISKit is a simple measure of business risk, i.e.
27
M.M. Rahaman
Corporate Failure
cash flow volatility, and EBIT DAit is firm’s earning before interest, tax, depreciation and amortization. The
second measure of business risk that we use is Shumway’s (2001) sigma measure which gives the idiosyncratic
standard deviation of firm’s stock returns. Shumway (2001) argues that firms with more volatile cash flows
should have higher sigma and higher sigma also implies higher operating leverage for firms. We follow
Shumway (2001) and regress each stock’s daily returns on the value-weighted NYSE/AMEX index returns
for the same quarter and calculate sigma as the standard deviation of the residuals of this regression.
Table 12 reports the estimates from the mediating instrument methodology for the risk channel. Column 1
reports the ‘total effect’ of excessive acquisitiveness on failure hazard while column 2-4 report the mediation
of the causality between the excessive acquisitiveness and the firm failure through the BRISK measure
and column 5-7 report the mediation of the causality through the sigma measure. It shows that both the
BRISK and the sigma are statistically significantly correlated with excessive acquisitiveness but only the
sigma measure is statistically significantly correlated with firm failure. Moreover, controlling for sigma along
with excessive acquisitiveness measure reduces the absolute size of the ‘total effect’ by 3% while remaining
statistically significant. This translates into a 9% decline in the odds ratio (we report the logarithm of odds
ratio in the table) of the ‘total effect’ of excessive acquisitiveness on firm failure. The results in this table
show evidence of partial mediation through sigma because the ‘indirect effect’ is still statistically different
from 0. Thus, instead of stabilizing, excessive use of M&A amplifies cash flow volatility (hence increasing
the business risk) and the causality from excessive use of M&A to firm failure gets mediated.
[Table 12 is about here]
6.3
Behavioral Channel
We construct two measures in the spirit of the behavioral channels to investigate the mediation of the
causality from the excessive use of M&A to the firm failure.
6.3.1
Managerial Cognitive Bias
In the spirit of the behavioral theory, which posits that when forecasting the outcomes of risky projects
executives all too easily fall victims to what psychologists call the planning fallacy and in its grip, managers
make decisions based on behavioral optimism or conservatism rather than on rational balance of gains,
losses and probabilities, we argue that each acquisition bid involves some cognitive bias and excessively
acquisitive firms are more prone to cognitive bias compared to their conservative counterparts. Through
this channel, excessively acquisitive managers accumulate greater cognitive bias and over time these decision
biases, with hindsight, get imputed into the operational efficiency of the firm creating structural imbalances
in the corporate assets and debt structure precipitating failure of firms. To measure managerial cognitive
bias, we assume that the bidding decision of the benchmark firm YB is governed by the following equation:
E YB = 1|X = F (Xβ) + ε
28
(13)
M.M. Rahaman
Corporate Failure
where X is the set of economic fundamentals and ε is a stochastic error independent of X that captures
noise and other unobservable, such as luck, and ε →iid N 0, σε2 . Acquiring decisions of the upward biased
firm-manager Yup and the downward biased firm-manager Ydown are given by:
E Yup = 1|X = F (Xβ) + ε + bias upi
(14)
E Ydown = 1|X = F (Xβ) + ε − bias downi
(15)
We assume that both bias upi and bias downi are independent of X and ε, and are distributed as bias upi →
2
2
N + µup , σup
and bias downi → N + µdown , σdown
with truncation at 0. From this specification, it is
obvious that both bias upi and bias downi act as non-negative shifters in these models where bias upi
captures unobservable that systematically pushes up the likelihood of M&A bids and bias downi captures
unobservable that systematically pulls down the likelihood of M&A bids compared to the benchmark firm.
We fit a liner probability model (LPM) of Yup and Ydown on a set of firm characteristics, industry fixed
effects, year fixed effects, and a set of industry and aggregate economic disturbance variables to extract
the bias upi and bias downi from the observed firm-managerial M&A bids. The ε term in the LPM is
assumed to have two components - one component is assumed to have a strictly non-negative distribution,
and the other component is assumed to have a symmetric distribution. In the econometrics literature, the
symmetric distribution is referred to as the idiosyncratic error and non-negative component is our measure
of managerial cognitive bias. From the bias upi and bias downi we construct our managerial cognitive bias
as mgt. biasi = bias upi + bias downi .
6.3.2
Managerial Attention Allocation
In the spirit of the bounded rationality theory, which posits that agents experience limitations in formulating
and solving complex problems and in processing (receiving, storing, retrieving, transmitting) information,
we argue that managers have limited attention spans or capacities to process information and excessively
acquisitive managers suffer from this limitation more severely than their conservative counterparts. Because
excessive acquisitiveness demands greater attention allocation from the limited attention span of managers
and it may divert managerial focus from the relevant economic functions of the firms. Thus, with hindsight,
managerial attention distortions may worsen operating performance and eventually mediate the causality
from the excessive use of M&A to the firm failure. In order to construct a proxy for the managerial attention
allocation, we use the cumulative number of lawsuits filed against the acquirer as a direct consequence of
the M&A bids normalized by the total number of deals conducted by the firm. 18
From our data set we could clearly identify 491 lawsuits filed against the acquirers as a result of their M&A
18 Litigation is an everyday fact of life for American corporations. According to the Fulbright & Jaworski’s Litigation Trends
Survey, 94% of U.S. counsels canvassed said that their companies had some form of legal dispute pending in a U.S. venue. For
89%, at least one new suit was filed against their company during the past year. One third of all companies and nearly 40% of
$1 billion-plus firms project the amount of litigation to increase next year. The survey also indicates that U.S. companies spend
71% of their overall estimated legal budgets on disputes. Large U.S. companies, typically the public firms that we study in this
paper, commit an average of $19.8 million to litigation, approximately 58% of total average legal spending of $34.2 million.
More than two-thirds of large companies surveyed reported at least one new suit involving $20 million or more in claims; 17%
faced a minimum of six suits in the $20 million-plus range. Given this gloomy state of corporate litigation involving U.S. firms,
we argue that litigations arising as a result of M&A bids may drain corporate resources and distract managers’ attention from
firm’s economic functions. Thus, limited attention span may rightly mediate the causality from the excessive use of M&A to
the eventual failure of firms.
29
M.M. Rahaman
Corporate Failure
bids. Table 13 reports the estimates from the mediating instrument methodology for the behavioral channel.
Column 1 reports the ‘total effect’ of the excessive acquisitiveness measure on failure hazard. Column 24 report the mediation of the causality from the excessive acquisitiveness to the firm failure through the
managerial cognitive bias measure, while column 5-7 report the mediation of the causality through the
managerial attention allocation measure. It shows that managerial decision bias relative to the average
firm is statistically significantly correlated with the excessive acquisitiveness and the firm failure hazard.
Moreover, controlling for managerial cognitive bias along with excessive acquisitiveness measure reduces the
absolute size of the ‘total effect’ by 9% while remaining statistically significant. This translates into a 26%
decline in the odds ratio (we report the logarithm of odds ratio in the table) of the ‘total effect’ of the excessive
acquisitiveness on firm failure hazard. Column 5-7 show that greater number of litigations (and hence greater
attention distortion) is statistically significantly correlated with the excessive acquisitiveness and the firm
failure hazard, that is excessively acquisitive firms suffer from greater attention distortion and more attention
distortion brings failure at a faster rate. However, controlling for the cumulative number of lawsuits along
with the excessive acquisitiveness measure reduces the absolute size of the ‘total effect’ by a meager 1%
in terms of the odds ratio. It implies that almost all of the variations in attention allocation measure is
explained by the excessive acquisitiveness measure. Thus, after controlling for the excessive acquisitiveness
measure there is very little variations left in our attention allocation measure to explain failure risk. The
results here show strong evidence of mediation through the managerial cognitive bias measure since inclusion
of this measure sizeably reduces the absolute size of the ‘total effect’ of the excessive acquisitiveness measure.
Column 8-9 of table 13 include all channels and show that sigma measure, managerial cognitive bias measure,
and attention allocation measure are statistically significantly correlated with firm failure. When these
measures enter the discrete-time hazard regression in column 8 along with our excessive acquisitiveness
measure, together they reduce the absolute size of the ‘total effect’ by 12% which translates into a 32.51%
reduction in the odds ratio of ‘total effect’. Overall, results from table 12 and table 13 show clear evidence of
mediation from the excessive use of M&A investment technology to the firm failure through the risk channel,
proxied by the sigma measure, and through the behavioral channel, proxied by managerial cognitive bias
measure, although the mediation process is stronger through the behavioral channel than the risk channel.
For robustness purpose, we bootstrap the change in ‘total effect’ due to mediation
through managerial
β−β 0
×100 after 1000 replications.
cognitive bias and sigma measures and figure 4 depicts the distribution of
β
It quite evidently shows that mediation takes place (absolute size of ‘total effect’ shrinks) with probability
1.00 with the managerial cognitive bias measure while mediation through the sigma measure occurs with
probability 0.90 bolstering the fact that mediation is stronger through the behavioral channel than the risk
channel.
[Figure 4 and Table 13 are about here]
7
Role of the Capital Market
In an efficient capital market any adverse effects of suboptimal managerial decisions should be fully incorporated into the security prices without any substantial delay. Moreover, the disciplinary role of the external
30
M.M. Rahaman
Corporate Failure
corporate control market may come into effect to arbitrage the managerial cognitive biases away by turning
the bad bidders into good targets, thus undoing the previous unprofitable acquisitions or preventing these
firms from making future unprofitable acquisitions [Jensen (1986)]. Mitchell and Lehn (1990) document
empirical evidence that firms that subsequently become takeover targets make acquisitions that significantly
reduce their equity value and firms that subsequently do not become takeover targets make acquisitions
that raise their equity value. More recently, Zhao and Lehn (2003) document strong inverse relationship
between acquiring firms’ returns and the likelihood that their CEOs are subsequently fired, buttressing the
disciplinary role of the internal corporate control to rein in bad acquiring CEOs.
We calculate the acquirers’ cumulative abnormal return CAR(−1,+1) around a three day event window
which includes one trading day prior to the bid announcement, the day of announcement, one trading day
after the bid announcement. To calculate the CAR(−1,+1) , we estimate a market model using stock returns
from 60 trading days (estimation window) prior to the event window and use the parameters from the market
model to calculate normal returns during the event window. We then subtract the estimated normal returns
from the observed returns during the event window to the calculate abnormal returns and cumulate the
abnormal returns over three days to come up with our CAR(−1,+1) measure. We regress CAR(−1,+1) on the
excessive acquisitiveness measure, various mediating instruments, and Gompers, Ishii and Metrick (2003)
governance score to investigate the capital market reactions in response to the managerial M&A actions.
Table 14 reports the estimates from the Ordinary Least Square (OLS) regression. In all regression results
presented in table 14, we control for industry and year fixed effects as well as 26 deal characteristics reported
in the SDC data set. We also correct for the clustering of deals by firms and use robust standard errors to test
the significance of the estimated parameters. All explanatory variables are lagged by one period. It shows
that the market reacts through CAR(−1,+1) negatively to deals if the firm has been excessively acquisitive
in the past. Gompers, Ishii and Metrick (2003) governance score has negative and statistically significant
effect on CAR(−1,+1) . Quite interestingly, cash flow volatility (BRISK) has negative effect on CAR(−1,+1)
while idiosyncratic standard deviation of stock return (Sigma) has positive effect on CAR(−1,+1) . Results
are similar in column 8-7 where we also control for deal value normalized by the market value of the firm.
To understand the confounding effects of underlying business risk measures on CAR(−1,+1) we estimate the
regression at various conditional quantiles of the CAR(−1,+1) distribution.
Quantile regression is a statistical technique intended to estimate, and conduct inference about, conditional
quantile functions.19 While the OLS enables us to estimate models for conditional mean functions, quantile
regression methods offer a mechanism for estimating models for the conditional median function, and the full
range of other conditional quantile functions. By estimating an entire family of conditional quantile functions,
quantile regression is capable of providing a more complete statistical analysis of the stochastic relationships
among CAR(−1,+1) and other random variables of our interest. Figure 5 gives the effects of the excessive
acquisitiveness on CAR(−1,+1) at various quantiles of the condition distribution of CAR(−1,+1) along with the
95% confidence intervals. It shows that the market reacts positively until the 30th conditional quantile while
the reaction becomes negative and increasingly stronger at the higher quantiles. The asymmetry of market
reaction at various conditional quantiles of CAR(−1,+1) is also evident in other measures of business risk and
behavioral biases (in figure 9). It reveals a sense of myopia in the capital market response in the sense that
even though the excessive use of M&A aggravates firms’ failure hazard and the Sigma mediates the causality
19 See
Koenker, R. and Hallock, K. (2001) for more about quantile regression
31
M.M. Rahaman
Corporate Failure
from the excessive use of M&A to the firm failure, the capital market reaction through CAR(−1,+1) does not
fully reflect the failure augmenting effects at all quantiles of the condition distribution of CAR(−1,+1) .
[Table 14 and Figure 5 to Figure 6 are about here]
Despite the seeming market myopia in fully incorporating the failure augmenting effects of the excessive
acquisitiveness and the sigma, the external market for corporate control seems to be effective in turning the
excessively acquisitive firms into future targets. In table 14 we re-estimate our discrete-time hazard model to
examine the effectiveness of external corporate control market in reining the excessively acquisitive managers.
For columns 1 to 6 of table 15, the dependant variable is a dichotomous variable which equals 1 for the fiscal
quarter in which firms exit through any means, otherwise it is 0. For columns 7 to 12, the dependent variable
equals 1 for the fiscal quarter in which firms get acquired, otherwise it is 0. We also control for industry and
year fixed effects and correct for clustering of observations by firms. Furthermore, we use robust standard
errors to test the significance of the estimated coefficients, where the coefficients are reported in logarithm of
odds ratio. The estimates from the discrete-time hazard model show that the excessive use of M&A increases
both the exit and the takeover hazard. After controlling for the mediating instruments, the indirect effects
of the excessive acquisitiveness on exit and takeover hazard are still statistically significant although the
effects decline in absolute value. These findings are consistent with the earlier evidence of Mitchell and Lehn
(1990). However, the internal corporate governance, measured by the Gompers, Ishii and Metrick (2003)
governance score, does not seem to have any bite on the exit and the takeover hazard when included in
the hazard regression along with the excessive acquisitiveness measure and the mediating instruments. It
seems that higher anti-takeover provisions, i.e. higher G-Index, decreases the exit or the takeover hazard
of the sample firms although not statistically significant in all cases except one. Succinctly, findings here
corroborate Jensen (1986) in the sense that the external corporate control market plays the role of preventing
excessively acquisitive firms from making future failure augmenting acquisitions by turning them into target
of takeover in its own way.
[Table 15 is about here]
8
Conclusion
Understanding the causes of business failure is tremendously important for investors, managers, and policymakers alike. Surprisingly, our understanding of this issue is very limited primarily because when firms
fail it is very difficult to disentangle failures that arise as a result of the adverse effects of managerial
rational decisions beyond the realm of managerial control from failures that result simply because of flawed
decision making. In this paper, we focus on a particular managerial action, i.e. mergers and acquisitions
(M&A), whose effect on firm value is arguably random and investigate (i) whether the excessive use of M&A
investment technology relative to an industry benchmark can precipitate corporate failure, and if it does, (ii)
what the possible channels are through which it catalyzes the eventual failure of firms. Although mergers
and acquisitions are widely used investment technologies at the disposal of managers pursuing aggressive
32
M.M. Rahaman
Corporate Failure
corporate growth strategies, the empirical literature in corporate finance has shown that the effects of M&A
on firm value creation is at best random - a number of firms create value while an equal or greater number of
firms also destroy value. Thus, by focusing on the excessive use of this investment technology for a sample of
firms that use it to pursue corporate growth strategies, we can meaningfully relate the hazard of corporate
failure, an extreme measure of firm value, with managerial actions and possibly shed light on the age old
question in finance of whether managers of failed businesses are villains or scapegoats.
