Digitized by the Internet Archive
in
2011 with funding from
Boston Library Consortium IVIember Libraries
http://www.archive.org/details/zombielendingdepOOcaba
DEWE'^
HB31
.M415
Massachusetts Institute of Technology
Department of Economics
Working Paper Series
Zombie Lending and
Depressed Restructuring
in
Ricardo
Japan
J.
Caballero
Takeo Hoshi
Anil K. Kashyap
Working Paper 06-06
March 8, 2006
RoomE52-251
50 Memorial Drive
Cambridge,
02142
MA
This
paper can be downloaded without charge from the
Network Paper Collection at
Social Science Research
httsp://ssrn.com/abstract=889727
3SACHUSETTS INSTITUTF
OF TECHNOLOGY
MAR
2 2 2006
LIBRARIES
I
'
Zombie Lending and Depressed Restructuring
Ricardo
J.
in
Japan
Caballero
Massachusetts Institute of Technology and
NBER
Takeo Hoshi
University of California at San Diego, Graduate School of International Relations and
Pacific Studies
Anil
and
NBER
K Kashyap
University of Chicago, Graduate School of Business, Federal Reserve
and
This
We
Bank of Chicago
NBER
draft:
March 2006
thank numerous seminar participants and our colleagues for useful comments.
especially benefited from helpful
Braun, Kenichiro Kobayashi,
Arai,
comments by
Hugh
Patrick,
Olivier Blanchard,
and Makoto Sakuragawa.
Munechika Katayama and Tatsuyoshi Okimoto
We
Roger Bohn, Toni
We
thank Yoichi
for expert research assistance.
Caballero thanks the National Science Foundation for research support. Hoshi thanks the
Research
Institute
Kashyap thanks
of Economy, Trade, and Industry (RIETI) for research support.
the Center for Research in Securities Prices and the Stigler Center both at
the University of Chicago Graduate School of Business for research support.
research was also fiinded in part by the
expressed in
this
of
Ewing Marion Kauffman Foundation. The views
paper are those of the authors and not necessarily of any of the
organizations with which
drafts
This
we
this
are affiliated or
paper
which sponsored
http://gsbwww.uchicago.edu/fac/anil.kashvap/research
this research.
be
will
.
First draft:
Future
posted
September 2003.
to
Zombie Lending and Depressed Restructuring
in Japan
Abstract:
In this paper,
we
propose a bank-based explanation for the decade-long Japanese
slowdown following
known
the asset price collapse in the early 1990s.
We
start
with the well-
observation that most large Japanese banks were only able to comply with capital
standards because regulators were lax in their inspections.
the banks often engaged in
sham loan
insolvent borrowers (that
we
call
of the congestion created by the zombies
their entry
is
lose
We
market share was thwarted.
Our
zombie problem. The counterpart
a reduction of the profits for healthy firms,
and investment.
not find good lending opportunities.
forbearance
zombies). Thus, the normal competitive outcome
the restructuring implications of the
which discourages
facilitate this
restructurings that kept credit flowing to otherwise
whereby the zombies would shed workers and
model highlights
To
In this context, even solvent banks will
confirm our story's key predictions that zombie-
dominated industries exhibit more depressed job creation and destruction, and lower
productivity.
We
present firm-level regressions showing that the increase in zombies
depressed the investment and employment growth of non-zombies and widened the
productivity gap between zombies and non-zombies.
1.
Introduction
This paper explores the role that misdirected bank lending played in prolonging
the Japanese
macroeconomic stagnation
began
that
in the early 1990s.
The
investigation
focuses on the widespread practice of Japanese banks of continuing to lend to otherwise
We
insolvent firms.
distorting effects
document
on healthy firms
Hoshi (2000) was the
that
that
began
is
its
first
phenomenon and
its
were competing with the impaired
paper to
call attention to this
firms.
by a number of observers of the Japanese
ramifications have been partially explored
economy. There
show
the prevalence of this forbearance lending and
agreement that the trigger was the large stock and land price declines
in early 1990s: stock prices lost roughly
60%
peak within three years, while commercial land prices
of their value from the 1989
fell
by roughly 50%
after their
1992 peak over the next ten years. These shocks impaired collateral values sufficiently
that
any banking system would have had tremendous problems adjusting.
the political and regulatory response
was
to
regulators were forced to recognize a
Aside from a couple of
to this rule is that
loan, they
were
against the
likely to
minimum
have
minimum
The
by
fear of falling
which
these firms
would recover
For instance,
or that the government
in 1997, at least 5 years after the
in turn
below the
to continue to extend credit to insolvent borrowers,
'
the regulators.
to call in a
many banks
Ministry of Finance was insisting that no public
periods
to
comply)
level of capital (the so-called
when banks wanted
to write off existing capital,
capital levels.
crisis
banks had to comply (or appear
with the international standards governing their
Basle capital standards). This meant that
Japan
few insolvencies and temporarily nationalize
the offending banks, the banks were surprisingly unconstrained
The one exception
in
deny the existence of any problems and delay
any serious reforms or restructuring of the banks.
when
But
would
bail
non-performing
pushed them up
capital standards led
gambling
them
that
out.^
somehow
Failing to
problem of non-performing loans was recognized, the
money would be needed to assist the banks. In February
1999 then Vice Minister of International Finance, Eisuke Sakakibara, was quoted as saying that the
Japanese banking problems "would be over within a matter of weeks." As
Services
Agency claimed
that Japanese
late as
2002, the Financial
banks were well capitalized and no more public money would be
necessary.
"
The banks
also tried to raise capital
by
issuing
more shares and subordinated
debt, as Ito
and Sasaki
(2002) document. When the banks raised new capital, however, almost all came from either related firms
(most notably life insurance companies) that are dependent on the banks for their financing, or the
rollover the loans also
the recession
would have sparked public
by denying
encouraged the banks
credit to
needy corporations.
to increase their lending to
were worsening
criticism that banks
Indeed, the government also
medium
small and
The continued
the apparent "credit crunch" especially after 1998.''
sized firms to ease
financing, or "ever-
greening," can therefore be seen as a rational response by the banks to these various
pressures.
A
simple measure of the ever-greening
is
shown
in Figure
percentage of bank customers that received subsidized bank credit.
of
how
the universe of firms considered here
real estate, retail,
is all
figure
30%
shows roughly
The lower
asset weighted figures, suggests that about
15% of assets
figures show, these percentages
were much lower
in the
these unprofitable borrowers (that
allowed them to distort competition throughout the
came
distortions
raising market
in
many ways,
defer the details
all that
matters
is
more
to the
panel,
of these firms were on
reside in these firms.