Using a discrete-time hazard framework, we find that failure comes at a faster rate for firms that use M&A
in an excessive manner relative to the median industry counterpart. After removing the failure risk arising
from idiosyncratic firm characteristics, industry and aggregate economic disturbances beyond the realm
of managerial control, a one standard deviation increase around the mean of the excessive acquisitiveness
measure can augment the conditional failure risk by 61% (conditional on other exogenous variables evaluated
at the mean). Furthermore, we find that although at the time of the bid announcement bidding firms are
larger in size, better in operating performance, have higher growth opportunities, more liquid assets, and
fewer debt obligations in the short term relative to their corresponding targets, tracking the evolution of
assets and debt structures of these bidding firms between the periods of their intense M&A activities reveals
that firms that eventually end up failing shrink in market value, do poorly in operating performance, and
decouple the balance between debt maturity and asset liquidity. They take on more short term debt while
having less liquid assets and sinking operating performances. The excessively acquisitive firms portray a
strikingly similar evolutions of assets and debt structures to those of the failed firms. This classic imbalance
between asset liquidity and debt maturity also explains why excessive acquisitiveness can trigger corporate
default in our sample even after controlling for default risk emanating from other determinants of financial
distress that are widely used in the bankruptcy prediction literature. A one standard deviation increase
around the mean of the excessive acquisitiveness measure can increase the conditional default risk by almost
34% (conditional on other exogenous variables evaluated at the mean).
In order to understand the channels through which the excessive use of M&A precipitates corporate failure,
we hypothesize, in the spirit of standard rational economic theory, that the frequency of poor outcomes is
an unavoidable result of managers taking rational risks in uncertain situations. Given the hard-to-predict
stochastic exogenous economic disturbances, firm failure is a phenomenon beyond the realm of managerial
control. In the spirit of the behavioral theory, we hypothesize that when forecasting the outcomes of risky
projects executives all too easily fall victims to what psychologists call the planning fallacy. In its grip,
managers make decisions based on behavioral optimism or conservatism rather than on rational balance of
gains, losses and probabilities thus paving the way for failure. And finally, in the spirit of the bounded
rationality theory, we hypothesize that managers have limited capacities to process information and excessively acquisitive managers suffer from this limitation more severely than their conservative counterparts,
because excessive acquisitiveness demands greater attention allocation and may divert managerial attention
away from the relevant economic functions of the firm. Attention distortion thus may worsen the operating
performance and eventually leading to failure. We construct proxies to test each of these propositions and
use a mediating instrument methodology following Baron and Kenny (1986) and Judd and Kenny (1981).
We find evidence of mediation of the causality through aggravated business risk and managerial cognitive
bias. We also find weak evidence of mediation through managerial attention distortion arising from the
increased number of lawsuits filed against the acquirers as a result of their M&A activities. From these
findings we argue that the causality from the excessive use of M&A investment technology to the firm failure
33
M.M. Rahaman
Corporate Failure
is channeled through aggravated business risk along with managerial cognitive bias and attention distortion.
However, the mediation process seems to be stronger through the behavioral channel than the underlying
business risk channel.
Finally, we study the capital market reaction to the managerial acquisitiveness and find evidence of capital
market myopia in incorporating the failure augmenting attributes of the excessive acquisitiveness into the
stock returns at the time of the bid announcement. In our sample, the market, on average, punishes
excessive acquisitiveness at the time of the bid announcement but it does not do so at all quantiles of the
conditional distribution of acquirers’ cumulative abnormal return from the announcement events. However,
despite investors’ myopia, the external corporate control market eventually reins in the excessive acquirers
by turning them into future targets of acquisition.
To understand the yet unresolved question of whether managerial actions can precipitate corporate failure,
we take a very narrow and specialized approach by focusing on a particular action that has random value
implications for firms and by limiting our investigation to a particular sample that uses that investment
technology. This strategy helps us to understand the value implication of M&A while shedding light on the
debate of whether managers of failed businesses are villains or scapegoats. However, we do not claim to have
fully resolved the debate about why firms fail and who to blame for failure. Rather, it is a step forward
towards understanding the complex interplay of forces that bring down a firm from the zenith of miracle to
the abyss of debacle.
34
M.M. Rahaman
Corporate Failure
References
[1] Agrawal, A.. and Jaffe, J.F.(1999), The Post Merger Performance Puzzle, Advances in Mergers and
Acquisitions, JAI Press
[2] Aghion, P. and Howitt, P. (1992), A Model of Growth through Creative Destruction, Econometrica, 60,
323-51.
[3] Altman, E. I. (1968), Financial Ratios, Discriminant Analysis, and the Prediction of Corporate
Bankruptcy, Journal of Finance, 23, 589-609.
[4] Altman, E. I.(2000), Predicting Financial Distress of Companies: Revisiting the Z-Score and Zeta Models,
Stern School of Business, New York University.
[5] Andrade, G., Mitchell, M., and Stafford, E. (2001) New Evidence and Perspectives on Mergers Journal
of Economic Perspectives 15, 103-120.
[6] Ang, J. and Cheng, Y. (2003), Direct Evidence on the Market-Driven Acquisitions Theory, Unpublished
working paper, Florida State University.
[7] Asquith, P., Gertner, R. and Scharfstein, D. (1994), Anatomy of Financial Distress: An Examination of
Junk Bond Issuers, Quarterly Journal of Economics, 109, 625-658.
[8] Baron, R. M. and Kenny, D. A. (1986), The Moderator-Mediator Variable Distinction in Social Psychological Research: Conceptual, Strategic and Statistical Considerations, Journal of Personality and Social
Psychology, 51, 1173-1182.
[9] Bhattacharjee, A., Higson, C., Holly, S. and Kattuman, P. (2002), Macro Economic Instability and
Business Exit: Determinants of Failures and Acquisitions of Large UK Firms, DAE Working Paper 0206,
Department of Applied Economics, University of Cambridge.
[10] Cabral, L.M.B. (1993), Experience Advantages and Entry Dynamics, Journal of Economic Theory, 59,
403-416.
[11] Coase, R. (1937), The Nature of Firm, Economica, 4, 386-405.
[12] Conlisk, J. (1996), Why Bounded Rationality, Journal of Economics Literature, 29, 233-250
[13] Cooley, T.F. and Quadrini, V. (2001), Financial Markets and Firm Dynamics, American Economics
Review, 90, 5, 1286-1310.
[14] Cox, D.R. (1972), Regression Models and Life Tables (with discussion), Journal of the Royal Statistical
Society, Series B 34, 187-220.
[15] Davydenko, S. and Rahaman, M.M. (2006), Managerial Discretion and Value Destruction in Financial
Distress, Rotman School of Management, University of Toronto.
[16] Denis, D.J. and Denis, D.K. (1995), Performance Changes Following Top Management Dismissals,
Journal of Finance, 50, 1029-1057.
35
M.M. Rahaman
Corporate Failure
[17] Dong, M., Hirshleifer, D., Richardson, S. and Teoh, S.H. (2003), Does Investor Misvaluation Drive the
Takeover Market?, Unpublished working paper, Ohio State University.
[18] Dunne, T., Roberts, M.J. and Samuelson, L. (1989), The Growth and Failure of U.S. Manufacturing
Plants, Quarterly Journal of Economics 104, 671-698.
[19] Ericson, R. and Pakes, A. (1992), An Alternative Theory of Firm and Industry Dynamics, Yale Cowles
Foundation Discussion Paper 1041.
[20] Evans, D.S. (1987), The Relationship Between Firm Growth, Size, and Age: Estimates for 100 Manufacturing Industries, Journal of Industrial Economics, 35, 567-581.
[21] Fama, E. (1998), Market efficiency, Long-term Returns, and Behavioural Finance, Journal of Financial
Economics 49, 283-306.
[22] Fama, E., French, K. (1997), Industry Costs of Equity, Journal of Financial Economics 43, 153-193.
[23] Franks, J., Harris, R. and Titman, S. (1991), The Post-merger Share-Price Performance of Acquiring
Firms, Journal of Financial Economics, 29, 81-96.
[24] Fries, S., Miller, M. and Perraudin, W. (1997), Debt in Industry Equilibrium, Review of Financial
Studies, 10(1), 39-67.
[25] Gompers, P. A., Ishii, J.L. and Metrick, A. (2003), Corporate Governance and Equity Prices, Quarterly
Journal of Economics, 118,1, 107-155.
[26] Gort, M. (1969), An Economic Disturbance Theory of Mergers, Quarterly Journal of Economics 83,
623-642.
[27] Hardin, J.W., Schmeidiche, H. and Carroll, R.J. (2003), Instrumental Variables, Bootstrapping, and
Generalized Linear Models, Stata Journal, 3, 4, 351-360
[28] Harford, J (2005), What Drives Merger Waves?, Journal of Financial Economics, 77, 529-560.
[29] Harford, J. (1999), Corporate Cash Reserves and Acquisitions, Journal of Finance 54, 1969-1997.
[30] Heaton, J.B. (2002), Managerial Optimism and Corporate Finance, Financial Management, 31, 33-45
[31] Hirshleifer, D. and Thakor, A.V. (1992), Managerial Conservatism, Project Choice, and Debt, Review
of Financial Studies, 5,3, 437-470.
[32] Hopenhayn, H.A. (1992), Entry, Exit, and Firm Dynamics in Long Run Equilibrium, Econometrica, 60,
1127-1150.
[33] Jensen, M. C. (1993), The Modern Industrial Revolution, Exit, and the Failure of Internal Control
Systems , Journal of Finance, 48, 3, 831-880
[34] Jensen, M.C. (1986), Agency Costs of Free Cash Flow, Corporate Finance, and Takeovers, American
Economic Review, 76, 323-329.
[35] Jovanovic, B. (1982), Selection and the Evolution of Industry, Econometrica, 50, 649-70
36
M.M. Rahaman
Corporate Failure
[36] Jovanovic, B. and Rousseau, P.L. (2001), Mergers and Technological Change: 1885-1998, Department
of Economics, Vanderbilt University, Working Paper No. 01-W16.
[37] Jovanovic, B. and Rousseau, P.L. (2002), The Q-Theory of Merger, American Economics Review, 92,
198-204.
[38] Judd, C. M. and Kenny, D. A. (1981), Process analysis: Estimating mediation in treatment evaluations,
Evaluation Review, 5, 602-619.
[39] Khanna, N. and Poulson, A. (1995), Managers of Financially Distressed Firms: Villains or Scapegoats?,
Journal of Finance, 50, 919-940.
[40] Koenker R and Hallock K (2001), Quantile regression, Journal of Economic Perspectives, 15, 14356.
[41] Lang, L. and Stulz, R. (1992), Contagion and Competitive Intra-Industry Effects of Bankruptcy Announcements, Journal of Financial Economics, 32,45-60.
[42] Lehn, K. and Zhao, M. (2006), CEO Turnover After Acquisition: Are Bad Bidders Fired?, Journal of
Finance, 61 (4) , 1759-1811.
[43] Levine, P. and Aaronovitch, S. (1981), The Financial Characteristics of Firms and Theories of Merger
Activity, Journal of Industrial Economics, 30(2), 149-172
[44] Loughran, T. and Vijh, A. (1997), Do Long-term Shareholders Benefit from Corporate Acquisitions?,
Journal of Finance 52, 1765-1790.
[45] Lovallo, D. and Kahneman, D. (2003), Delusions of Success: How Optimism Undermines Executive’s
Decision, Harvard Business Review, July, Page 57.
[46] MacKay, P. and Phillips, G.M. (2005), How Does Industry Affect Firm Financial Structure? Review of
Financial Studies, 18(4), 1433-1466.
[47] MacKie-Mason, J. K. (1990), Do Taxes Affect Corporate Financing Decisions?, Journal of Finance,
45(5), 1471-1493.
[48] Maksimovic, V. and J. Zechner, J. (1991), Debt, Agency Costs, and Industry Equilibrium, Journal of
Finance, 46, 1619-1643.
[49] Malmendier, U. and Tate, G. (2003), Who Makes Acquisition? A Test of the Overconfidence Hypothesis,
Working Paper, Stanford University.
[50] Malmendier, U. and Tate, G. (2005), CEO Overconfidence and Corporate Investment, Journal of Finance, 60, 2661-2700
[51] Marshall, M. (1890), Principles of Economics, London: Macmillan and Co., Ltd., 1920. Eighth edition
[52] Meek, G. (1977), Disappointing Marriage: A Study of the Gain from Merger, New York, Cambridge
University Press.
[53] Mitchell, M. and Mulherin, J.H. (1996), The Impact of Industry Shocks on Takeover and Restructuring
Activity, Journal of Financial Economics 41, 193-229.
37
M.M. Rahaman
Corporate Failure
[54] Mitchell, M. and Stafford, E. (2000), Managerial Decisions and Long-term Stock Price Performance,
Journal of Business 73, 287-329.
[55] Mitchell, M.L. and Lehn, K. (1990), Do Bad Bidders Become Good Targets?, The Journal of Political
Economy, 98, 372-398.
[56] Moeller, S.B., Schlingemann, F.P. and Stulz, R.M. (2005), Wealth Destruction on a Massive Scale? A
Study of Acquiring-Firm Returns in the Recent Merger Wave, Journal of Finance, 60, 757-782.
[57] Nelson, R.R. and Winter, S.G. (1978), Forces Generating and Limiting Concentration under Schumpeterian Competition, Bell Journal of Economics, 9, 524-548.
[58] Olly, S. and Pakes, A. (1996), The Dynamics of Productivity in the Telecommunications Equipment
Industry, Econometrica, 6, 1263-1297
[59] Opler, T. and Sherdian, T. (1994), Financial Distress and Corporate Performance, Journal of Finance,
49, 1015-1040.
[60] Rau, P.R. and Vermaelen, T. (1998), Glamour, Value and the Post-acquisition Performance of Acquiring
Firms, Journal of Financial Economics 49, 223-253.
[61] Ravn, M.O. and Uhlig, H. (2002), On Adjusting the Hodrick-Prescott Filter for the Frequency of
Observations, Review of Economics and Statistics, 84, 371-376
[62] Rhodes-Kropf, M., Robinson, D. and Viswanathan, S. (2004), Valuation Waves and Merger Activity:
The Empirical Evidence, Journal of Financial Economics, forthcoming.
[63] Roll, R. (1986), The Hubris Hypothesis of Corporate Takeovers, Journal of Business, 59, 197-216.
[64] Schumpeter, J. (1942), The Process of Creative Destruction, Unwin
[65] Shleifer, A. and Vishny, R. (2003), Stock Market Driven Acquisitions, Journal of Financial Economics
70, 295-311.
[66] Shleifer, A. and Vishny, R.. (1992), Liquidation Values and Debt Capacity, Journal of Finance 32,
337-347.
[67] Shumway, T. (2001), Forecasting Bankruptcy More Accurately: A Simple Hazard Model, Journal of
Business, 74, 101-124.
[68] Silverman, B. (2002), Technological Change and the Evolution of Corporate Innovation: The Structure
of Patenting 1890-1990, Academy of Management Review, 27(2), 311-313
[69] Verter, G. (2002), Timing of Merger Waves, Harvard University, Mimeo.
[70] Williams, J. (1995), Financial and Industrial Structure with Agency, Review of Financial Studies, 8(2),
431-474.
[71] Zmijewski, M. E. (1984), Methodological Issues Related to the Estimation of Financial Distress Prediction Models, Journal of Accounting Research, 22, 59-82.
38
M.M. Rahaman
Corporate Failure
Figure 1: Cumulative Size Growth of the Excessively and Conservatively Acquisitive Firms
This graph compares the cumulative size (book value of total assets) of the excessively acquisitive vis-aFigure 1: Cumulative Size Growth of the Excessively Acquisitive and Conservatively
vis the conservatively acquisitive firms. It clearly shows the aggressive growth strategies pursued by the
Acquisitive
Sample counterparts.
excessively acquisitive firms compared to their
relatively conservative
Excessively Acquisitive Firms
Conservatively Acquisitive Firms
900%
800%
Cumulative Size Growth
700%
600%
500%
400%
300%
200%
100%
Bid1-Bid9
Bid1-Bid8
Bid1-Bid7
39
Bid1-Bid6
Bid1-Bid5
Bid1-Bid4
Bid1-Bid3
Bid1-Bid2
0%
M.M. Rahaman
Corporate Failure
Figure 2: Failure Risk Profiles of the Acquiring and Non-acquiring Sample
This graph compares the failure risk profiles of the acquiring and the non-acquiring sample under various
baseline hazard model specifications. It clearly shows that, on average, acquiring sample has lower failure
risk than the non-acquiring sample.
Weibull Hazard Model
0
Hazard function
.005 .01 .015 .02 .025
Hazard function
.0025 .003 .0035 .004 .0045
Constant Hazard Model
25
50
75
100
125
Analysis time
Acq. Sample
150
175
25
Non-acq. Sample
75
100
125
Analysis time
Acq. Sample
175
Non-acq. Sample
Hazard function
.005
.01
0
0
Hazard function
.005
.01
150
Log-normal Hazard Model
.015
Log-logistic Hazard Model
.015
50
25
50
75
100
125
Analysis time
Acq. Sample
150
175
25
Non-acq. Sample
50
75
100
125
Analysis time
Acq. Sample
40
150
175
Non-acq. Sample
M.M. Rahaman
Corporate Failure
Figure 3: Excessive Use of M&A and Shift in the Failure Risk Profiles
This graph compares the failure risk profiles of the acquiring and the non-acquiring sample under various
baseline hazard model specifications. It also shows how the failure risk profile of the acquiring sample changes
as the firms in the acquiring sample become more and more aggressively acquisitive. From the graph it is
obvious that, all else equal, the more aggressive the acquiring sample becomes the more likely it is that they
are going to fail more often than the non-acquiring sample.
0
Hazard function
.01 .02 .03 .04 .05
Weibull Hazard Model
Hazard function
.002 .004 .006 .008 .01
Constant Hazard Model
25
50
75
100
125
Analysis time
150
175
25
50
75
100
125
Analysis time
150
175
Acq. Sample
Non-acq. Sample
Acq. Sample
Non-acq. Sample
excess_acq=0
excess_acq=1
excess_acq=0
excess_acq=1
excess_acq=2
excess_acq=3
excess_acq=2
excess_acq=3
Log-normal Hazard Model
0
0
Hazard function
.005 .01 .015 .02
Hazard function
.005 .01 .015 .02
Log-logistic Hazard Model
25
50
75
100
125
Analysis time
150
175
25
50
75
100
125
Analysis time
150
175
Acq. Sample
Non-acq. Sample
Acq. Sample
Non-acq. Sample
excess_acq=0
excess_acq=1
excess_acq=0
excess_acq=1
excess_acq=2
excess_acq=3
excess_acq=2
excess_acq=3
41
M.M. Rahaman
Corporate Failure
Figure 4: Probability of Mediation by the Mediating Instrument
0
This graph shows the bootstrap distribution of percentage changes β−β
× 100 in the ‘Total Effect’ of
β
excessive acquisitiveness
as of
a result
of mediation
through
cognitive
bias and theAcquisitiveness
sigma
Figure 8: Bootstrap
Distribution
Percentage
Change
inthe
themanagerial
Total Effects
of Excessive
channels. It clearly shows that ‘Total Effect’ decreases with 100% of the time using the managerial cognitive
β −β /
channel whereas
‘TotalManagerial
Effect’ declinesCognitive
only 90% of the
time
using
the sigma
channel. given
In otherby
words,
due tobias
Mediation
through
Bias
and
Sigma
Channels
×100
β
the mediation process seems to be stronger through the behavioral channel than the risk channel.
(
.9
.8
.7
.6
.5
.4
.3
.2
.1
0
0
.1
.2
.3
.4
.5
.6
.7
.8
.9
1
Sigma Channel
1
Mgt. Cognitive Bias Channel
-13
-12
-11
-10
-9
-8
-7
% Change
-6
-5
-4
-3
-6
42
-5
-4
-3
-2
% Change
-1
0
1
2
)
M.M. Rahaman
Corporate Failure
Figure 5: Conditional Quantile Functions of Excessive Acquisitiveness Measure
This graph shows the effect of excessive acquisitiveness on the various conditional quantile functions of the
FigureCumulative
9: EffectsAbnormal
of Managerial
Excessive
on Various
Quantiles
Conditional
Return (CAR)
of theAcquisitiveness
acquirers from the M&A
announcement
events.ofItthe
shows
that
capital market
does
not
always
react
negatively
to
the
acquirer’s
excessive
M&A
behavior
even
though
this
Distribution of Cumulative Abnormal Return (CAR) of Acquirers
-.04
-.02
0
.02
action eventually augments the conditional failure risk of the firm.
(+) CAR Region
-.06
(-) CAR Region
.1
.2
.3
.4
.5
Quantile
Lower/Upper Bound
43
.6
.7
.8
Excessive Acq.
.9
M.M. Rahaman
Corporate Failure
Figure 6: Conditional Quantile Functions of the Mediating Instruments
-.01
-1
0
-.005
1
0
2
3
.005
This graph shows the effects of various mediating instruments on the various conditional quantile functions of
theFigure
Cumulative
Abnormal
ReturnSigma,
(CAR)Mgt.
of the
acquirers
from Allocation
the M&A announcement
events. Itof
captures
10: Effects
of BRISK,
Bias,
and Attn.
on Various Quantiles
the
that there seem
to
be
inconsistencies
in
terms
of
how
the
capital
market
reacts
to
these
various
mediating
Conditional Distribution of Cumulative Abnormal Return (CAR) of Acquirers
instruments and how these instruments eventually affect the conditional failure risk of the firm.
.1
.2
.3
.4
.5
Quantile
.6
.7
.9
.1
.2
.3
.4
.5
Quantile
.6
.7
Lower/Upper Bound
.8
.9
Sigma
-.02
-.01
0
-.01
.01
0
.02
.01
BRISK
.03
Lower/Upper Bound
.8
.1
.2
.3
.4
.5
Quantile
Lower/Upper Bound
.6
.7
.8
.9
.1
Mgt. Bias
.2
.3
.4
.5
Quantile
Lower/Upper Bound
44
15
.6
.7
.8
Attn. Allocation
.9
Table 1: Bidders and Target Sample Characteristics
Size and Performance Measures
log(Tot. Assets)
log(Mkt. Value of Assets)
Net Income/Tot. Assets
EBITDA/Tot. Assets
Market-to-Book
Leverage and Liquidity Measures
Tot. Liab./Tot. Assets
Book Leverage
Market Leverage
Cash/Tot. Assets
Cash/Curr. Liab.
Curr. Assets/Curr. Liab.
St. Debt/Tot. Liab.
Lt. Debt/Tot. Liab.
PPE/Tot. Assets
Deal Characteristics
Tot. Deal Announcement
Num. of qtr. survived
Cumulative Abnormal Return
6.22
6.69
0.01
0.03
1.40
0.56
0.56
0.38
0.06
0.30
1.93
0.04
0.29
0.18
3.00
39.00
0.00
0.56
0.56
0.41
0.13
1.10
2.67
0.09
0.33
0.24
6.00
47.63
0.01
Medain
(2)
6.29
6.76
0.00
0.03
1.97
Mean
(1)
45
10779
10779
63613
61221
61255
60510
60960
47860
47661
56076
60683
58500
61289
60510
61131
51641
60510
N
Bidding Firms
10779
10779
10779
10352
10368
10248
10329
8445
8367
9998
10321
10137
10374
10248
10359
9403
10248
Num.
Firms
1.67
36.16
0.14
0.59
0.59
0.47
0.14
1.29
2.79
0.10
0.30
0.27
5.83
6.23
-0.01
0.01
1.83
Mean
(3)
1.00
30.00
0.10
0.59
0.59
0.46
0.06
0.29
2.01
0.05
0.27
0.23
5.74
6.04
0.00
0.02
1.22
Median
(4)
2124
3582
2124
2953
2951
2899
2933
2388
2382
2792
2926
2884
2962
2899
2928
2485
2899
N
Target Firms
2124
2124
2124
1774
1771
1747
1762
1309
1303
1669
1754
1725
1782
1747
1780
1540
1747
Num.
Firms
4.33***
11.47***
-0.13***
-0.03***
-0.03***
-0.05***
-0.01***
-0.19**
-0.13
-0.01
0.03***
-0.03***
0.45***
0.53***
0.02***
0.01***
0.14***
(1-3)
7.04
7.06
22.30
3.05
3.02
4.98
2.17
1.71
1.12
1.46
2.55
2.88
5.08
5.60
5.04
5.54
2.18
Absolute
t-stat
2.00***
9.00***
-0.09***
-0.03***
-0.03***
-0.08***
-0.00***
0.00
-0.09***
-0.01***
0.02***
-0.06***
0.48***
0.65***
0.01***
0.01***
0.18***
(2-4)
Difference
10.12
21.28
38.45
3.63
3.49
8.63
2.02
0.12
2.55
6.28
2.67
7.85
9.22
15.26
33.23
15.31
14.51
Absolute
t-stat
This table reports the differential firm characteristics of the bidding and the target firms at the time of bid announcement. Among the size and performance measures, total
assets is defined to be the total book value of firms assets at the end of the fiscal quarter in which firm announces the bid. Market value is defined to be the sum of market value
of equity and the book value of debt. Net income is earning after all interest and tax payment while EBITDA is earning before interest, tax, depreciation and amortization.
Market-to-book ratio is calculated by diving the market value of firm’s assets with its book value. Among the leverage and liquidity measures, total liabilities measure all
outstanding liabilities owed to outsiders other than the shareholders of the firm. Book leverage is defined to be the ratio of firm’s total outstanding short and long term debt
to book value of total assets whereas market leverage is defined to be the ratio of total outstanding short and long term debt to the market value of firm’s total assets. Cash
is defined to be the value of cash and other cash equivalent marketable securities, current assets are cash plus account receivables, current liabilities are short term debt plus
account payable, short term debts are debt obligations maturing within one year while long term debts are debt obligations maturing in two years or more time. PPE is defined
to be the net book value of firm’s Properties, Plants, and Equipments. And finally, cumulative abnormal return is calculated against a market model around a 3-day event
window period consisting of the announcement date, one trading day prior to the announcement date and one trading day after the announcement date.
M.M. Rahaman
Corporate Failure
Table 2: Excessive and Conservative Acquisitive Sample Characteristics
Size and Performance Measures
log(Tot. Assets)
log(Mkt. Value of Assets)
Net Income/Tot. Assets
EBITDA/Tot. Assets
Market-to-Book
Leverage and Liquidity Measures
Tot. Liab./Tot. Assets
Book Leverage
Market Leverage
Cash/Tot. Assets
Cash/Curr. Liab.
Curr. Assets/Curr. Liab.
St. Debt/Tot. Liab.
Lt. Debt/Tot. Liab.
PPE/Tot. Assets
Deal Characteristics
Cumulative Abnormal Return
46
6.36
6.83
0.01
0.03
1.40
0.56
0.56
0.38
0.05
0.28
1.91
0.04
0.30
0.17
0.00
0.56
0.56
0.42
0.12
1.01
2.55
0.09
0.33
0.24
0.01
Medain
(2)
6.41
6.88
0.00
0.03
1.94
Mean
(1)
53494
51564
51597
51019
51344
40175
39977
47049
51108
49161
51630
51019
51490
43139
51019
N
8651
8297
8313
8227
8275
6789
6717
7998
8273
8116
8319
8227
8303
7462
8227
Num.
Firms
Excessively Acquisitive Firms
0.02
0.54
0.54
0.40
0.16
1.61
3.26
0.08
0.28
0.23
5.65
6.12
0.00
0.02
2.12
Mean
(3)
0.01
0.53
0.53
0.35
0.07
0.40
2.09
0.03
0.21
0.15
5.52
6.02
0.01
0.03
1.39
Median
(4)
10087
9628
9629
9464
9587
7656
7655
8998
9546
9310
9630
9464
9612
8477
9464
N
5227
5013
5013
4925
4997
3998
3991
4767
4983
4858
5014
4925
5008
4514
4925
Num.
Firms
Conservatively Acquisitive Firms
-0.01***
0.03***
0.03***
0.02***
-0.03***
-0.61***
-0.71***
0.01***
0.05***
0.01**
0.75***
0.76***
0.00***
0.01***
-0.17***
(1-3)
6.80
5.14
5.17
3.21
10.49
5.14
4.37
3.69
8.37
2.38
14.27
14.08
3.64
6.81
4.00
Absolute
t-stat
-0.01***
0.03***
0.03***
0.03***
-0.01***
-0.11***
-0.18***
0.01***
0.08***
0.02***
0.84***
0.81***
0.00***
0.00***
0.02*
(2-4)
Difference
9.25
9.27
9.20
6.06
10.65
12.83
9.36
8.85
14.95
7.35
31.10
26.59
8.26
10.66
1.67
Absolute
t-stat
This table reports the differential firm characteristics of the excessively and the conservatively acquisitive sample firms at the time of bid announcement. Among the size and
performance measures, total assets is defined to be the total book value of firms assets at the end of the fiscal quarter in which firm announces the bid. Market value is defined
to be the sum of market value of equity and the book value of debt. Net income is earning after all interest and tax payment while EBITDA is earning before interest, tax,
depreciation and amortization. Market-to-book ratio is calculated by diving the market value of firm’s assets with its book value. Among the leverage and liquidity measures,
total liabilities measure all outstanding liabilities owed to outsiders other than the shareholders of the firm. Book leverage is defined to be the ratio of firm’s total outstanding
short and long term debt to book value of total assets whereas market leverage is defined to be the ratio of total outstanding short and long term debt to the market value of
firm’s total assets. Cash is defined to be the value of cash and other cash equivalent marketable securities, current assets are cash plus account receivables, current liabilities
are short term debt plus account payable, short term debts are debt obligations maturing within one year while long term debts are debt obligations maturing in two years or
more time. PPE is defined to be the net book value of firm’s Properties, Plants, and Equipments. And finally, cumulative abnormal return is calculated against a market model
around a 3-day event window period consisting of the announcement date, one trading day prior to the announcement date and one trading day after the announcement date.
M.M. Rahaman
Corporate Failure
47
Table 3: What Drives Firm-Level M&A Propensity?