As
these
1980s and early 1990s.
we
rest
call
"zombies")
alive, the
of the economy.
banks
The zombies'
generally, congesting the markets
at the current
where they
came from guaranteeing
participated.
the deposits of
supported the zombies served as a very inefficient program to sustain
employment.
Thus, the normal competitive outcome whereby the zombies would shed
workers and lose market share was thwarted.
wages reduce
the profits that
More
new and more
importantly, the low prices and high
productive firms could earn, thereby
government when banks received capital injections. See Hoshi and Kashyap (2004, 2005) for more on
"double-gearing" between banking and life insurance sectors.
Subsequently when the Long-Term Credit Bank was returned
sale
was
the
life
which shows comparable
workers whose producfivity
Effectively the growing government liability that
^
that
including depressing market prices for their products,
wages by hanging on
firms declined and,
that
We
publicly traded manufacturing, construction,
support from the banks in the early 2000s.
By keeping
which reports the
wholesale (excluding nine general trading companies) and service
The top panel of the
sector firms.
banks
now
the firms are identified until the next section, but for
1,
new owners would maintain
lending to small and
to private
medium
this
ownership, a condition for the
borrowers. The
new owners
tightened credit standards and the government pressured them to continue supplying funds, see Tett (2003)
for details.
See Aheame and Shinada (2004) for some direct evidence suggesting that inefficient firms in the nonmanufacturing sector gained market share in Japan in the 1990s. See also Kim (2004) and Restuccia and
Rogerson (2003) for attempts to quantify the size of these types of distortions.
"
discouraging their entry and investment.
good lending opportunities
particularly
section,
we
in Japan.
we document and
In the remainder of the paper
even solvent banks saw no
In addition,
formalize this story.
There are a number of
describe the construction of our zombie measure.
potential proxies that could be used to identify zombies.
measurement problems confoxond most of these
In the next
As we
explain, however,
alternatives.
Having measured the extent of zombies, we then model
The model
their effects.
on
is
a standard variant of the type that
is
designed to contrast the adjustment of an industry to a negative shock with and without
is
studied in the literature
We model the presence
the presence of zombies.
surge in destruction that would arise in the
or credit shock.
The main
effect
sufficiently to re-equilibrate the
economy
is
"sclerosis"
— and
of that constraint
is
some of those
technological,
that job
creation
demand,
must slow
This means that during the adjustment the
economy.
the associated "scrambling"
that are less productive than
to the
wake of an unfavorable
Hammour
that
—
(1998, 2000) have called
would not be saved without
the preservation of production units that
banks' subsidies
It
of zombies as a constraint on the natural
characterized by what Caballero and
—
creative destruction.
the
the retention of firms and projects
do not enter or are not implemented due
congestion caused by the zombies.
In the fourth section of the paper,
the model.
In particular,
we
assess the
main aggregate implications of
study the interaction between the percentage of zombies in
the
economy and
We
find that the rise of the zombies has
the
we
amount of restructuring, both over time and across
been associated with
different sectors.
falling levels
of aggregate
restructuring, with job creation being especially depressed in the parts of the
We
with the most zombies firms.
performance measures.
In section 5
we
then explore the impact of zombies on sectoral
We find that the prevalence of zombies
lowers productivity.
analyze firm-level data to directly look for congestion effects of
the zombies on non-zombie firms' behavior.
growth for healthy firms
falls
We
find that investment
as the percentage of
zombies
in
and employment
their industry rises.
Moreover, the gap in productivity between zombie and non-zombie firms
percentage of zombies
activity the
most
economy
rises.
for the fastest
Most
strikingly, the presence
growing healthy
firms. All
rises as the
of the zombies depresses
of these findings are consistent
with the predictions that zombies crowd the market and that the congestion has real
effects
on the healthy firms in the economy. Simple extrapolations using our regression
coefficients suggest that cumulative size of the distortions (in terms of investment, or
employment)
is
substantial.
In the final section of the paper
we conclude by summarizing
our results and
discussing the implications of our findings for Japan's outlook.
2.
Identifying zombies
Our
story can be divided into
supporting zombie firms.
two
First, the
parts.
banks misallocated credit by
Second, the existence of zombie firms interfered with the
process of creative destruction and stifled growth.
Our measure of zombie should not
only capture the misallocation of credit but also be useftil in testing the effect of zombies
on corporate
profitability
2.1 Defining
There
in
and growth.
Zombies
is
a growing literature examining the potential misallocation of
Japan (see Sekine, Kobayashi, and Saita (2003) for a survey).
indirect.
Much of the
bank
credit
evidence
is
For instance, several papers (including Hoshi (2000), Fukao (2000), Hosono
and Sakuragawa (2003), Sasaki (2004)) study the distribution of loans across industries
and note that underperforming industries
bank
credit than other sectors that
like real estate or construction received
were performing
Peek and Rosengren (2005)
offer the
the potential misallocation of bank credit.
most
They
firms often increased between 1993 and 1999.
to lend to the firms than other
more
better (such as manufacturing).^
direct
find that
and systematic study
bank
credit to
to date
on
poor performing
These firms' main banks are more likely
banks dealing with these firms when the firm's profitability
Other indirect evidence comes from studies such as Smith (2003), Schaede (2005) and Jerram (2004) that
document that loan rates in Japan do not appear to be high enough to reflect the riskiness of the loans.
Sakai, Uesugi and Watanabe (2005), however, show that poorly performing firms (measured by operating
profits or net worth) still pay higher bank loan rates and are more likely to exit compared with better
performing firms, at least for small firms. Finally, see also Hamao, Mei and Xu (forthcoming) who show
that firm-level equity returns became less volatile during the 1990s and argue that this is likely due to a lack
of restructuring in the economy.
^
is
This pattern of perverse credit allocation
declining.
own balance
affiliated
sheet
is
weak
or
banks do not show
We
when
the borrower
is
more
likely
when
the bank's
a keiretsu affiliate. Importantly, non-
is
this pattern.
depart from past studies by trying to identify zombies by classifying firms
only based on our assessment of whether they are receiving subsidized credit, and not by
This strategy permits us to evaluate the
looking at their productivity or profitability.
effect
of zombies on the economy.
If instead
we were
zombies based on
to define
their
operating characteristics, then almost by definition industries dominated by zombie firms
would have low
this correlation,
profitability,
we want to
The challenge
and
test for
it.
approach
for our
have low growth. Rather than hard-wiring
likely also
to use publicly
is
determine which firms are receiving subsidized
little
incentive to reveal that a loan
which banks could
is
credit:
miss-priced.
available information to
banks and
To
get
borrowers have
Because of the myriad of ways
many ways
transfer resources to their clients, there are
attempt to measure subsidies.
their
some guidance we used
the Nikkei
that
we
in
could
Telecom 21
to
search the four newspapers published by the Nihon Keizai Shimbun-sha (Nihon Keizai
Shimbun, Nikkei Kin'yH Shimbun, Nikkei Sangyo Shimbun, Nikkei RyUtsU Shimbun)
between January 1990 and
assistance"
and
"reconstruction").^
either
May 2004
for all
news
articles containing the
"management reconstruction plan"
The summary of our
Our search uncovers 120
("corporation"
or
findings are given in Table
separate cases.
types of assistance that were included.
As
In most of
and
1.
them
the table shows,
words "financial
there
were multiple
between
interest rate
concessions, debt-equity swaps, debt forgiveness, and moratoriums on loan principal or
interest,
most of these packages involve reductions
in interest
payments or outright debt
forgiveness by the troubled firms.^
The decision by
a
bank
to restructure the loans to distressed
companies
in these
ways, rather than just rolling over the loans, helps reduce the required capital needed by
the bank.
Without such restructuring, banks would be forced
The Japanese phrases were Kin'yu Shien
'
These patterns
AND (Keiei
are consistent with the claim
by
Saiken Keikaku
to classify the loans to
OR (Kigyo AND
Saiken)).
Tett and Ibison (2001) that almost one-half of the public
funds injected into the banking system in 1998 and 1999 were allowed to be passed on to troubled
construction companies in the form of debt forgiveness.
those borrowers as "at risk", which usually would require the banks to set aside
With
the loan value as loan loss reserves.
70%
of
banks need only move the
restructuring, the
loans to the "special attention" category, which requires reserves of at most 15%.
In light of the evidence in Table
We
involves a direct interest rate subsidy.
hypothetical lower
bound
quality borrowers.
We then compare this
we make
Finally,
between actual
we
concentrate on credit assistance that
proceed in three steps.
payments (R
for interest
several econometric
interest rate (r)
1,
)
we
calculate a
we
expect only for the highest
to the
observed interest payments.
that
lower bound
First,
assumptions to use the observed difference
and notional lower bound
rate (r
)
to infer cases
where we
believe subsidies are present.
2.2 Detecting
Zombies
The minimum required
interest
payment
for each firm each year,
R *,
is
,,
defined
as:
5
f
R%=>'s,.,BS,,,.,+
where BS.
,
,
BL.,_,
BL. and Bonds.
,
bank loans (more than one
,
+ 7'cZ>„i„ „,„ ,^,
are short-term
and
year),
total
5
y^^
,
*
Bonds,, _,
bank loans
(less
bonds outstanding (including convertible
bonds (CBs) and warrant-attached bonds) respectively of firm
rs,
,
rl,
,
and rcb^m
over the
last 5 years,
I
bond issued
i
at the
end of year
are the average short-term prime rate in year
average long-term prime rate in year
convertible corporate
than one year), long-term
t,
and the minimum observed coupon
in the last five years before
(short-term
we know
about the firms' debt structure
bank borrowing, long-term borrowing
that are
is
t,
and
the
on any
t.
This estimate for the lower bound reflects the data constraints
particular, all
rate
t,
we
face.
In
the type of debt instrument
due
in
one year and remaining
long-term bank borrowing, bonds outstanding that are due in one year and remaining
bonds outstanding, and commercial paper outstanding). In other words, we do not know
the exact interest rates on specific loans, bonds or commercial paper, nor do
we know
the
exact maturities of any of these obligations.
measure include
all interest,
payments we can
Finally, the interest
and discount expenses, including those related
fee
to trade
credit.
The general
principle guiding the choices
we make
^*
are extremely advantageous for the borrower, so that
would pay
firms
in the
absence of subsidies.
financing takes place at rcbmin
over the
last 5 years,
I
fact less than
is in
we
that firms
assuming not only
are
interest rates
bonds are issued when rates are
To
we examined
categorize firms
for robustness
we compare
(Ri,,)
with our hypothetical lower bound.
total
borrowing
where
CP,,,.;
at the
the
is
we
measure
is
the alternative
made by
x^
,
"conservative" because
s
—
Ri,
=
(5/.,_/
55,.
,_,
+ 5Z,. ,_, +
5o;7£fe,.
Note
+CP,,,./),
at the
/
Accordingly
-R*
—-
we assume
r.
^-r.,, as the interest rate gap. This
minimum
the
expenses on items beyond our concept of
expenses on trade
,_,
for the firm
interest
rates
that are
Rt,,
includes
extremely advantageous to the firm and because the interest payment,
interest
the firms
normalize the difference by the amount of
so that the units are comparable to interest rates.
refer to the resulting variable,
We provide
1.
amount of commercial paper outstanding
t,
to the conversion
R* and
the actual interest payments
beginning of the period
beginning of the period
due
at their lowest.
check in Appendix
We
bond
that
additional discussion of the data choices used in constructing
approaches that
what most
For instance, by assuming
borrow using convertible bonds (which carry lower
option), but also that these
to select interest rates that
is
total
borrowing (such as
interest
credit).
that given our procedure to construct
/-*
we
will not
be able
to detect all
types of subsidized lending.^ In particular, any type of assistance that lowers the current
period's interest payments can be detected: including debt forgiveness, interest rate
concessions, debt for equity swaps, or moratoriums on interest rate payments,
which appeared
to
be prevalent in the cases studied
in
Table
1
.
On
all
of
the other hand, if a
In addition to the cases studied below, Hoshi (2006) examines the potential problems that might arise
from rapid changes in interest rates. For example, if interest rates fell sharply and actual loan terms moved
as well, then our gap variable could be misleading about the prevalence of subsidized loans. He constructs
an alternative measure (that would be more robust to within year interest rate changes) and concludes that
this sort of problem does not appear to be quantitatively important.
bank makes new loans
loans, then our
from a
assets
We
t
client at overly generous prices our
if a
to
pay off past
bank buys other
proxy will not detect the assistance.
Our baseline procedure
its
interest rate
gap
is
negative
classifies a firm
<
(x,,
The
0).
/
zombie
as a
measured lower bound, then only a firm
that receives a subsidy
If r*
zombie remains even under
Thus we
non-zombies. In
a
as non-
this perfect scenario.
resort to a second approach,
this
is
can have a
However, the problem of labeling a firm with xu just above zero
negative gap.
for
justification for this
the conservative philosophy underlying the construction of r*.
is
perfectly
used
explore two strategies for identifying the set of zombie firms from the
whenever
strategy
interest rates that are then
gap variable will not capture the subsidy. Likewise,
calculated interest rate gaps.
year
normal
to a firm at
second approach
which
we assume
is
more robust
that the set
of
to misclassification
of zombies
is
a "fiizzy"
set.
In the classical set theory, an element either belongs or does not belong to a particular set
so that a 0-1 indicator function can be used to define a subset.