(2)
(3)
(4)
(5)
(6)
(7)
(8)
(9)
(10)
(11)
(12)
0.2631*** 0.2614*** 0.2635*** 0.2637*** 0.2634*** 0.2786*** 0.2617*** 0.2632*** 0.2636*** 0.2635*** 0.2631*** 0.2396***
[36.51]
[32.93]
[36.55]
[36.03]
[36.05]
[33.98]
[36.28]
[36.51]
[36.56]
[36.56]
[36.51]
[31.46]
Net Income/Tot. Assets
0.4299**
0.6827***
0.4271**
0.4706**
0.4706**
0.4821***
0.4624**
0.4301**
0.4215**
0.4179**
0.4295**
0.6756***
[2.41]
[4.61]
[2.39]
[2.41]
[2.41]
[3.55]
[2.44]
[2.40]
[2.41]
[2.41]
[2.40]
[3.28]
Tot. Liab./Tot. Assets
-0.9870*** -1.0775*** -0.9891*** -0.9895*** -0.9887*** -0.9453*** -0.9894*** -0.9871*** -0.9884*** -0.9874*** -0.9870*** -0.9316***
[15.47]
[14.97]
[15.49]
[15.36]
[15.35]
[13.44]
[15.54]
[15.47]
[15.50]
[15.49]
[15.47]
[14.95]
Cash/Tot. Assets
-0.2880*** -0.1547** -0.2890*** -0.2954*** -0.2966*** -0.1492* -0.3023*** -0.2884*** -0.2886*** -0.2875*** -0.2883*** -0.2856***
[3.97]
[2.17]
[3.99]
[4.04]
[4.06]
[1.92]
[4.18]
[3.98]
[3.98]
[3.97]
[3.98]
[3.97]
Lt. Debt/Tot. Assets
0.6680*** 0.8049*** 0.6699*** 0.6666*** 0.6668*** 0.7279*** 0.6633*** 0.6681*** 0.6673*** 0.6672*** 0.6679*** 0.6238***
[11.72]
[12.58]
[11.76]
[11.61]
[11.62]
[11.51]
[11.67]
[11.72]
[11.71]
[11.71]
[11.72]
[11.15]
PPE/Tot. Assets
-1.0641*** -1.1801*** -1.0646*** -1.0639*** -1.0634*** -1.1931*** -1.0324*** -1.0645*** -1.0659*** -1.0654*** -1.0641*** -0.8517***
[12.44]
[12.98]
[12.45]
[12.36]
[12.34]
[12.63]
[12.11]
[12.44]
[12.45]
[12.45]
[12.44]
[10.05]
Market-to-Book
0.0434***
0.0434*** 0.0434*** 0.0434*** 0.0394*** 0.0413*** 0.0434*** 0.0430*** 0.0430*** 0.0434*** 0.0382***
[5.20]
[5.21]
[5.16]
[5.16]
[4.66]
[5.02]
[5.20]
[5.16]
[5.17]
[5.20]
[4.75]
Change in TFP
0.0773***
[3.07]
Deregulation Dummy
0.1554***
[4.25]
Ind. Demand Shock
-0.0001***
[4.64]
Ind. Supply Shock
-0.0002***
[2.81]
Ind. Tech. Shock
0.0525***
[6.70]
Ind. Merger Wave Dummy
0.1070***
[5.98]
Agg. Demand Shock
0.2396***
[3.19]
Agg. Supply Shock
2.4408***
[5.82]
Agg. Equity Shock
-0.0569***
[5.65]
Agg. Debt Shock
0.0096***
[4.10]
Agg. Equity Momentum
0.2982***
[21.13]
Constant
-2.9692*** -2.8365*** -2.9706*** -2.9584*** -2.9589*** -2.9925*** -2.9105*** -2.9694*** -2.9684*** -2.9667*** -2.9692*** -4.8050***
[11.14]
[10.35]
[11.14]
[11.04]
[11.05]
[10.74]
[10.99]
[11.14]
[11.13]
[11.13]
[11.14]
[17.75]
Ind. Fixed Effect
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Year Fixed Effect
No
No
No
No
No
No
No
No
No
No
No
No
N
418346
354539
418346
415000
415259
326968
410347
418346
418346
418346
418346
418346
Num. of Firms
10439
8923
10439
10355
10356
8759
10439
10439
10439
10439
10439
10439
Pseudo-R2
0.06
0.06
0.06
0.06
0.06
0.06
0.06
0.06
0.06
0.06
0.06
0.06
2
Wald-χ
2375.90
2179.23
2378.95
2374.08
2357.04
2109.26
2369.99
2382.67
2391.64
2401.98
2385.41
3438.53
log(Tot. Assets)
(1)
0.2719***
[31.06]
0.5530***
[3.70]
-0.9535***
[13.93]
-0.1406*
[1.87]
0.6836***
[11.22]
-0.9713***
[10.61]
0.0264***
[3.58]
(14)
0.1887***
0.1036*
[2.97]
[1.67]
-0.0001*
-0.0000
[1.72]
[0.06]
-0.0001
-0.0002
[0.56]
[1.59]
0.0523***
0.0185**
[6.54]
[2.31]
0.2155***
0.1036***
[9.69]
[4.58]
0.3173***
0.0485
[3.64]
[0.49]
3.6973***
2.9368***
[5.54]
[3.61]
-0.1177***
0.0035
[10.92]
[0.23]
0.0007
0.0078
[0.18]
[1.23]
0.3636***
6.3454***
[22.58]
[33.04]
-5.1432*** -34.6772***
[18.45]
[34.79]
Yes
Yes
No
Yes
316753
316728
8677
8677
0.07
0.09
3161.45
4562.34
0.2514***
[28.70]
0.8449***
[6.06]
-0.9159***
[13.62]
-0.2051***
[2.72]
0.6794***
[10.97]
-0.9347***
[10.12]
0.0294***
[3.98]
(13)
This table reports the estimates from a multi-period logit model to ascertain the determinants of M&A propensity of the sample firms. The dependent variable in the regression is 1 if during the current
fiscal quarter firm makes an M&A bid, otherwise it it 0. Total assets is defined to be the book value of firms assets, net income is income from operation after all taxes and interest payment, total
liabilities are all obligations due to outsiders other than the shareholders of the firms, cash are the value of cash plus other marketable securities hold by the firm, long term debts are debt obligations
due in two or more years time, PPE is defined to be the net book value of firm’s Plant, Property and Equipments, and market value is calculated as market value of equity plus the book value of firm’s
debt. TFP stands for total factor productivity of the firm estimated following the methodology developed by Olly and Pakes (1996). Industry classifications are based on Fama-French (1997). We use
the total industry net sales from the quarterly COMPUSTAT data item 2 as proxy for industry demand and the total industry cost of good sold from the quarterly COMPUSTAT data item 30 as proxy
for industry supply. We also collect information about all patents for the period of 1963-2002 from the NBER patent database and convert the assigned technology class of each of those patents into
international patent class using the methodology developed by Silverman (2002). From the international patent class we covert them back into 1987 Standard Industry Classifications and assign the
patents by grant year to each of our 49 Fama and French (1997) industries. We then decompose the series into trend and irregular components using the Hodrick-Prescott (HP) filter. After decomposing
the trend and irregular components of the series, we calculate series instability by estimating the acceleration (change of change) of the irregular component. We use major deregulatory initiatives
during the sample period as a proxy for regulatory shocks. We use the quarterly real GDP data from the Federal Reserve Bank of St. Louis as proxy for aggregate demand and the real price of crude
petroleum in the U.S. from the U.S. Energy Information Administration as a proxy for aggregate supply. Utilizing the HP filter, we then calculate the aggregate demand and supply shock. As a proxy
for aggregate equity and debt market instability we apply the HP filter on the Dow Jones Industrial average and bank prime lending rate, respectively. To capture the momentum in equity market,
we apply the HP filter on S&P 500 index and use the smoothed trend portion of series as our proxy for momentum in aggregate equity market. We also construct measures of industry merger wave
utilizing the X-12-ARIMA, a seasonal adjustment software produced and maintained by the U.S. Census Bureau. We detail the construction of all variables in the data section of the paper. Robust z
statistics are given in brackets and “*” denotes significance at 10%; “**” denotes significance at 5%; “***” denotes significance at 1% level.
Table 4: What Drives Firm-Level M&A Propensity? Decomposition of the Economic Shocks
This table decomposes the economic shocks into positive and negative components and reports the estimates from a multi-period logit
model. The dependent variable in the regression is 1 if during the current fiscal quarter firm makes an M&A bid, otherwise it is 0. Total
assets is defined to be the book value of firms assets, net income is income from operation after all taxes and interest payment, total
liabilities are all obligations due to outsiders other than the shareholders of the firms, cash are the value of cash plus other marketable
securities hold by the firm, long term debts are debt obligations due in two or more years time, PPE is defined to be the net book value
of firm’s Plant, Property and Equipments, and market value is calculated as market value of equity plus the book value of firm’s debt.
TFP stands for total factor productivity of the firm estimated following the methodology developed by Olly and Pakes (1996). Industry
classifications are based on Fama-French (1997). We use the total industry net sales from the quarterly COMPUSTAT data item 2 as
proxy for industry demand and the total industry cost of good sold from the quarterly COMPUSTAT data item 30 as proxy for industry
supply. We also collect information about all patents for the period of 1963-2002 from the NBER patent database and convert the
assigned technology class of each of those patents into international patent class using the methodology developed by Silverman (2002).
From the international patent class we covert them back into 1987 Standard Industry Classifications and assign the patents by grant
year to each of our 49 Fama and French (1997) industries. We then decompose the series into trend and irregular components using
the Hodrick-Prescott (HP) filter. After decomposing the trend and irregular components of the series, we calculate series instability by
estimating the acceleration (change of change) of the irregular component. We use major deregulatory initiatives during the sample
period as a proxy for regulatory shocks. We use the quarterly real GDP data from the Federal Reserve Bank of St. Louis as proxy for
aggregate demand and the real price of crude petroleum in the U.S. from the U.S. Energy Information Administration as a proxy for
aggregate supply. Utilizing the HP filter, we then calculate the aggregate demand and supply shock. As a proxy for aggregate equity
and debt market instability we apply the HP filter on the Dow Jones Industrial average and bank prime lending rate, respectively. If
the actual series has been above the HP filtered trend components for at least three consecutive periods then we treat the shock as
positive and if the actual series has been below the trend components for three consecutive periods then we treat the shock as negative.
To capture the momentum in equity market, we apply the HP filter on S&P 500 index and use the smoothed trend portion of series as
our proxy for momentum in aggregate equity market. We also construct measures of industry merger wave utilizing the X-12-ARIMA,
a seasonal adjustment software produced and maintained by the U.S. Census Bureau. We detail the construction of all variables in the
data section of the paper. Robust z statistics are given in brackets and “*” denotes significance at 10%; “**” denotes significance at
5%; “***” denotes significance at 1% level.
log(Tot. Assets)
(1)
(2)
(3)
(4)
(5)
(6)
(7)
0.2627***
[36.16]
0.4713**
[2.42]
-0.9831***
[15.39]
-0.2949***
[4.08]
0.6588***
[11.63]
-1.0480***
[12.36]
0.0426***
[5.11]
0.0033***
[9.27]
-0.0018***
[5.51]
0.2632***
[36.06]
0.4704**
[2.41]
-0.9893***
[15.37]
-0.2965***
[4.06]
0.6668***
[11.63]
-1.0618***
[12.32]
0.0432***
[5.15]
0.2784***
[34.02]
0.4803***
[3.54]
-0.9451***
[13.44]
-0.1486*
[1.92]
0.7276***
[11.50]
-1.1924***
[12.63]
0.0391***
[4.64]
0.2633***
[36.51]
0.4307**
[2.41]
-0.9874***
[15.48]
-0.2882***
[3.97]
0.6686***
[11.73]
-1.0647***
[12.44]
0.0435***
[5.21]
0.2645***
[36.46]
0.4236**
[2.40]
-0.9905***
[15.51]
-0.2854***
[3.93]
0.6668***
[11.69]
-1.0705***
[12.45]
0.0430***
[5.16]
0.2638***
[36.54]
0.4218**
[2.40]
-0.9872***
[15.48]
-0.2853***
[3.94]
0.6664***
[11.70]
-1.0665***
[12.46]
0.0427***
[5.14]
0.2631***
[36.49]
0.4269**
[2.41]
-0.9878***
[15.48]
-0.2879***
[3.97]
0.6681***
[11.73]
-1.0627***
[12.42]
0.0433***
[5.19]
(8)
0.2509***
[28.89]
Net Income/Tot. Assets
0.8438***
[6.15]
Tot. Liab./Tot. Assets
-0.9123***
[13.71]
Cash/Tot. Assets
-0.1997***
[2.68]
Lt. Debt/Tot. Assets
0.6726***
[11.03]
PPE/Tot. Assets
-0.9223***
[10.16]
Market-to-Book
0.0288***
[3.98]
Pos. Ind. Demand Shock
0.0027***
[7.09]
Neg. Ind. Demane Shock
-0.0022***
[5.00]
Pos. Ind. Supply Shock
0.0022**
0.0042***
[2.17]
[3.03]
Neg. Ind. Supply Shock
-0.0010
0.0004
[0.91]
[0.41]
Pos. Ind. Tech. Shock
-0.0052
0.0354***
[0.36]
[2.63]
Neg. Ind. Tech. Shock
0.0020
0.0476**
[0.10]
[2.05]
Pos. Agg. Demand Shock
0.5565***
0.1588
[5.08]
[1.43]
Neg. Agg. Demand Shock
0.0105
-0.1019
[0.07]
[0.43]
Pos. Agg. Supply Shock
-0.7101
-3.1745**
[0.60]
[2.09]
Neg. Agg. Supply Shock
9.2105***
4.8126***
[11.52]
[4.86]
Pos. Agg. Equity Shock
-0.0732***
-0.0292*
[4.23]
[1.83]
Neg. Agg. Equity Shock
-0.1249***
-0.3047***
[5.56]
[11.46]
Pos. Agg. Debt Shock
0.0107*
0.0384***
[1.69]
[4.75]
Neg. Agg. Debt Shock
-0.1104*** -0.1216***
[6.94]
[6.28]
Deregulation Dummy
0.2054***
[3.28]
Ind Merger Wave Dummy
0.2164***
[9.85]
Agg. Equity Momentum
0.3537***
[22.00]
Constant
-2.9636*** -2.9601*** -2.9913*** -2.9695*** -2.9804*** -2.9665*** -2.9627*** -5.0925***
[11.23]
[11.12]
[10.74]
[11.14]
[11.16]
[11.14]
[11.12]
[18.81]
Ind. Fixed Effect
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Year Fixed Effect
No
No
No
No
No
No
No
No
N
415000
415259
326968
418346
418346
418346
418346
316753
Num. of Firms
10355
10356
8759
10439
10439
10439
10439
8677
Pseudo-R2
0.06
0.06
0.06
0.06
0.06
0.06
0.06
0.07
2
Wald-χ
2417.18
2350.48
2068.55
2388.21
2544.65
2451.05
2444.72
3184.62
48
(9)
0.2708***
[31.24]
0.5435***
[3.66]
-0.9483***
[14.00]
-0.1453*
[1.96]
0.6750***
[11.25]
-0.9546***
[10.58]
0.0258***
[3.57]
0.0030***
[8.31]
-0.0014***
[3.14]
0.0042***
[2.97]
0.0013
[1.21]
0.0486***
[3.56]
0.0057
[0.23]
0.1491
[1.17]
0.0281
[0.12]
13.7910***
[6.64]
1.4380
[1.24]
0.0097
[0.53]
0.6126***
[8.89]
0.0554***
[5.00]
-0.1296***
[5.66]
0.0923
[1.50]
0.1038***
[4.65]
6.5266***
[33.94]
-35.5046***
[35.67]
Yes
Yes
316728
8677
0.09
4989.05
M.M. Rahaman
Corporate Failure
Table 5: Why Some Firms Are More Acquisitive Than Others?