In contrast, in fuzzy set
theory an element can belong to a particular subset to a certain degree, so that the
indicator fiinction can take any value in the interval [0,
indicator function are confined to {0, 1}, a set defined
Using
a "crisp" set.
"crisp."
some
this
terminology, our
Our second approach, on
1].
When
by the indicator function
approach assumes the
first
the images of the
the other hand, assumes the set
is
set
is
called
of zombies
is
"fuzzy," allowing
firms to be more-or-less zombie-like.^
The
function,"
indicator
which we assume
1
dn
z{x;d^,d^)^
^
-X"
a fuzzy
function that defines
to
if X
ift/,
be
<
(for the set
subset
is
called
"membership
of zombie firms):
£/,
<x<J2
whereJ,
<0<J2
(1)
dj -d^
if X
The shape of
the
> d^
membership fimction
is
determined by the two parameters, dj and
d2.
Figure 2 shows this membership function along with the indicator function implicit in our
first
approach.
when
'
di
It is
easy to see the second approach degenerates to our
and d2 are both zero.
See Nguyen and Walker (2006)
for an introduction to the
fuzzy
set theory.
first
approach
.
The second approach
is
appealing given the fUzzy nature of the concept of
"zombie firms." These are defined
from
their creditors to survive in spite
how much
to specify
to
to
much more
acknowledges
zombie
status
financial help
information than
this limitation
be those firms that receive sufficient financial help
of their poor
is
profitability.
we do about
individual firms.
and assigns numbers between
case,
{di, dj)
we assume
50 basis points
=
1
we had
access
Our fuzzy approach
to those firms
whose
in the construction of r*,
we
=
(0,
what follows we show
results for {dj, dj)
(-25bp, 75bp), where bp stands for basis points.
Thus, in the
first
a firm with x„ below zero is a definite zombie and a firm with x„ above
is
definitely a
non-zombie: any firm with
points has "zombiness" between
2.3
and
if
ambiguous.
is
that di is closer to zero than d?. In
50bp) and
inherently difficult
considered to be sufficient, even
Given the asymmetry (toward conservatism) inherent
assume
It is
and
x,,
between zero and 50 basis
1
Quantifying the prevalence of zombies
Figure
1
shows the aggregate estimate of the percentage of zombies using our
baseline procedure.
As mentioned
earlier, treating all firms
equally
we
see that the
percentage of zombies hovered between 5 and 15 percent up until 1993 and then rose
sharply over the
mid 1990s
year after 1994.
zombies
is
so that the zombie percentage
was above 25 percent
for every
In terms of the congestion spillovers, a size weighted measure of
likely to
be more important.
Weighting firms by
their assets
we
see the
same
general pattern but with the overall percentage being lower, closer to 15 percent in the
latter part
of the sample.
We
plausibility
Figure
1
view the cross-sectional prevalence of zombies as another way
of our
definition.
into five industry
To conduct
this
assessment,
we
largest general trading companies),
recall that all the firms included here are publicly traded.
industry
is
aggregated the data used in
groups covering manufacturing, construction, real
and wholesale (other than the nine
to assess the
estate, retail
and services -
The zombie index
for an
constructed by calculating the share of total assets held by the zombie firms
-
10
and
for the
remainder of the paper
we
concentrate on asset weighted zombie indices.
addition to showing the industry distribution,
implied by our second procedure with
{di, dz)
we
=
(0,
compute the zombie percentages
also
50bp) and
(c//,
d2)
=
(-25bp, 75bp).
We
Figure 3 shows the zombie index for each industry from 1981 to 2002.
three
main conclusions from these graphs.
shows
Starting with the upper left
baseline case) and the two fuzzy measures share similar time series
between the
crisp
draw
hand panel
that
zombie measure (our
the data for the entire sample, first notice that the crisp
correlation
In
movements (with
measure and the two fuzzy measures exceeding
the
0.99).
Second, the other five panels show that the proportion of zombie firms increased in the
late
1990s
more
in
every industry.
serious
The
third
non-manufacturing
for
key conclusion
firms
than
that the
is
zombie problem was
manufacturing
for
firms.
In
manufacturing, the crisp measure suggests that zombie index only rose from 3.11%
(1981-1993 average) to 9.58% (1996-2002 average).
In the construction
industry,
however, the measure increased from 4.47% (1981-1993 average) to 20.35% (1996-2002
average).
Similar large increases occurred for the wholesale and
retail, services,
and
real
estate industries.
There are a variety of potential explanations for these cross-sectional differences.
For instance, Japanese manufacturing firms face global competition and thus could not
easily be protected without prohibitively large subsidies.
For example, many of the
troubled Japanese automakers were taken over by foreign firms rather than rescued by
their
banks during the 1990s.
In contrast, there
is
very
little
foreign competition in the
other four industries.
A
sectors.
second important factor was the nature of the shocks
For instance, the construction and
real estate industries
hitting the different
were forced
the huge run-up and subsequent collapse of land prices mentioned earlier.
to deal with
Thus, the
adjustment for these industries was likely to be more wrenching than for the other sectors.
But the most important point about the differences shown
in Figure 3
is
that they
confirm the conventional wisdom that bank lending distortions were not equal across
sectors
and
that the
problems were
for further discussion.
less acute in
manufacturing - see Sekine
et al
Thus, regardless of which explanation one favors as to
(2003)
why
this
11
might be the case,
this
we view
as particularly reassuring that our
it
zombie index confirms
conventional view.
Figure 4, our
last plausibility
zombies
for the firms that are
To keep
the graphs readable
show
similar patterns.
check, shows the asset weighted percentages of
above and below the median
we show
only the crisp measures, but the other measures
In manufacturing the differences are not very noticeable, with
In the remaining industries,
fewer high profit firms being labeled as zombies.
slightly
profit rate for their industry.
particularly in real estate
and construction,
it
appears that our measure of zombies
is
identifying firms that are systematically less profitable than the non-zombies, particularly
fi-om the
3.
mid- 1 990s onward.
A model of the effect of zombie firms on restructuring
To analyze
the effect of zombies
we
study a very simple environment that
involves entry and exit decisions of both incumbent firms and potential
we
later
start
with a normal environment where
the operating profits from running a firm.
where some incumbent firms
3.1
to
We
which
all
As
a
decisions are based purely on
then contrast that environment to one
(for an unspecified reason) receive a subsidy that allows
remain in business despite negative operating
profits.
The Environment
The
indexed by
units.
essential points
i ).
A
of interest can be seen in a model where time
(representative) period
The productivity of
productivity for firm
yl
firms,
extend to analyze expansion and contraction decisions of existing firms.
benchmark we
them
new
i
t
starts
is
discrete (and
with a mass m^ of existing production
the incumbents varies over time and the current level of
in year
t,
y^^, is:
= A+ ^,
12
,
where
e"^
an idiosyncratic shock that
is
main predictions from
shocks, so
we assume
this
is
distributed uniformly
model do not depend on
these shocks are
on the unit
The
interval.
the persistence of the productivity
i.i.d.
In addition to the incumbents, there are also a set of potential entrants and
normalize their mass to be
The
V2.
The productivity
before deciding whether to enter or not.
year
/is:
with
B
>
assumed
new
and g^ distributed uniformly on the unit
to
more productive (and more
for i* potential
interval.
have no persistence. These assumptions imply
firms will be
draw a productivity
potential entrants each
that
is
an entry cost,
k >
,
Finally, both
where
A'^,
that they
new and
represents the
N and
new
The shock g^
firm in
is
again
on average the potential
we
also
assume
one
that there
to start up.
old units must incur a cost p{N^) in order to produce,
number of production
of the existing units that do not
respect to
must pay
level, y^
profitable) than the incumbents (for
period only, then they become incumbents as well). However,
we
exit
and new
units in operation at time
entrants.
The
cost
p{N)
is
t
,
i.e.,
the
sum
increasing with
Indeed p(N)
captures any scarce input such as land, labor or capital.
captures any reduction in profits due to congestion or competition. '° For our purposes,
all
the predictions
we emphasize
will hold as long as
continuous function of A^. For simplicity,
p{N,) =
we
p{N)
is
a strictly increasing
adopt the linear fiinction:
N,+M-
'"
For example, we can motivate jOfTV) as the reduction in profits due to competition in the output market.
Suppose the price of output is given by D'' (N), a decreasing function ofN, and that the cost of production
for each production unit is a constant, C. Under our assumption on productivity, an incumbent decides to
stay in the market (and a potential entrant decides to enter the market) if D'' (N){A+
equivalently,
A+
s
-C/D''{N) >
0.
In this specific sxample,
p(N)
is
i) -C >
C/D''{N), which
is
0,
or
increasing with
respect to N.
13
,
where the
intercept
// is
potential shift variable that captures cost changes
and other
profit
shocks.
3.2 Decisions
This basic model will quickly generate complicated dynamics because the
existing firms
new
main
have paid the entry cost and thus face a different decision problem than the
firms for which the entry cost
predictions, so
we assume
is
that
not sunk. These dynamics are not essential for our
B
= k
.
In this case, the exit decision
and the entry decision by potential entrants become
shocks are
i.i.d.
investment
is
and there
exactly offset
is
no advantage
by a lower
fully
by incumbents
myopic. Since productivity
fi'om being an insider (the
sunk cost of
productivity), both types of units look only at
current profits to decide whether to operate.
Letting
and y" denote the reservation productivity of incumbents and
y°
potential entrants, respectively,
y"
we
have:
-K- p{N) = 0.
In this case
it
is
straightforward to find the
mass of
exit,
D^
,
and
entry,
Hf
respectively:
A
='>^i
1-
di
f
= mXv{N,)-A),
=
''^=\L-/'
V'-^^^^^)-^))2
2
(2)
(3)
Jp(iv,)-,4
Adding
units created to the surviving incumbents yields the total
operating at time t
number of units
:
14
N,=H,+m,-D,=(^ + m}(l-{p{N,)~A)).
3.3
Equilibrium and Steady State
We
first
to
(4)
step
is
can
now
solve for the steady state of the normal version of the economy.
to replace
p{N) with
be composite shock that
costs (higher
/u)
is
N + /u
equal to
in (4).
A-/.i
.
The notation
Note
is
The
simplified if we define S
lower S indicates either higher
that a
or lower average productivity (smaller A).
This yields the equilibrium
number of units:
^l/2 + m,^
N,=
+
(l
S).
(5)
3/2 + mj
Given
the total
number of operating
destruction and creation
A
l/2 + m^
by
we can
substituting (5) into (2)
and
solve for equilibrium rates of
(3):
-S
=T^t
(6)
3/2 + m,
H=-
1
+5
(7)
3/2
+ m,
The dynamics of this system
^M
units,
= Nf
are determined by:
(8)
In steady state, the mass of incumbents remains constant at
m" =
iV"
,
which
requires that creation and destruction exactly offset each other or, equivalently, that
15
.
m^ = N^. Using the
latter
condition and
has a unique positive solution
which
2(1
+ 5)
we
can approximate the above by:
3
2
In our subsequent analysis
and
state
A
we
will
that the initial (pre-shock) value
corresponding steady state will be
nif^
assume
of
that the
S, So, is 0.
= N^ = 1/2 and
i/g
economy begins
Given
=
£>(,
in a
steady
this normalization, the
=1/4.
(permanent) Recession
We
now
can
construction the
changes
analyze the adjustment of the economy to a profit shock.
model
treats
in n, as equivalent.
aggregate productivity
shifts,
changes
in
By
A, and cost shocks,
So what follows does not depend on which of these occurs.
separate the discussion to distinguish between the short- and long-run impact of a
decline in S from
we mean
i",,
for a fixed
has adjusted to
It is
dS
,
2
^-1 + -S.
to"
We
m"
=
For small values of 5,
3.4
a quadratic equation for
of:
+
m
(5), yields
A
its
=
to 5,
m = mo =
new
<
(lower productivity or higher costs).
1/2.
By
the "long-run,"
steady state value
rrij
=1/2 +
By
the "short-run"
on the other hand, we mean
after
m
(2/ 3)S^
easy to see from (6) and (7) that in the short-run:
dS'
16
That
is,
when S
drops, creation falls and destruction rises, leading to a decline in
economy, negative
In other words, in a normal
(4)).
profit
A'^
(see
shocks are met with both
increased exit by incumbents and reduced entry of new firms.
Over time,
incumbents
(recall
between destruction and creation reduces the number of
the gap
from
(4)
and
(8) that
AN=H-D), which
lowers the cost of inputs
(p(N)) and eventually puts an end to the gap between creation and destruction caused
by
the negative shock.
Across steady
dm
dN
2
dS
dS
2,
states,
units falls
is
impact as time goes by and the
that since
A'^
falls
not enough to offset the direct effect of a lower S on creation. That
creation falls in the long run.
And
run, the initial surge in destruction
falling
below
its
is,
since creation and destruction are equal in the long
is
temporary and ultimately destruction also ends up
pre-shock level."
Zombies
Suppose now
that
destruction brought about
be accomplished.