This table shows the correlation structure of managerial acquisitiveness with firm’s productivity shocks, investment and acquisition
expenditure, future growth opportunity, and governance proxies. We discuss the construction of the excessive acquisitiveness measure
in detail in the data section of the paper. TFP stands for total factor productivity estimated using the methodology developed by Olly
and Pakes (1996). Optimism driven bid is a dummy variable which equals 1 if the firm announces an acquisition bid even if it receives
a negative productivity shock in that period while growth driven bid is another dummy variable which equals 1 if the firm announces
an acquisition bid when market-to-book ratio is greater than 1. Firm level capital expenditure and acquisition expenditure are from
COMPUSTAT data item 90 and data item 94, respectively. Governance index (G) is from Gompers, Ishii and Metrick (2003). P-values
are given in bracket
Managerial Acquistiveness
Change in TFP
Optimism Driven Bid
Growth Driven Bid
Capital Expenditure
Acquisition Expenditure
Acquirer’s G-Index
Excessive Acq. Sample
Non-Excessive Acq. Sample
0.00670
[0.00]
0.13160
[0.00]
0.28210
[0.00]
0.07090
[0.00]
0.10220
[0.00]
0.04680
[0.00]
-0.00550
[0.00]
-0.07300
[0.00]
-0.14650
[0.00]
-0.00580
[0.00]
-0.01450
[0.00]
-0.02690
[0.00]
49
Table 6: Excessive Use of M&A Investment Technology and Corporate Failure
50
-2.1960**
[2.44]
Yes
Yes
412416
10439
0.14
9803.81
Ind. Dummy
Year Dummy
N
Num. of Firms
Pseudo-R2
Wald-Chi2
-0.4945**
[2.33]
-0.6399***
[6.98]
-0.2080***
[2.96]
0.9312
[1.44]
-0.6792
[0.75]
-0.4256
[0.08]
0.1388
[0.08]
-0.5918***
[3.40]
Constant
Agg. Equity Momentum
Agg. Debt Shock
Agg. Equity Shock
Agg. Supply Shock
Agg. Demand Shock
Ind. Tech. Shock
Ind. Supply Shock
Ind. Demand Shock
Deregulation Dummy
Governance Score
Excessive Acq. IV
Excessive Acq.
Distance to N. Hedge
Market-to-Book
PPE/Tot. Assets
Lt. Debt/Tot. Assets
Cash/Tot.Assets
Tot. Liab./Tot. Assets
Net Income/Tot. Assets
log(Age)
log(Tot. Assets)
(1)
Yes
Yes
408589
10439
0.17
7436.56
-4.8474***
[5.73]
-0.5728***
[2.93]
-0.5057***
[6.18]
-0.3289
[0.81]
0.9249***
[2.90]
-0.3066
[0.34]
-0.5097
[0.10]
0.3889
[0.22]
-0.7134***
[4.05]
2.9351***
[5.59]
(2)
Yes
Yes
408589
10439
0.18
5624.42
-3.3751***
[4.27]
3.1217***
[6.21]
-0.5916***
[2.92]
-0.4664***
[5.87]
-0.1456***
[3.18]
0.9232
[1.62]
-0.1813
[0.21]
-0.6270
[0.12]
0.5073
[0.29]
-0.5964***
[3.43]
(3)
No
No
409080
-1.1064
[1.19]
48.4192***
[3.49]
-1.5172***
[4.97]
-0.4324***
[3.23]
0.8086*
[1.88]
1.4626**
[2.10]
3.9651***
[3.17]
-0.1513
[0.91]
1.6802***
[2.87]
-0.7501***
[5.34]
(4)
Yes
Yes
127394
2661
0.21
-13.203***
[10.27]
-0.033
[1.06]
-0.415***
[6.25]
-0.784***
[6.15]
-0.891***
[3.61]
3.212***
[11.87]
0.472
[0.99]
-1.800***
[4.61]
1.221**
[2.42]
-1.968***
[7.76]
(5)
Yes
Yes
125861
2658
.22
-13.4186***
[10.58]
-0.0297
[0.95]
1.9575***
[7.53]
-0.5104***
[7.51]
-0.7445***
[5.50]
-0.8620***
[3.56]
3.1509***
[11.95]
0.8115*
[1.69]
-1.9001***
[4.84]
1.5018***
[2.87]
-1.9471***
[7.92]
(6)
Yes
Yes
318928
8678
0.14
9213.49
-0.2395
[1.10]
0.0005**
[2.28]
-0.0005
[1.31]
0.0336
[1.09]
1.1507***
[2.84]
0.9461
[0.31]
-0.1523***
[3.07]
-0.0535
[1.38]
1.2905
[1.53]
-8.1319**
[2.26]
-0.4993*
[1.91]
-0.6113***
[5.37]
-0.5216*
[1.93]
0.6704
[0.71]
-1.3080
[1.11]
-0.3927
[0.06]
0.1537
[0.07]
-0.5522***
[3.07]
(7)
Yes
Yes
315595
8678.00
0.17
6908.89
-0.2242
[1.02]
0.0006**
[2.53]
-0.0006
[1.51]
0.0395
[1.32]
1.2745***
[3.17]
1.8302
[0.62]
-0.1329**
[2.56]
-0.0547
[1.28]
0.8935
[0.98]
-8.7534**
[2.12]
-0.5806**
[2.32]
-0.4342***
[4.04]
-1.1425*
[1.94]
0.6979
[0.74]
-0.7847
[0.74]
-0.4458
[0.07]
0.3913
[0.20]
-0.5586***
[3.21]
3.1055***
[5.11]
(8)
Yes
Yes
315595
8678
0.19
5109.62
-0.2085
[0.97]
0.0005**
[2.56]
-0.0006
[1.38]
0.0352
[1.16]
1.1733***
[2.90]
1.2174
[0.40]
-0.1339***
[2.61]
-0.0434
[1.15]
1.0292
[1.24]
-7.9488**
[2.21]
3.3286***
[5.16]
-0.6098**
[2.36]
-0.4150***
[3.83]
-0.1365**
[2.30]
0.7856
[1.27]
-0.6743
[0.57]
-0.7121
[0.11]
0.5790
[0.28]
-0.5551***
[2.62]
(9)
No
No
315740
0.1748
[0.63]
0.0008***
[3.39]
-0.0007
[1.53]
0.0473*
[1.86]
0.7867*
[1.80]
-1.1690
[0.45]
-0.0687
[1.20]
-0.0458*
[1.68]
1.0019***
[14.87]
-7.3653***
[12.66]
28.1494***
[2.80]
-1.1919***
[5.01]
-0.3733***
[2.81]
0.2000
[1.00]
1.0252***
[8.69]
1.8604*
[1.95]
-0.0523
[0.85]
2.0648***
[2.97]
-0.7435***
[9.68]
(10)
Yes
Yes
94806
2341
0.23
-0.028
[0.78]
0.744
[0.93]
0.000
[0.37]
-0.000
[0.07]
0.075
[0.90]
-0.297*
[1.81]
1.345
[0.97]
-29.337**
[1.96]
0.066
[0.33]
0.197
[0.09]
-14.451
[0.90]
-0.372***
[5.17]
-0.751***
[5.30]
-0.704***
[2.87]
3.671***
[10.40]
-0.201
[0.36]
-1.935***
[4.42]
1.434***
[2.63]
-2.082***
[6.38]
(11)
Yes
Yes
93554
2337
0.25
-0.0232
[0.63]
0.8799
[1.09]
0.0003
[0.41]
-0.0001
[0.06]
0.0669
[0.80]
1.3913
[0.97]
-29.0461*
[1.92]
-0.2876*
[1.73]
0.0502
[0.25]
0.2466
[0.11]
-13.6694
[.]
2.3355***
[8.49]
-0.4927***
[6.69]
-0.6957***
[4.48]
-0.6623**
[2.54]
3.5406***
[10.30]
0.1631
[0.29]
-2.0571***
[4.65]
1.6752***
[2.89]
-2.0458***
[6.50]
(12)
This table reports the estimates from a discrete-time hazard model to determine the effect of managerial acquisitiveness on firm’s failure hazard. The dependent variable in the regression is 1 for the
period in which the firm fails and 0 otherwise. The definitions of all the explanatory variables are same as in the previous tables. The Distance to Natural Hedge variable is constructed using the
equation 2 in the paper. Robust z statistics are given in brackets and “*” denotes significance at 10%; “**” denotes significance at 5%; “***” denotes significance at 1% level.
M.M. Rahaman
Corporate Failure
Table 7: Excessive Use of M&A Investment Technology and Corporate Failure: Robustness Tests
51
Ind. Dummy
Year Dummy
Firm Dummy
N
Num. of Firms
Pseudo-R2
Wald-Chi2
R-squared
Adjusted-R2
F-stat
Constant
Agg. Equity Momentum
Agg. Debt Shock
Agg. Equity Shock
Agg. Supply Shock
Agg. Demand Shock
Ind. Tech. Shock
Ind. Supply Shock
Ind. Demand Shock
Deregulation Dummy
Winner
Excessive Acq.
Conserv Acq.
Market-to-Book
PPE/Tot. Assets
Lt. Debt/Tot. Assets
Cash/Tot.Assets
Tot. Liab./Tot. Assets
Net Income/Tot. Assets
log(Age)
log(Tot. Assets)
-3.9188***
[4.65]
Yes
Yes
No
408589
10439
0.18
7521.59
-0.5653***
[2.58]
-0.6191***
[6.62]
-0.2722
[0.77]
1.0557*
[1.89]
-0.2657
[0.29]
-0.5757
[0.11]
0.3842
[0.21]
-0.6813***
[3.30]
-19.4124***
[10.74]
-0.2621
[1.07]
0.0005**
[2.42]
-0.0006
[1.35]
0.0290
[0.91]
1.0830***
[2.62]
0.2183
[0.07]
-0.1594***
[3.05]
-0.0524
[1.19]
1.3759
[1.40]
-10.2596**
[2.30]
Yes
Yes
No
315595
8678
0.18
7208.90
-0.5604**
[2.06]
-0.5713***
[4.72]
-1.0614*
[1.75]
0.8300
[0.80]
-0.7880
[0.73]
-0.5341
[0.09]
0.3939
[0.19]
-0.6043***
[2.96]
-21.8251***
[13.14]
(1)
(2)
Non-linearity
0.07
0.07
39.35
-0.0843
[0.00]
Yes
Yes
Yes
412195
10439
0.07
0.07
56.69
0.0351***
[17.40]
-0.0110***
[23.82]
0.0469***
[27.80]
0.0014
[1.05]
0.0036
[1.44]
-0.0151***
[7.10]
-0.0001
[0.78]
0.0069***
[2.63]
-0.0007**
[2.54]
-0.0018*
[1.92]
0.0000***
[3.14]
-0.0000
[1.63]
0.0002
[1.47]
0.0086***
[2.71]
0.0145
[0.79]
-0.0010**
[2.02]
-0.0001
[1.63]
-0.0037
[1.06]
-0.1060***
[6.37]
Yes
Yes
Yes
318760
8678
0.0314***
[18.39]
-0.0101***
[25.22]
0.0346***
[31.48]
0.0012
[0.94]
0.0041
[1.45]
-0.0148***
[7.25]
-0.0002
[0.87]
0.0043*
[1.90]
-0.0008***
[2.67]
(3)
(4)
Linear Probability Model
with firm FE
-3.3209***
[5.48]
Yes
Yes
No
264077
8135
0.17
3145.97
3.0707***
[34.06]
-0.5300***
[30.39]
-0.4610***
[10.90]
-0.0977*
[1.75]
1.6296***
[13.19]
0.2150
[1.38]
-0.9990***
[8.57]
0.6663***
[4.92]
-0.5927***
[8.69]
0.0318
[0.15]
0.0005**
[2.16]
-0.0005
[1.11]
0.0240
[0.71]
1.3199***
[2.89]
-0.5887
[0.16]
-0.1219**
[2.23]
-0.0468
[0.95]
1.2226*
[1.77]
-9.1438*
[1.86]
Yes
Yes
No
194710
6646
0.18
2776.26
3.2698***
[33.64]
-0.5297***
[27.46]
-0.3947***
[8.39]
-0.1031***
[3.21]
1.6775***
[13.91]
-0.1810
[1.00]
-1.1863***
[9.36]
0.7475***
[5.03]
-0.5361***
[7.18]
(5)
(6)
Focusing on the 1980s
sample only
-3.3643***
[4.26]
Yes
Yes
No
408589
10439
0.18
5729.12
3.1102***
[6.15]
2.3251***
[5.51]
-0.5933***
[2.92]
-0.4664***
[5.84]
-0.1457***
[3.16]
0.9232
[1.63]
-0.1801
[0.21]
-0.6234
[0.12]
0.5135
[0.29]
-0.5971***
[3.44]
3.3168***
[5.12]
2.2976***
[4.39]
-0.2051
[0.95]
0.0006***
[2.64]
-0.0006
[1.43]
0.0350
[1.15]
1.1264***
[2.78]
1.3826
[0.45]
-0.1319***
[2.58]
-0.0426
[1.13]
0.9269
[1.16]
-7.4636**
[2.14]
Yes
Yes
No
315595
8678
0.19
5164.33
-0.6114**
[2.36]
-0.4151***
[3.81]
-0.1369**
[2.27]
0.7858
[1.27]
-0.6747
[0.58]
-0.7110
[0.11]
0.5879
[0.28]
-0.5563***
[2.63]
(7)
(8)
Winners’ Curse
Explanation
-3.1624***
[3.84]
Yes
Yes
No
397852
3.0461***
[4.96]
-0.5737***
[2.67]
-0.5586***
[5.18]
-0.2832
[0.62]
0.7771
[1.19]
-0.1572
[0.18]
-0.4360
[0.08]
0.4242
[0.21]
-0.6389***
[3.96]
-0.2623
[1.23]
0.0006***
[3.38]
-0.0008*
[1.88]
0.0418
[1.39]
1.3457***
[3.68]
0.9994
[0.35]
-0.1401***
[3.20]
-0.0347
[1.29]
0.7343
[0.87]
-5.8726*
[1.66]
Yes
Yes
No
310301
3.1429***
[4.68]
-0.5488**
[2.31]
-0.4490***
[3.43]
0.0667
[0.17]
0.5609
[0.92]
-0.7490
[0.63]
-0.4633
[0.08]
0.4312
[0.21]
-0.3526**
[1.97]
(9)
(10)
Two Dimensional Clustering
(by firm and size)
This table reports the robustness tests of the causal effects of excessive acquisitiveness on firm failure. Definitions of all the variables are same as in the previous tables. We measure conservative
acquisitiveness using CON SERV ACQijt = DIST. N Hijt × I(Xijt −M edian(X−ijT )<0) . To measure winner, we use cumulative number of completed contested bids normalized by the total number
of deals completed by the firm. Robust z statistics are given in brackets and “*” denotes significance at 10%; “**” denotes significance at 5%; “***” denotes significance at 1% level.
M.M. Rahaman
Corporate Failure
Table 8: Excessive Use of M&A Investment Technology and Corporate Failure: Marginal Effects
52
Ind. Fixed Effect
Year Fixed Effect
log(Tot. Assets)
log(Age)
Net Income/Tot. Assets
Tot. Liab./Tot. Assets
Cash/Tot. Assets
Lt. Debt/Tot. Assets
PPE/Tot. Assets
Market-to-Book
Excessive Acq.