We
"banks" choose to protect incumbents from the
by the decHne
assume
the additional units that
'
initial
than one for one with S, the long run reduction in the input cost due to reduced
competition
'
beyond the
between destruction and creation closes gradually. Note
positive gap
3.5
that:
^
The number of production
less
we have
that the
in S.
banks do
initial
There are a variety of ways that
this
surge in
this
might
by providing just enough resources
would have been scrapped so
that they
to
can remain in operation.
undone when creation and destruction are measured as ratios over A', as is
However, the qualitative aspects of the short run results are preserved since
divided by either initial employment or a weighted average of initial and final
This long run level effect
is
often done in empirical work.
empirically the flows are
employment.
17
With
this
assumption, a firm that does receive a subsidy
is
indifferent to exiting
and
operating, and thus entry and exit decisions remain myopic.
The maximum
would show
have
short run effect
would be on impact, when
a spike in destruction (see (5)).
Under
the normal
economy
the zombie-subsidy assumption,
we
that:
Di=Do-\-
The post-shock
destruction remains the
same
adjustment on the destruction margin means that
Ni=H;^+m,-l/4
Replacing
this
=
as the pre-shock level.
now
creation
Hi+l/A.
expression into
(3),
we
must do
all
The lack of
the adjustment:
(10)
can solve out for H:
This can be compared to the impact change in creation that occurs in the absence of
zombies. Doing
dHi
we
see:
^1^1^ dH,,
dS
That
so,
4
3
is,
zombies.
dS
a decline in S has a
This result
is
much
larger negative effect
a robust feature of this type of model.
quahtative prediction would hold even if
allowed persistence
in the productivity
productivity advantage of
new
shock causes the labor market
suppressed,
then the
on creation
to clear
presence of
In particular, the
same
not suppressed the dynamics and had
shocks and a gap between entry costs and the
Intuitively, this is the case
firms.
labor market
we had
in the
because the adverse
with fewer people employed.
clearing
can only occur
if
If destruction
is
job creation drops
precipitously.
18
—
As Caballero and Hammour
(1998, 2000) emphasize, both this "sclerosis"
preservation of production units that
and the associated "scrambling"
some of those
—
do not enter due
that
would not be saved without
the banks' subsidies
to the congestion
when
a normally functioning
caused by the zombies - are robust
there are fi-ictions against contracting.
economy, we have shown the existence of
zombies softens a negative shock's impact on destruction and exacerbates
What
creation.
dS
That
3
is,
is
the net effect
the
the retention of firms that are less productive than
implications of models of creative destruction
Compared with
—
on the number of firms?
It is
its
impact on
straightforward to show:
dS
2
in response to a negative shock,
that in the presence of
# falls by less if there are zombies, which means
zombies the reduced destruction
This
additional drop in creation.
because as job creation
falls,
is
is
not fully matched by the
another intuitive and robust result.
the marginal
entrant's
This occurs
productivity rises.
This high
productivity allows the marginal entrant to operate despite the higher cost induced by
(comparatively) larger N.
A
(net
final
important prediction of the model
is
the existence of a gap in profitability
of entry costs) between the marginal entrant and the marginal incumbent when there
are zombies.'^
At impact,
the destruction does not change, so that all the firms with
idiosyncratic productivity shocks above the old threshold (1/2) remain in the industry.
On
the other hand,
new
entrants have to clear a higher threshold to
negative shock in S (which
negative shock).
As
is
compensate
for the
only partially offset by the lower congestion following the
a resuh, the profitability of the marginal entrant
higher than that of the marginal incumbent. The difference
is
is
inefficiently
given by:
Note that a wedge like this one also arises when there is a credit constraint on potential entrants but not
on incumbents. In our model depressed entry results from the congestion due to zombies, and the gap is due
to the subsidy to incumbents. Clearly, however, if the two mechanisms coexist they would reinforce each
other, as congestion would reduce the collateral value of potential entrants.
'^
19
'x+vl-s,
In
3
2
summary,
the
model makes two robust
predictions.
The
first
is
that the
presence of zombies distorts the normal creation and destruction patterns to force larger
Second,
creation adjustments following shocks to costs, productivity or profits.
distortion depresses productivity
as the
inefficient units at the
expense of more
Accordingly, productivity will be lower
productive potential entrants.
more zombies and
by preserving
this
when
there are
zombies become more prevalent they will generate larger and
larger distortions for the non-zombies.
By
how
slightly re-interpreting
by zombies
the congestion effects caused
profitability. Instead
of a
what a "firm" means
set
of projects, some of which are
projects that are hit
which projects
to
will affect firms with different levels
in place
that
many new
Higher
become
(exits)
and
which
new
that firms differ in the quality
many
projects,
projects
of their projects.
start
(entries).
In particular,
some
Low profitability
firms
this
very
however, are more likely to have some new projects
however, could be non-monotonic because
projects might
to
and the presence of zombies may not influence
profitability firms,
its
in
projects that are unusually profitable, but
profitable each period that might be
projects, then
of
(incumbents) but the others have not been
other (low profitability) firms have only a few profitable projects.
much.
can also see
by productivity shocks every period and firms are deciding
(high profitability) firms have
will not start
we
Then, the above model can be re-interpreted as a model
terminate
Suppose further
some
model,
of assuming that a firm has only one project, suppose a firm consists
started (potential entrants).
which
in our
still
crowded out by the zombies. This
if a firm
be worth
effect,
has a sufficiently good mix of
initiating.
We
will also test for
whether
higher quality firms are disproportionately harmed by the zombies, but (because of the
potential non-monotonicity)
we
see this prediction as less robust than the previous two.
20
—
4.
The
effect of zombies
We
that the
on job creation, destruction and productivity
use the two robust predictions of the model to guide our search for evidence
zombie problem has affected Japan's economic performance
begin by looking
at
aggregate cross-industry differences.
firm-level data to characterize
how
the behavior of the
significantly.
In the next section,
we
We
study
non-zombie firms has been altered
by the presence of zombie competitors.
Because our zombie indices
exist
from 1981 onwards, we
start
by
calculating the
average of the crisp zombie index for each industry from then until 1993 and compare
We
that to the average for the late 1990s (1996-2002).
use the differences in these two
averages to correct for possible biases in the level of zombie index and any industryspecific effects.
It
makes
particular, the results
little
difference as to
we show would be
how we
define the pre-zombie period. In
very similar if we took the normal (non-zombie)
period to be 1981 to 1990, or 1990 to 1993. Our evidence consists of relating creation,
destruction,
and productivity data
these measures
are
more
to this
change in the zombie index,
in the
distorted
industries
in order to see if
where zombie prevalence has
increased the most.
Our most
direct evidence
on
this point is in
creation and destruction against the change in the
measures constructed by Genda
in
our model.
et al.
Figure
5,
which
zombie index.
plots the rate of job
We
use the job flow
(2003) as proxies for the concepts of entry and exit
Their measures are based on The Survey of Employment Trends,
conducted by the Ministry of Welfare and Labor biannually on a large sample of
establishments that employ five or more regular workers.