Deregulation Dummy
Ind. Demand Shock
Ind. Supply Shock
Ind. Tech. Shock
Agg. Demand Shock
Agg. Supply Shock
Agg. Equity Shock
Agg. Debt Shock
Agg. Equity Momentum
Yes
Yes
Yes
Yes
-0.0009***
-0.0007***
-0.0002***
0.0014
-0.0003
-0.0010
0.0008
-0.0009***
0.0048***
1/2 Std. Below
the Mean
At
the Mean
-0.0012***
-0.0009***
-0.0003***
0.0019
-0.0004
-0.0013
0.0010
-0.0012***
0.0064***
(2)
(1)
Yes
Yes
-0.0016***
-0.0013***
-0.0004***
0.0025
-0.0005
-0.0017
0.0014
-0.0016***
0.0085***
1/2 Std. Above
the Mean
(3)
Yes
Yes
-0.0007***
-0.0006***
-0.0002***
0.0011
-0.0002
-0.0007
0.0006
-0.0007***
0.0037***
1 Std. Around
the Mean
(4)
Yes
Yes
-0.0013***
-0.0009***
-0.0003***
0.0017
-0.0014
-0.0015
0.0012
-0.0012***
0.0070***
-0.0004
0.0000***
0.0000
0.0001
0.0025***
0.0027
-0.0003***
-0.0001
0.0023
At
the Mean
(5)
Yes
Yes
-0.0009***
-0.0006***
-0.0002***
0.0012
-0.0010
-0.0011
0.0009
-0.0009
0.0052***
-0.0003
0.0000***
0.0000
0.0001
0.0018***
0.002
-0.0002***
-0.0001
0.0017
1/2 Std. Below
the Mean
(6)
Yes
Yes
-0.0017***
-0.0012***
-0.0004***
0.0022
-0.0019
-0.0020
0.0017
-0.0016
0.0095***
-0.0006
0.0000***
0.0000
0.0001
0.0034***
0.0036
-0.0004***
-0.0001
0.0031
1/2 Std. Above
the Mean
(7)
Yes
Yes
-0.0008***
-0.0006***
-0.0002***
0.0010
-0.0009
-0.0009
0.0008
-0.0007
0.0043***
-0.0003
0.0000***
0.0000
0.0000
0.0016***
0.0016
-0.0002***
0.0000
0.0014
1 Std. Around
the Mean
(8)
This table reports the estimates from a discrete-time hazard model to determine the marginal effects of managerial acquisitiveness and other exogenous variables on firm’s failure hazard. The dependent
variable in the regression is 1 for the period in which firm fail and 0 otherwise. Total assets is defined to be the book value of firms assets, net income is income from operation after all taxes and
interest payment, total liabilities are all obligations due to outsiders other than the shareholders of the firms, cash are the value of cash plus other marketable securities hold by the firm, long term
debts are debt obligations due in two or more years time, PPE is defined to be the net book value of firm’s Plant, Property and Equipments, and market value is calculated as market value of equity
plus the book value of firm’s debt. Managerial excessive and conservative acquisitiveness measures are discussed in details in the data section. Industry classifications are based on Fama-French (1997).
We use the total industry net sales from the quarterly COMPUSTAT data item 2 as proxy for industry demand and the total industry cost of good sold from the quarterly COMPUSTAT data item
30 as proxy for industry supply. We also collect information about all patents for the period of 1963-2002 from the NBER patent database and convert the assigned technology class of each of those
patents into international patent class using the methodology developed by Silverman (2002). From the international patent class we covert them back into 1987 Standard Industry Classifications and
assign the patents by grant year to each of our 49 Fama and French (1997) industries. We then decompose the series into trend and irregular components using the Hodrick-Prescott (HP) filter. After
decomposing the trend and irregular components of the series, we calculate series instability by estimating the acceleration (change of change) of the irregular component. We use major deregulatory
initiatives during the sample period as a proxy for regulatory shocks. We use the quarterly real GDP data from the Federal Reserve Bank of St. Louis as proxy for aggregate demand and the real price
of crude petroleum in the U.S. from the U.S. Energy Information Administration as a proxy for aggregate supply. Utilizing the HP filter, we then calculate the aggregate demand and supply shock. As
a proxy for aggregate equity and debt market instability we apply the HP filter on the Dow Jones Industrial average and bank prime lending rate, respectively. To capture the momentum in equity
market, we apply the HP filter on S&P 500 index and use the smoothed trend portion of series as our proxy for momentum in aggregate equity market. We also construct measures of industry merger
wave utilizing the X-12-ARIMA, a seasonal adjustment software produced and maintained by the U.S. Census Bureau. We detail the construction of all variables in the data section of the paper. Robust
z statistics are given in brackets and “*” denotes significance at 10%; “**” denotes significance at 5%; “***” denotes significance at 1% level.
M.M. Rahaman
Corporate Failure
M.M. Rahaman
Corporate Failure
Table 9: Can the Deal Characteristics Discriminate Between the Failed and Non-Failed Sample?
This table reports the deal characteristics of the ‘Failed’ (F) and ‘Non-Failed’ (NF) sample. Panel-A reports the differences in deal
characteristics of average as well as median ‘Non-Failed’ firms from the those of the ‘Failed’ firms (F-NF). Panel-B, on the other hand,
reports the difference in deal characteristics generated from various dummy variables. Deal value is the reported deal value in million
U.S. dollars from the SDC. Total assets is defined to be the book value of all assets while market value is calculated by adding the
market value of equity with the book value of debt at the end of each fiscal quarter. In the table, “*” denotes significance at 10%; “**”
denotes significance at 5%; “***” denotes significance at 1% level using t statistics.
Failed Sample Deals (F)
(1)
(2)
Deal Value ($ Million)
Deal Value/Tot. Assets
Deal Value/ Mkt. Value
Deal Value/Equity Value
Days to Completion
Completed Deals
Target is in the Similar Industry
Acquisition is Merger Wave
Pure Cash Finance Deal
Pure Stock Finance Deal
Financing through Borrowing
Financing through Internal Funds
Financing through Line of Credit
Stock Swap
Block Purchase
Divestiture of Target
Division Sell-off of Target
Financial Acquirer
Mean
Median
41.37***
[11.79]
0.42***
[8.52]
0.25***
[5.16]
0.71**
[2.86]
46.47***
[26.67]
7.93***
[29.18]
0.09***
[43.35]
0.06***
[48.55]
0.10***
[47.02]
0.00
[.02]
N
%
6,455
5,614
1,942
644
842
174
173
211
941
604
1,822
476
118
70.79
61.56
21.30
7.06
9.23
1.91
1.90
2.31
10.32
6.62
19.98
5.22
1.29
Non-Failed Sample Deals (NF)
(3)
(4)
Panel-A: Deal Summary Statistics
Mean
Median
225.99***
[17.72]
0.15***
[26.87]
0.08***
[40.49]
4.09
[1.04]
66.23***
[52.81]
24.42***
[90.47]
0.04***
[81.45]
0.03***
[101.58]
0.06***
[132.03]
12.00***
[22.13]
Panel-B: Deal Characteristics Dummy
N
%
38,417
35276
11791
5502
4270
1218
1626
1299
5211
5075
12343
4521
1145
53
70.50
64.73
21.64
10.10
7.84
2.24
2.98
2.38
9.56
9.31
22.65
8.30
2.10
Difference (F-NF)
(1-3)
(2-4)
F-NFM ean
F-NFM edian
-184.62***
[13.96]
0.27***
[5.47]
0.17***
[3.56]
-3.38
[0.86]
-19.76***
[9.20]
-16.49***
[22.33]
0.05***
[33.04]
0.03***
[37.91]
0.04***
[35.96]
-12.00***
[11.36]
N (1-3)
% (2-4)
-31962
-29662
-9849
-4858
-3428
-1044
-1453
-1088
-4270
-4471
-10521
-4045
-1027
0.29
-3.17
-0.34
-3.04
1.39
-0.33
-1.08
-0.07
0.76
-2.69
-2.67
-3.08
-0.81
Table 10: Evolution of Firms’ Assets and Debt Structure
54
Performance Measures
log(Market Value)
Net Income/Tot. Assets
EBITDA/Tot. Assets
Market-to-Book
Leverage Measures
Totl Liab./Tot. Assets
Book Leverage
Market Leverage
St. debt/Tot. Liab.
Lt. Debt/Tot. Liab
Liquidity and Risk Measures
Cash/Tot. Assets
Cash/Curr. Liab.
Curr. Assets/Curr. Liab.
PPE/Tot. Assets
Cash Flow Volatility
-1.474***
-0.011***
-0.013***
-0.089**
0.017
0.015
0.058**
0.062***
-0.013
-0.008
-0.188***
-0.434***
0.008
0.829***
-0.097***
-0.093***
-0.100***
0.024***
-0.039
0.032**
0.044
0.010
-0.022
0.558***
(2)
Qtr. After
Last Bid
-1.112***
-0.004***
-0.003
0.328***
(1)
Qtr. Before
First Bid
-0.040**
-0.232*
-0.444**
0.030
0.271
0.114**
0.108**
0.158***
0.038***
0.026
-0.362*
-0.007***
-0.010**
-0.417***
Diff-in-Diff
(3)
2.20
1.78
2.60
1.20
1.34
2.65
2.58
4.14
3.39
0.50
1.75
3.61
2.67
4.80
(4)
Absolute
t-stat
-125.00%
-527.27%
-4440.00%
136.36%
48.57%
117.53%
116.13%
158.00%
158.33%
66.67%
-32.55%
-175.00%
-333.33%
-127.13%
(5)
% Change From
First to Last Bid
Diff. Between the Failed and Non-Failed Sample (F-NF)
-0.004
-0.104
0.152
0.017
0.324***
-0.034
-0.032
-0.066**
0.017***
0.053
-0.464***
0.002
0.007***
0.117
(1)
Qtr. Before
First Bid
-0.024***
-0.215***
-0.241**
0.048***
0.447***
0.059**
0.060**
0.051*
0.030***
0.077**
-0.597***
-0.001*
0.001
-0.023
(2)
Qtr. After
Last Bid
-0.020
-0.111
-0.393**
0.031
0.123
0.093**
0.092**
0.117**
0.013*
0.024
-0.133
-0.003*
-0.006*
-0.140*
Diff-in-Diff
(3)
1.19
0.97
2.44
1.35
0.75
2.57
2.60
2.92
1.63
0.51
0.70
1.88
1.61
1.64
(4)
Absolute
t-stat
-500%
-107%
-259%
182%
38%
274%
288%
177%
76%
45%
-29%
-150%
-86%
-120%
(5)
% Change From
First to Last Bid
Diff Between the Excess Acq. and Non-Excess Acq. Sample (X-NX)
This table reports the evolution of firms’ debt and assets structure from one quarter before the first M&A bid to one quarter after the last M&A bid for the median firms in our sample. All firms
for which differential firm characteristics are reported here makes exactly 3 acquisition bids because the median number of bids firms make in our sample is 3. First panel on left of the table reports
the difference in firm characteristics of non-failed median firms from the failed median firms (F-NF). Second panel on the right of the table reports the difference in firm characteristics of the median
excessively acquisitive sample from the non-excessively acquisitive sample (X-NX). Market value is calculated by adding the market value of equity with the book value of debt at the end of each fiscal
quarter. Net income is earning after all interest and tax payment while EBITDA is earning before interest, tax, depreciation, and amortization. Market-to-book ratio is calculated by diving the market
value of firms assets with the book value of its assets. Among the leverage and liquidity measure, total liabilities measure all outstanding liabilities owed to outsider other than the shareholders of the
firm. Book leverage is defined to be the ratio of firm’s total outstanding short-term and long-term debt to book value of total assets whereas market leverage is defined to be the ratio of total outstanding
short and long-term debt to the market value of firm’s total assets. Cash is defined to be the value of cash as well as other cash equivalent marketable securities, current assets are cash plus account
receivables, current liabilities are short-term debt plus account payable, short term debts are debt obligations maturing within one year while long term debts are debt
obligations maturing in two years
or more in time. PPE refers to net book value of firm’s Property, Plant and Equipment. Cash flow volatility is calculated as log abs(EBIT DAit − EBIT DAit−1 ) for each firm i and time period t.
In the table, “*” denotes significance at 10%; “**” denotes significance at 5%; “***” denotes significance at 1% level using t statistics.
M.M. Rahaman
Corporate Failure
M.M. Rahaman
Corporate Failure
Table 11: Excessive Use of M&A Investment Technology and Corporate Default
This table reports the estimates from the discrete-time hazard regression to determine the effects managerial excessive acquisitiveness
and various determinants financial distress on firm’s default hazard. Total assets is defined to be the book value of firms assets while
market value is book value of debt plus market value of equity, net income is income from operation after all taxes and interest payment,
total liabilities are obligations due to outsiders other than the shareholders of the firms. Current assets are cash plus account receivables,
current liabilities are short-term debt plus account payable. ZSCORE is calculated from Altman (2000). Sigma and excess return are
calculated following Shumway (2001). Governance score is from Gompers, Ishii and Metrick (2003) and the excessive acquisitiveness
measure is explained in detail in the data section of the paper. Robust z statistics are given in brackets and “*” denotes significance
at 10%; “**” denotes significance at 5%; “***” denotes significance at 1% level.
Altman (1968)
(1)
(2)
log(Age)
Excessive Acq.
Excessive Acq. - Marginal Effect
0.188
[1.58]
1.450***
[6.29]
0.001***
[5.77]
Governance Score
Governance Score - Marginal Effect
ZSCORE
-0.006***
[3.67]
0.369
[1.64]
Zmijewski (1984)
(3)
(4)
0.108
[0.98]
1.335***
[5.91]
0.001***
[3.44]
-0.061
[1.58]
0.000
[1.46]
-0.030***
[4.43]
Net Income/Tot. Assets
Curr. Assets/Curr. Liab.
Excess Return
Sigma
Ind. Fixed Effect
Year Fixed Effect
N
Num. of Firms
Pseudo-R2
-21.770***
[18.10]
-20.509
[.]
-21.400***
[17.49]
-25.223
[.]
Yes
Yes
263343
7606
0.07
Yes
Yes
104519
2076
0.09
Yes
Yes
340092
8233
0.08
Yes
Yes
134005
2226
0.20
55
0.407**
[2.14]
-0.042
[1.28]
0.000
[123]
-1.775***
[3.99]
2.210***
[7.41]
0.030
[0.57]
log(Mkt. Value)
Constant
0.203**
[2.04]
1.862***
[9.85]
0.001***
[8.43]
-0.068
[1.58]
0.000
[1.53]
-0.024
[0.80]
0.018*
[1.83]
-0.363**
[2.49]
Tot. Liab./Tot. Assets
0.244
[1.02]
Shumway (2001)
(5)
(6)
-0.046
0.031
[1.36]
[0.48]
-1.708
-2.255
[1.51]
[1.38]
9.708***
22.205***
[6.97]
[9.01]
-25.733*** -29.727***
[21.51]
[21.68]
Yes
Yes
423442
10502
0.11
Yes
Yes
170016
2787
0.16
M.M. Rahaman
Corporate Failure
Table 12: Mediating the Causality: The Risk Channel
This table reports the estimates from a mediating instrument methodology to determine the channels through which managerial
excessive acquisitiveness catalyzes firm’s failure. Total assets is defined to be the book value of firms assets, net income is income from
operation after all taxes and interest payment, total liabilities are all obligations due to outsiders other than the shareholders of the
firms, cash are the value of cash plus other marketable securities hold by the firm, long term debts are debt obligations due in two or
more years time, PPE is defined to be the net book value of firm’s Plant, Property and Equipments, and market value is calculated
as market value of equity plus the book value of firm’s debt. Managerial excessive acquisitiveness
measure is discussed in detail in
the data section of the paper. BRISK is defined to be as log abs(EBIT DAit − EBIT DAit−1 ) for each firm i and time period t.