The
series
used for our
analysis include not only the job creation (destruction) at the establishments that were
included in the survey in both at the beginning and
estimated job creation (and destruction) by
exited).
To
new
end of the year, but also the
entrants (and the establishments that
control for the industry specific effects in job creation/destruction,
the difference
between the average job creation (destruction)
period and the average for the 1991-1993 period.
93 data as a control because figures of Genda
because that
at the
is
We
rate for the
we
look
at
1996-2000
et al. start
are restricted to using the 1991
only in 1991 and
we
stop in
2000
the last year they cover.
21
The top of Figure
from
5
shows
that the job destruction rate in the late 1990s increased
1990s in every industry, as
that in the early
More
unfavorable shock to the economy.
was smaller
destruction
in the industries
expected, the presence of zombies slows
The second panel of Figure
5
we would
expect to see following an
importantly, the graph
shows
that the surge in
where more zombies appeared.
down job
shows
we
Thus, as
destruction.
that the presence
of zombies depresses job
Creation declined more in the industries that experienced sharper zombie
creation.
In manufacturing,
growth.
which suffered the
least
from the zombie problem, job
creation hardly changed from the early 1990s to the late 1990s. In sharp contrast, job
creation exhibits extensive declines in non-manufacturing sectors, particularly in the
construction sector.
Of
course not
slowdown
prices and the
that followed
disproportionately from the
shock during the 1990s.
were equally affected by the Japanese crash
sectors
all
A
boom
it.
in asset
For example, construction, having benefited
years, probably also
was
hit
by
the largest recessionary
large shock naturally raises job destruction
and depresses job
creation further. Despite this source of (for us, unobserved) heterogeneity, the general
patterns
shock
is
we
expected from job flows hold.
by checking whether
through job creation
more zombie-affected
larger. In this metric,
is
creation has borne a
in
One way of
much
larger share
it
is
controlling for the size of the
sectors, the relative adjustment
quite clear
from Figure
5 that job
of the adjustment in construction than
in
manufacturing.
Our evidence on productivity
given in Figure
6.
distortions caused
In the model, zombies are the
by
the interest rate subsidies
low productivity
units that
would
is
exit
the market in the absence of help from the banks. Their presence lowers the industry's
average productivity both directly by continuing to operate and indirectly by deterring
entry of
more productive
firms.
The productivity data here
Harada (2004) who study productivity growth
average grow^th of the
'^
Our simple model assumes
zombie
industry.
rescued by banks.
It is
in
total factor productivity
22
are
indusfries.
from Miyagawa,
Figure
(TFF) from 1990
that the job destruction rate stays the
same even
6,
to
2000 against the
90%
None of the major results would change. Job destruction would
much as it would under the normal environment.
and
which plots the
after a negative
straightforward to relax this by assuming, for example, that
Ito
rise
shock in a
of zombies are
following a
negative shock but not as
22
change in the crisp zombie index, shows that the data are consistent with the model's
implication: the regression line in the figure confirms the visual impression that industries
where zombies became more important were the ones where TFP growth was
5.
worst.'"*
Firm-level zombie distortions
We
read the evidence in Figures 5 and 6 as showing that zombies are distorting
ways
industry patterns of job creation and destruction, as well as productivity in the
suggested by the model.
To
model's predictions,
test directly the
level data to see if the rising presence
of zombies
in the late
we
next look
at firm-
1990s had discernible effects
on the healthy firms (which would suffer from the congestion created by the zombies).
The data we analyze
are
from the Nikkei Needs Financial dataset and are derived
from income statements and balance sheets
Tokyo Stock Exchange.
sections of the
on the
for firms listed
The sample runs from 1981
contains between 1,844 and 2,506 firms depending on the year.
variables:
employment growth (measured by
the
number of
We
first
to
and second
2002, and
it
concentrate on three
full-time employees), the
investment rate (defined as the ratio of investment in depreciable assets to beginning of
year depreciable assets measured
(computed as the log of
employment).
In
all
sales
book
at
minus
1/3
value),
and a crude productivity proxy
minus 2/3 the log of
the log of capital
the regressions reported
below we dropped observations
in the top
and bottom 2.5% of the distribution of the dependent variable.
The simplest regression
Activity .j,=
where
activity
5'Dj,
we
y^nonz.^,
study
dummy
is
Of course
%
+ ^
i^,
(11)
percentage change in employment, or
of annual indicator variables and a
the probability that the firm
the percentage of industry assets residing in
'*
(^nonz^,
rate, the
Djt includes a set
variables, nonzijt
is:
+ jZ^, +
can be either the investment
our productivity proxy,
industry
+
that
zombie
is
set
non-zombie, and
of
2,, is
firms.
could arise because industries that had the worst shocks wound up with the
can disentangle these explanations by using firm-level data (see below).
this correlation
most zombies.
We
23
Because of the reduced form nature of both the regression equation and the
modeling of the subsidies
zombies,
to the
coefficients in these regressions.
For instance,
unspecified aggregate shocks. Likewise,
so large that they
firms; so
we do
we do
we
not attempt to interpret most of the
include the year
we can imagine
wind up investing more
(or adding
that the
dummies
to
allow for
zombies' subsidies are
more workers) than
the healthy
not propose to test the theory by looking at the estimates for p, the
coefficient for the non-zombies.
productivity specification the
The one exception
model
to this general principle
clearly predicts that
is
that for the
non-zombies will have higher
average productivity than zombies.
We
rising
instead focus on
what we see
zombie congestion should harm
would be negative
as the novel prediction of the theory: that the
the non-zombies.
This prediction suggests that
cp
investment and employment regressions, and be positive in the
in the
productivity specification. Note that for the investment (employment) specification one
might normally suspect
that as the percentage
of sick firms in the industry
rises, the
healthy firms would have more (relative to the sick ones) to gain from investing
(expanding employment). Thus, under normal circumstance there would be good reasons
to expect
(p
to
be positive rather than negative.
The main
reason, other than ours,
percentage in the industry
To
operating in the industry.
First,
somehow
is
for finding negative
cp
is
if the
zombie
standing in for the overall (un)attractiveness of
this potential objection to
our results
we
note two things.
our definition of zombies, by virtue of only using interest rate payments, does not
guarantee
percentage
that
is
growth opportunities are necessarily bad just because the zombie
high. Second, in order to be consistent with our findings, the reaction to
industry conditions must be different for zombies and non-zombies. In particular, non-
zombies must be more affected by an industry downturn than zombies
negative.