Sigma is calculated following Shumway (2001). Industry classifications are based on Fama-French (1997). We use the total industry
net sales from the quarterly COMPUSTAT data item 2 as proxy for industry demand and the total industry cost of good sold from
the quarterly COMPUSTAT data item 30 as proxy for industry supply. We also collect information about all patents for the period
of 1963-2002 from the NBER patent database and convert the assigned technology class of each of those patents into international
patent class using the methodology developed by Silverman (2002). From the international patent class we covert them back into 1987
Standard Industry Classifications and assign the patents by grant year to each of our 49 Fama and French (1997) industries. We then
decompose the series into trend and irregular components using the Hodrick-Prescott (HP) filter. After decomposing the trend and
irregular components of the series, we calculate series instability by estimating the acceleration (change of change) of the irregular
component. We use major deregulatory initiatives during the sample period as a proxy for regulatory shocks. We use the quarterly
real GDP data from the Federal Reserve Bank of St. Louis as proxy for aggregate demand and the real price of crude petroleum in
the U.S. from the U.S. Energy Information Administration as a proxy for aggregate supply. Utilizing the HP filter, we then calculate
the aggregate demand and supply shock. As a proxy for aggregate equity and debt market instability we apply the HP filter on the
Dow Jones Industrial average and bank prime lending rate, respectively. To capture the momentum in equity market, we apply the HP
filter on S&P 500 index and use the smoothed trend portion of series as our proxy for momentum in aggregate equity market. We also
construct measures of industry merger wave utilizing the X-12-ARIMA, a seasonal adjustment software produced and maintained by
the U.S. Census Bureau. We detail the construction of all variables in the data section of the paper. Robust z statistics are given in
brackets and “*” denotes significance at 10%; “**” denotes significance at 5%; “***” denotes significance at 1% level.
Total Effect
Mediating Through BRISK Measure
Dependent Variable
Estmination Methodology
FAILURE
(1) LOGIT
BRISK
(2) OLS
FAILURE
(3) LOGIT
FAILURE
(4) LOGIT
BRISK
(5) TOBIT
FAILURE
(6) LOGIT
FAILURE
(7) LOGIT
log(Tot. Assets)
-0.6098**
[2.36]
-0.4150***
[3.83]
-0.1365**
[2.30]
0.7856
[1.27]
-0.6743
[0.57]
-0.7121
[0.11]
0.5790
[0.28]
-0.5551***
[2.62]
3.3286***
[5.16]
0.7665***
[146.75]
-0.0200
[1.40]
-0.2948**
[2.22]
0.2637***
[7.81]
0.1525***
[2.94]
-0.0149
[1.10]
-0.0157
[0.31]
0.0241***
[3.03]
0.1003***
[2.59]
-0.5617
[1.38]
-0.7216***
[4.42]
-0.1211
[0.49]
1.8658**
[2.44]
-0.6475
[0.51]
-0.7184
[0.09]
0.2437
[0.09]
-0.6811***
[4.18]
-0.6375*
[1.73]
-0.5156***
[3.37]
0.0556
[0.12]
1.5649**
[1.96]
-0.2321
[0.21]
-0.7301
[0.09]
0.5210
[0.19]
-0.4369***
[3.71]
3.2067***
[3.71]
0.1324
[1.06]
-0.0058***
[237.53]
-0.0034***
[42.73]
-0.0038***
[20.16]
0.0021***
[28.19]
-0.0042***
[16.09]
-0.0001*
[1.73]
0.0006**
[2.46]
-0.0005***
[34.14]
0.0048***
[20.07]
-0.3884
[1.58]
-0.6052***
[5.32]
-0.1538
[1.29]
0.8088**
[2.36]
-1.0699
[0.87]
-0.4797
[0.08]
0.1525
[0.07]
-0.5978***
[2.66]
-0.5122**
[2.26]
-0.3893***
[3.46]
-0.0657
[0.20]
0.5942
[1.15]
-0.6132
[0.57]
-0.4149
[0.07]
0.4264
[0.21]
-0.4505***
[2.60]
3.2329***
[5.22]
8.9590***
[6.01]
-0.1997
[0.93]
0.0005**
[2.26]
-0.0005
[1.25]
0.0301
[0.87]
0.9496**
[2.15]
-0.4744
[0.15]
-0.1758***
[3.25]
-0.0246
[0.56]
1.6027**
[2.08]
-12.1997***
[4.02]
8.6466***
[4.52]
-0.1821
[0.93]
0.0006**
[2.51]
-0.0006
[1.33]
0.0347
[1.06]
0.9810**
[2.32]
-0.0401
[0.01]
-0.1550***
[2.91]
-0.0177
[0.46]
1.2625*
[1.72]
-10.2615***
[3.38]
Yes
Yes
318825
8677
0.16
18782.93
Yes
Yes
315492
8677
0.20
5948.85
log(Age)
Net Income/Tot. Assets
Tot. Liab./Tot. Assets
Cash/Tot.Assets
Lt. Debt/Tot. Assets
PPE/Tot. Assets
Market-to-Book
Excessive Acq.
BRISK
0.1532
[1.28]
Mediating Through Sigma Measure
Sigma
Deregulation Dummy
Ind. Demand Shock
Ind. Supply Shock
Ind. Tech. Shock
Agg. Demand Shock
Agg. Supply Shock
Agg. Equity Shock
Agg. Debt Shock
Agg. Equity Momentum
Constant
Ind. Fixed Effect
Year Fixed Effect
N
Num. of Firms
R2 /Pseudo-R2
Wald-χ2
-0.2085
[0.97]
0.0005**
[2.56]
-0.0006
[1.38]
0.0352
[1.16]
1.1733***
[2.90]
1.2174
[0.40]
-0.1339***
[2.61]
-0.0434
[1.15]
1.0292
[1.24]
-7.9488**
[2.21]
-0.0363*
[1.66]
-0.0002***
[7.78]
-0.0002***
[3.06]
-0.0014
[0.45]
0.0796*
[1.77]
-1.3947***
[4.34]
0.0187***
[2.91]
-0.0046**
[2.48]
-1.7568***
[17.09]
4.7463***
[9.56]
-0.0420
[0.19]
0.0006***
[2.63]
-0.0004
[0.96]
0.0448
[1.17]
1.0392**
[2.19]
-3.6694
[1.00]
-0.1649***
[2.78]
-0.0198
[0.41]
1.7702**
[2.55]
-12.2673***
[3.93]
-0.0187
[0.09]
0.0006***
[2.90]
-0.0005
[1.06]
0.0452
[1.24]
1.0161**
[2.17]
-2.6982
[0.79]
-0.1472***
[2.62]
-0.0132
[0.33]
1.6141**
[2.33]
-10.4138***
[3.21]
Yes
Yes
315595
8678
0.19
5109.62
Yes
Yes
249716
8148
0.60
Yes
Yes
249128
8148
0.17
14410.87
Yes
Yes
247548
8144
0.20
4171.29
56
-0.0013***
[4.03]
0.0000
[1.17]
-0.0000***
[3.93]
0.0002***
[4.41]
0.0108***
[13.39]
0.0721***
[12.43]
0.0002
[1.38]
-0.0003***
[8.53]
-0.0124***
[10.43]
0.1232***
[21.86]
Yes
Yes
318656
8677
Table 13: Mediating the Causality: The Behavioral Channel
57
Industry Dummy
Year Dummy
N
Num. of Firms
Adj-R2 /Pseudo-R2
Wald-χ2
Constant
Agg. Equity Momentum
Agg. Debt Shock
Agg. Equity Shock
Agg. Supply Shock
Agg. Demand Shock
Ind. Tech. Shock
Ind. Supply Shock
Ind. Demand Shock
Deregulation Dummy
Attn. Allocation
Mgt. Bias
Sigma
Excessive Acq.
Market-to-Book
PPE/Tot. Assets
Lt. Debt/Tot. Assets
Cash/Tot.Assets
Tot. Liab./Tot. Assets
Net Income/Tot. Assets
Yes
Yes
315595
8678
0.19
5109.62
-0.2085
[0.97]
0.0005**
[2.56]
-0.0006
[1.38]
0.0352
[1.16]
1.1733***
[2.90]
1.2174
[0.40]
-0.1339***
[2.61]
-0.0434
[1.15]
1.0292
[1.24]
-7.9488**
[2.21]
-0.6098**
[2.36]
-0.4150***
[3.83]
-0.1365**
[2.30]
0.7856
[1.27]
-0.6743
[0.57]
-0.7121
[0.11]
0.5790
[0.28]
-0.5551***
[2.62]
3.3286***
[5.16]
log(Tot. Assets)
log(Age)
FAILURE
(1) LOGIT
Dependent Variable
Estimation Methodology
Total Effect
Yes
Yes
318760
8678
0.0048***
[3.21]
-0.0000***
[14.31]
-0.0000***
[2.92]
-0.0004
[1.50]
0.0016
[0.42]
0.0449*
[1.65]
0.0005
[0.99]
0.0003*
[1.75]
0.0899***
[16.15]
0.3697***
[14.02]
-0.0020***
[17.19]
-0.0020***
[5.41]
-0.0009
[1.09]
0.0031***
[8.78]
0.0105***
[8.53]
0.0001
[0.57]
0.0012
[0.99]
-0.0012***
[18.09]
0.0911***
[81.92]
MGT. BIAS
(2) TOBIT
Yes
Yes
318928
8678
0.15
15864.67
-0.2734
[1.20]
0.0005**
[2.40]
-0.0005
[1.20]
0.0352
[1.12]
1.1288***
[2.75]
0.6993
[0.23]
-0.1584***
[3.09]
-0.0593
[1.31]
1.0984
[1.21]
-12.6808***
[3.08]
4.7467***
[12.97]
-0.4864*
[1.88]
-0.5244***
[3.94]
-1.1539*
[1.85]
0.6928
[0.65]
-1.2118
[1.09]
-0.4246
[0.07]
0.2269
[0.11]
-0.5896***
[3.08]
FAILURE
(3) LOGIT
Yes
Yes
315595
8678
0.20
5029.25
-0.2242
[1.07]
0.0006***
[2.73]
-0.0006
[1.26]
0.0376
[1.22]
1.1294***
[2.76]
1.2217
[0.40]
-0.1374***
[2.70]
-0.0451
[1.20]
0.7422
[0.94]
-9.6126***
[2.64]
3.4729***
[9.52]
-0.6077**
[2.36]
-0.3517***
[3.12]
-0.1280*
[1.69]
0.7410
[1.20]
-0.6849
[0.58]
-0.7310
[0.11]
0.6234
[0.30]
-0.5356***
[2.62]
3.0300***
[4.47]
FAILURE
(4) LOGIT
Mediating Through Mgt. Bias Measure
Yes
Yes
318760
8678
0.02
-0.0001
[0.10]
0.0000
[0.74]
-0.0000
[0.90]
-0.0000
[1.08]
0.0011**
[2.24]
-0.0002
[0.03]
-0.0000
[0.19]
0.0000
[0.25]
-0.0022*
[1.95]
-0.0011
[0.12]
0.0007***
[4.39]
0.0027***
[4.57]
0.0007
[1.44]
0.0011
[1.49]
0.0006
[0.32]
0.0000
[0.71]
-0.0019
[0.66]
-0.0001*
[1.89]
0.0168***
[7.60]
ATTN. ALLOCA
(5) OLS
Yes
Yes
318928
8678
0.14
13751.80
1.6759**
[2.44]
-0.2467
[1.17]
0.0005**
[2.57]
-0.0005
[1.31]
0.0349
[1.10]
1.2150***
[2.86]
0.9825
[0.32]
-0.1585***
[3.16]
-0.0658
[1.56]
1.3659*
[1.88]
-9.5868***
[3.34]
-0.5188**
[2.00]
-0.6465***
[5.24]
0.1181
[0.26]
1.0388**
[2.24]
-1.0712
[0.92]
-0.4869
[0.08]
0.1999
[0.09]
-0.7154***
[3.02]
FAILURE
(6) LOGIT
Yes
Yes
315595
8678
0.19
5136.19
1.3479**
[2.13]
-0.2091
[0.98]
0.0005**
[2.55]
-0.0006
[1.38]
0.0357
[1.17]
1.1802***
[2.92]
1.2309
[0.40]
-0.1348***
[2.63]
-0.0435
[1.15]
1.0408
[1.26]
-7.9721**
[2.21]
-0.6108**
[2.37]
-0.4208***
[3.98]
-0.1358**
[2.24]
0.7815
[1.26]
-0.6766
[0.58]
-0.7132
[0.11]
0.5850
[0.28]
-0.5529***
[2.61]
3.3191***
[5.13]
FAILURE
(7) LOGIT
Mediating Through Attn. Allocation Measures
Yes
Yes
318825
8677
0.18
19122.14
8.5150***
[5.28]
4.6408***
[12.96]
1.5587***
[2.66]
-0.2403
[1.08]
0.0005**
[2.44]
-0.0005
[1.10]
0.0319
[0.92]
0.9470**
[2.15]
-0.6726
[0.22]
-0.1756***
[3.22]
-0.0292
[0.67]
1.2687*
[1.68]
-14.6042***
[4.62]
-0.3940
[1.62]
-0.5348***
[4.35]
-0.1426
[1.48]
0.7520**
[2.28]
-1.0570
[0.86]
-0.4962
[0.08]
0.2271
[0.11]
-0.5523**
[2.57]
FAILURE
(8) LOGIT
Yes
Yes
315492
8677
0.21
5898.97
-0.5130**
[2.27]
-0.3369***
[2.91]
-0.0675
[0.19]
0.5470
[0.92]
-0.6231
[0.60]
-0.4253
[0.07]
0.4561
[0.22]
-0.4316***
[2.61]
2.9354***
[4.52]
8.4336***
[4.08]
3.2867***
[10.18]
1.2460*
[1.96]
-0.2081
[1.01]
0.0006**
[2.55]
-0.0005
[1.14]
0.0365
[1.11]
0.9797**
[2.30]
-0.1487
[0.05]
-0.1567***
[2.94]
-0.0226
[0.59]
0.9141
[1.24]
-11.4410***
[3.55]
FAILURE
(9) LOGIT
All Channels
This table reports the estimates from a mediating instrument methodology to determine the channels through which managerial excessive acquisitiveness catalyzes firm’s failure. Total assets is defined
to be the book value of firms assets, net income is income from operation after all taxes and interest payment, total liabilities are all obligations due to outsiders other than the shareholders of the
firms, cash are the value of cash plus other marketable securities hold by the firm, long term debts are debt obligations due in two or more years time, PPE is defined to be the net book value of
firm’s Plant, Property and Equipments, and market value is calculated as market value of equity plus the book value of firm’s debt. Managerial excessive acquisitiveness,
managerial cognitive bias, and
managerial attention allocation measures are discussed in detail in the data section of the paper. BRISK is defined to be as log abs(EBIT DAit − EBIT DAit−1 ) for each firm i and time period t.
Sigma is calculated following Shumway (2001). Industry classifications are based on Fama-French (1997). We use the total industry net sales from the quarterly COMPUSTAT data item 2 as proxy
for industry demand and the total industry cost of good sold from the quarterly COMPUSTAT data item 30 as proxy for industry supply. We also collect information about all patents for the period
of 1963-2002 from the NBER patent database and convert the assigned technology class of each of those patents into international patent class using the methodology developed by Silverman (2002).
From the international patent class we covert them back into 1987 Standard Industry Classifications and assign the patents by grant year to each of our 49 Fama and French (1997) industries. We then
decompose the series into trend and irregular components using the Hodrick-Prescott (HP) filter. After decomposing the trend and irregular components of the series, we calculate series instability by
estimating the acceleration (change of change) of the irregular component. We use major deregulatory initiatives during the sample period as a proxy for regulatory shocks. We use the quarterly real
GDP data from the Federal Reserve Bank of St. Louis as proxy for aggregate demand and the real price of crude petroleum in the U.S. from the U.S. Energy Information Administration as a proxy
for aggregate supply. Utilizing the HP filter, we then calculate the aggregate demand and supply shock. As a proxy for aggregate equity and debt market instability we apply the HP filter on the Dow
Jones Industrial average and bank prime lending rate, respectively. To capture the momentum in equity market, we apply the HP filter on S&P 500 index and use the smoothed trend portion of series
as our proxy for momentum in aggregate equity market. We also construct measures of industry merger wave utilizing the X-12-ARIMA, a seasonal adjustment software produced and maintained by the
U.S. Census Bureau. We detail the construction of all variables in the data section of the paper. Robust z statistics are given in brackets and “*” denotes significance at 10%; “**” denotes significance
at 5%; “***” denotes significance at 1% level.
M.M. Rahaman
Corporate Failure
Table 14: The Capital Market Reaction
58
0.014
[0.70]
Yes
Yes
Yes
63556
10771
0.01
0.01
8.13
Deal Charact. Dummy
Ind. Fixed Effect
Year Fixed Effect
N
Num. of Firms
R2
Adj-R2
F-stat
-0.019***
[12.23]
Constant
Deal Value/Mkt. Equity
Governance Score
Attn. Allocation
Mgt. Bias
Sigma
BRISK
Excessive Acq.
(1)
Yes
Yes
Yes
43784
9184
0.02
0.02
8.80
0.039***
[3.95]
-0.003***
[16.24]
(2)
Yes
Yes
Yes
63580
10774
0.02
0.02
7.52
0.001
[0.03]
0.533***
[5.81]
(3)
Yes
Yes
Yes
63581
10774
0.01
0.01
6.52
0.011
[0.51]
0.003
[0.80]
(4)
Yes
Yes
Yes
63581
10774
0.01
0.01
6.96
0.013
[0.61]
-0.007***
[8.10]
(5)
Yes
Yes
Yes
33771
2925
0.02
0.02137
0.005
[0.21]
-0.001***
[4.51]
(6)
Yes
Yes
Yes
23534
2730
0.03
0.02957
0.054***
[3.29]
-0.004**
[2.43]
-0.001***
[3.11]
0.318***
[4.00]
-0.006
[1.47]
-0.001
[1.17]
-0.000***
[2.86]
(7)
Yes
Yes
Yes
38585
9148
0.02
0.02
160.58
0.000
[1.17]
0.037
[1.19]
-0.020***
[9.91]
(8)
Yes
Yes
Yes
27910
7846
0.02
0.02
150.53
0.000
[1.09]
0.048***
[4.59]
-0.004***
[13.27]
(9)
Yes
Yes
Yes
38600
9153
0.04
0.03
126.53
0.000
[1.01]
0.017
[0.54]
0.745***
[7.64]
(10)
Yes
Yes
Yes
38601
9153
0.02
0.01
149.60
0.000
[0.84]
0.031
[0.97]
0.005
[1.43]
(11)
Yes
Yes
Yes
38601
9153
0.02
0.01
142.51
0.000
[0.84]
0.035
[1.12]
-0.007***
[6.46]
(12)
Yes
Yes
Yes
20447
2821
0.03
0.03
-0.001***
[4.62]
-0.009**
[2.45]
0.019
[0.68]
(13)
Yes
Yes
Yes
14839
2615
0.04
0.03
-0.005*
[1.82]
-0.001***
[3.48]
0.244**
[2.57]
-0.005
[0.90]
-0.001
[0.88]
-0.001***
[3.51]
-0.012***
[2.75]
0.070***
[2.87]
(14)
This table reports the estimates from OLS regression to determine the market reactions to managerial acquisitiveness and various mediating instruments. Deal value is in U.S.$ million from the SDC
database. Market value is defined to be market value of equity plus book value of debt. Managerial excessive acquisitiveness,
managerial cognitive bias, and managerial attention allocation measures are
discussed in detail in the data section of the paper. BRISK is defined to be as log abs(EBIT DAit − EBIT DAit−1 ) for each firm i and time period t. Sigma is calculated following Shumway (2001).
Governance index is from Gompers, Ishii and Metrick (2003). Robust z statistics are given in brackets and “*” denotes significance at 10%; “**” denotes significance at 5%; “***” denotes significance
at 1% level.
M.M. Rahaman
Corporate Failure
Table 15: Excessive Use of M&A and Exit/Takeover Hazard
59
Ind. Fixed Effect
Year Fixed Effect
N
Num. of Firms
Pseudo-R2
Constant
Agg. Equity Momentum
Agg. Debt Shock
Agg. Equity Shock
Agg. Supply Shock
Agg. Demand Shock
Ind. Tech. Shock
Ind. Supply Shock
Ind. Demand Shock
Deregulation Dummy
Governance Score
Mgt. Bias
Sigma
Excessive Acq.
Market-to-Book
PPE/Tot. Assets
Lt. Debt/Tot. Assets
Cash/Tot.Assets
Tot. Liab./Tot. Assets
Net Income/Tot. Assets
log(Age)
log(Tot. Assets)
Yes
Yes
315595
8678.00
0.11
-0.1145
[0.81]
0.0007***
[4.38]
-0.0006**
[2.04]
-0.0091
[0.43]
1.2341***
[4.24]
0.8588
[0.40]
-0.0615*
[1.65]
-0.0046
[0.20]
-0.5935
[1.00]
-2.4213
[1.01]
-0.2929***
[3.36]
-0.1113
[0.97]
-0.0489
[1.40]
0.4291
[0.88]
-0.4345
[0.66]
-0.7692
[0.23]
0.4737
[0.45]
-0.2146**
[2.29]
3.2386***
[8.88]
(1)
Yes
Yes
318825
8677.00
0.06
-0.0879
[0.61]
0.0006***
[4.07]
-0.0005*
[1.66]
-0.0140
[0.63]
1.0819***
[3.65]
-0.2498
[0.11]
-0.0915**
[2.40]
-0.0022
[0.09]
-0.1288
[0.31]
-16.4334
[.]
8.0078***
[8.85]
-0.0989***
[6.17]
-0.2423***
[8.82]
-0.0856
[1.53]
0.3024
[0.60]
-0.8190***
[3.30]
-0.0057***
[2.73]
-0.1628*
[1.77]
-0.2094***
[4.02]
(2)
Yes
Yes
318928
8678.00
0.07
-0.1395
[0.95]
0.0007***
[4.49]
-0.0005*
[1.71]
-0.0084
[0.38]
1.2187***
[4.10]
0.3607
[0.17]
-0.0872**
[2.32]
-0.0184
[0.73]
-0.5536
[0.97]
-7.9457***
[3.43]
4.8504***
[20.27]
-0.1723*
[1.88]
-0.1890
[1.46]
-0.4812
[1.06]
0.3931
[0.67]
-0.8852
[1.31]
-0.2295
[0.07]
-0.0099
[0.01]
-0.2134***
[2.60]
(3)
Yes
Yes
109991
2554.00
0.08
-0.0271*
[1.68]
-0.5418
[1.46]
0.0002
[0.79]
-0.0002
[0.38]
-0.0155
[0.30]
2.0120***
[3.24]
6.6360
[0.91]
-0.0838
[1.09]
-0.2458**
[2.34]
-3.4164***
[3.26]
-4.1891
[.]
-0.1173***
[4.26]
0.1714**
[2.10]
-0.5889***
[2.73]
1.2234***
[8.05]
0.4067
[1.49]
-0.0072***
[3.88]
0.0848
[0.35]
-0.2894***
[6.76]
(4)
Exit through any Routes
Yes
Yes
315492
8677.00
0.12
-0.1112
[0.79]
0.0007***
[4.51]
-0.0006*
[1.74]
-0.0111
[0.51]
1.0999***
[3.69]
0.1613
[0.07]
-0.0742*
[1.95]
0.0034
[0.16]
-0.7883
[1.50]
-5.0096**
[2.34]
-0.2261***
[3.17]
-0.0216
[0.19]
-0.0300
[0.36]
0.2482
[1.04]
-0.4614
[0.69]
-0.6967
[0.21]
0.4287
[0.40]
-0.1675*
[1.84]
2.9020***
[7.48]
6.7708***
[4.73]
3.3285***
[13.04]
(5)
Yes
Yes
108328
2552.00
0.19
-0.2601***
[8.05]
0.5548***
[5.77]
-0.1964
[0.36]
0.6591***
[3.77]
0.1162
[0.42]
-0.0063***
[3.34]
0.3805
[1.46]
-0.1375***
[4.11]
1.3140***
[7.82]
8.9385***
[3.24]
13.4498***
[16.73]
0.0139
[0.86]
-0.6306
[1.64]
0.0003
[0.90]
-0.0002
[0.30]
-0.0202
[0.39]
2.0008***
[3.00]
6.6401
[0.90]
-0.1011
[1.26]
-0.2148**
[2.00]
-4.0524***
[3.75]
-12.7910
[.]
(6)
Yes
Yes
315595
8678.00
0.08
-0.1453
[0.82]
0.0008***
[4.19]
-0.0007*
[1.94]
-0.0397
[1.50]
0.8346**
[2.37]
2.6950
[0.97]
-0.0282
[0.62]
-0.0217
[0.71]
-1.7358***
[3.88]
-10.6293
[.]
-0.1259***
[10.51]
-0.1075***
[3.00]
0.1147**
[2.25]
0.2434***
[7.19]
0.2169*
[1.77]
-0.0038**
[2.08]
0.1973*
[1.65]
-0.1570***
[7.13]
2.8617***
[35.26]
(7)
Yes
Yes
318825
8677.00
0.04
-0.1375
[0.78]
0.0007***
[3.96]
-0.0006*
[1.72]
-0.0412
[1.55]
0.8222**
[2.37]
2.1503
[0.78]
-0.0479
[1.06]
-0.0292
[0.99]
-1.2488***
[2.66]
-12.0359
[.]
1.5236***
[3.16]
0.0137
[1.26]
-0.2432***
[7.64]
0.0385
[0.64]
0.2451***
[7.83]
-0.3308***
[2.74]
-0.0045***
[2.60]
-0.1824
[1.64]
-0.1643***
[7.41]
(8)
Yes
Yes
318928
8678.00
0.07
-0.1687
[0.95]
0.0008***
[4.37]
-0.0006
[1.64]
-0.0375
[1.41]
0.8094**
[2.31]
2.0456
[0.74]
-0.0541
[1.19]
-0.0316
[1.07]
-1.6872***
[3.49]
-16.1476
[.]
6.3609***
[23.21]
-0.0077
[0.76]
-0.1267***
[3.97]
0.0740
[1.01]
0.1822***
[3.08]
-0.3657***
[3.12]
-0.0050***
[2.87]
-0.0780
[0.72]
-0.1300***
[6.24]
(9)
Yes
Yes
109991
2554.00
0.08
-0.0238
[1.39]
-0.7880*
[1.85]
0.0003
[0.99]
-0.0001
[0.15]
-0.0372
[0.69]
1.6853**
[2.56]
11.1539
[1.41]
-0.0382
[0.46]
-0.2849**
[2.45]
-4.2379***
[3.74]
-0.2512
[.]
-0.0687**
[2.46]
0.1626*
[1.87]
0.4943
[1.07]
0.6647***
[4.28]
0.2499
[0.87]
-0.0080***
[3.41]
0.0850
[0.33]
-0.2320***
[5.95]
(10)
Exit through Acquisition
Yes
Yes
315492
8677.00
0.10
-0.1652
[0.91]
0.0008***
[4.44]
-0.0007*
[1.82]
-0.0374
[1.40]
0.8044**
[2.25]
2.4948
[0.90]
-0.0354
[0.77]
-0.0240
[0.80]
-2.1033***
[4.56]
-13.2986
[.]
-0.1202***
[8.97]
0.0018
[0.05]
0.1800**
[2.34]
0.1408**
[2.06]
0.1303
[1.06]
-0.0041**
[2.26]
0.2132*
[1.77]
-0.1312***
[6.08]
2.4058***
[27.23]
0.1778
[0.26]
4.9235***
[17.14]
(11)
Yes
Yes
108328
2552.00
0.18
-0.2457***
[7.97]
0.5446***
[5.38]
0.9716**
[2.57]
0.2490
[1.31]
0.0616
[0.21]
-0.0074***
[3.16]
0.3965
[1.42]
-0.1099***
[3.48]
1.2899***
[7.18]
3.4705**
[2.20]
13.9031***
[16.56]
0.0188
[1.08]
-0.8599*
[1.96]
0.0004
[1.14]
-0.0001
[0.19]
-0.0364
[0.68]
1.6371**
[2.34]
11.6031
[1.45]
-0.0584
[0.68]
-0.2739**
[2.33]
-4.8626***
[4.21]
-9.9552
[.]
(12)
This table reports the estimates from discrete-time hazard model of exit and takeover. Total assets is defined to be the book value of firms assets, net income is income from operation after all taxes and
interest payment, total liabilities are all obligations due to outsiders other than the shareholders of the firms, cash are the value of cash plus other marketable securities hold by the firm, long term debts
are debt obligations due in two or more years time, PPE is defined to be the net book value of firm’s Plant, Property and Equipments, and market value is calculated as market value of equity plus the
book value of firm’s debt. Managerial excessive acquisitiveness and managerial cognitive bias measures are discussed in detail in the data section of the paper. Sigma is calculated following Shumway
(2001). Governance score is from Gompers, Ishii and Metrick (2003). Industry classifications are based on Fama-French (1997). We use the total industry net sales from the quarterly COMPUSTAT
data item 2 as proxy for industry demand and the total industry cost of good sold from the quarterly COMPUSTAT data item 30 as proxy for industry supply. We also collect information about all
patents for the period of 1963-2002 from the NBER patent database and convert the assigned technology class of each of those patents into international patent class using the methodology developed
by Silverman (2002). From the international patent class we covert them back into 1987 Standard Industry Classifications and assign the patents by grant year to each of our 49 Fama and French
(1997) industries. We then decompose the series into trend and irregular components using the Hodrick-Prescott (HP) filter. After decomposing the trend and irregular components of the series, we
calculate series instability by estimating the acceleration (change of change) of the irregular component. We use major deregulatory initiatives during the sample period as a proxy for regulatory shocks.
We use the quarterly real GDP data from the Federal Reserve Bank of St. Louis as proxy for aggregate demand and the real price of crude petroleum in the U.S. from the U.S. Energy Information
Administration as a proxy for aggregate supply. Utilizing the HP filter, we then calculate the aggregate demand and supply shock. As a proxy for aggregate equity and debt market instability we apply
the HP filter on the Dow Jones Industrial average and bank prime lending rate, respectively. To capture the momentum in equity market, we apply the HP filter on S&P 500 index and use the smoothed
trend portion of series as our proxy for momentum in aggregate equity market. We also construct measures of industry merger wave utilizing the X-12-ARIMA, a seasonal adjustment software produced
and maintained by the U.S. Census Bureau. We detail the construction of all variables in the data section of the paper. Robust z statistics are given in brackets and “*” denotes significance at 10%;
“**” denotes significance at 5%; “***” denotes significance at 1% level.
M.M. Rahaman
Corporate Failure