Nonetheless,
we
specification.
is
to
cp
to
come
out
seek to find other controls for business opportunities for the
healthy firms to minimize this potential omitted variable bias.
address this problem
for
Our main
add current sales growth of each firm
Thus, our alternative regression
control to
to the regression
is:
24
Activityy,=
5'Dj,+ ^nonz.j, +^Zj,
V^nonz,j,*Sij,
where
Syt is
The
7t
is
0s,.^
+ ^^i/Z., + 7rnonz.,^^*Z.*s^^^ +
+
v,
(12)
the growth rate of sales and the other variables are defined as in equation (11).
coefficient
different
n
in (12) reveals
from zero, then
growing healthy firms are
mentioned
+ ^onZy,*Zj, +
earlier,
distortions should
implies that faster growing healthy firms and slower
it
differentially affected
a natural
interpretation
by
(11) for the crisp
of the model suggests that the zombie
fourth columns of Table 2
zombie index.
First, as
We
draw
As
the presence of the zombies.
be larger for the healthiest firms. This would be the case
The second through
specification.
an additional potential effect for the zombies. If
three
shows our estimates
main conclusions
if Tt
<
0.
for equations
fi-om this simple
predicted by the theory, increases in percentages of zombie firms
operating in an industry significantly reduces both investment and employment growth
Our second
for the healthy firms in the industry.
finding,
shown
in
column
4, is that the
non-zombies have significantly higher productivity than the zombies. Finally, the same
column shows
that the productivity
gap between zombies and non-zombies
percentages of zombies in an industry
rises.
rises as the
These findings are consistent with the main
predictions of our model.
As mentioned above,
equation
(1
1
) is
a competing explanation for the sign of the estimated
that the industry
zombie percentage
is
in
an indirect measure of the growth
opportunities in the industry, even for the healthy firms.
We
including controls that directly capture growth opportunities.
estimates of equation (12),
cp
which include contemporaneous
address this concern by
Columns
5
and 6 report
firm-specific sales growth as
the potential growth proxy; for the investment specification, this type of accelerator
specification generally performs quite well in a-theoretic horse-races
specifications (see
Bemanke, Bohn and Reiss
among competing
(1988)).
We ran a similar regression using investment rates for US firms covered in the Compustat database
between 1995 and 2004. In this regression cp was insignificantly different from zero. The limited
information on debt structure in Compustat no doubt introduces noise in zombie assignments and we did
explore many alternatives to deal with this. But this result suggests to us that there is not a mechanical
'^
reason to find that
cp is
significantly negative in this type of regression.
25
.
In both columns the estimated coefficient
and
in
each equation the/?
columns 2 and
(p's
3.
is
on
sales
growth
highly significant,
is
nearly twice as high as that in the simpler specifications in
In the specifications with sales growth, the estimated magnitude of the
drops compared to
the
simpler
specifications,
but they
remain negative and
This indicates that while some of the interaction term's significance
significant.
have been due
More
to omitting proxies for
growth opportunities,
it is
may
not the sole reason.
substantively, in both of these specifications the estimated values for n are
This
significantly negative.
triple interaction
suggests that the fastest growing non-
zombie (healthy) firms are the most impaired by the widespread presence of zombie
firms in their industry.
In
Appendix 2 we
report a long
list
of robustness exercises, including
fiizzy
versions of equations (11) and (12), regressions omitting marginal zombies, as well as
using different measures of minimum required interest rates in the construction of zombie
and some of the point estimates vary across
indicators. Wliile the level of significance
these multiple scenarios, the general flavor of the results does not.
estimates for
cp
More
tend to be negative and significant for the investment and employment
The estimates of
regressions and positive and significant for the productivity regressions.
71
more
are
specifically, the
sensitive to the exact specification,
and vary more
employment
for the
regressions than for the investment specifications.
In the remainder of our discussion
firms on investment and
manufacturing
we
industries,
employment
alternative
estimate
if there
attempt to quantify the impact of zombie
employment growth of non-zombies.
where
our
particularly high in the late 1990s.
industries,
we
weighted
asset
We
focus on the five non-
of zombies
measures
For a typical non-zombie firm
how much more
the
in each of these
non-zombie would have invested or increased
had not been so many zombies
low zombies scenarios. In "Case
1,"
in the industry.
we assume
that the
We
consider two
zombie index stayed
at its
average value from 1981 through 1992 for each industry and calculate
more
a typical
years.
" More
by (z +
how much
non-zombie firm would have invested (or employed) over the next ten
In "Case 2,"
we assume
specifically, the investment (or
(p){actiial
were
zombie
that the
zombie index
employment)
index - alternative
is
for the industry
was
the
same
as
estimated to have been higher than the actual level
zombie
index)
26
from 1993
that for manufacturing for each year
We
to 2002.
investment under these two scenarios and compare
it
calculate the cumulative
investment (defined as the average of the median rates) during
employment, we compare the cumulative decline attributable
typical annual
amount of annual
to the typical
this
to the
zombies with the
change over the period (again defined as the average of the median
we
In all of these calculations
For
period.
rates).
take the regression estimates based on the crisp zombie
indices in Table 2, and ignore any feedback from industry equilibrium considerations.
Table 3 shows both investment and employment growth in non-zombie firms
would have been higher
in all these industries
had there been
the wholesale industry the cumulative investment loss
slightly
typical
higher by 3.0 percentage points
risen (which can
be compared
non-zombie
real estate developer
end of the period
at the
In
some
in
was about 12.1% of capital, which
more than one year worth of investment during
employment growth of a
zombies.
For example, for the typical non-zombie firm
industries, the difference is quite large.
was
less
if the
The
period.
this
would have been about
zombie percentage had not
average hiring in the industry of 0.62% per year).
to the
Overall, these effects are substantial.
In our
main
specifications
we
find the effect of
zombie
infestation
on non-
zombies depends on the level of sales growth of the non-zombie (negative coefficient
estimates on the three
way
interaction).
robust than the double interactions,
differential impacts suggested
it
by our
While these
is still
triple interaction results are less
interesting to
document
magnitude of the
the
estimates.
Figure 7 uses estimates from Table 2 for equation (12) to infer the differential
effect of varying degrees
of zombie infestation for non-zombies with different levels of
sales growth; formally, this
amounts
= ^ + ns
to studying
.
The
left
panel
dnonzdZ
shows
the
zombie
distortion
the dotted lines in the graph
on investment
show
is
significantly
worse
for fast
growing firms;
the 95 percent (asymptotic) confidence intervals.
only are these marginal effects significant, the overall quantitative impact
instance, for a firm with ten percent sales growth, if the industry
to increase
from
0.1 to 0.2,
investment would
fall
by
1.3
is large.
Not
For
zombie percentage were
percentage points per year; if the
firm instead had 15 percent armual sales growth, the investment drop would be 1.55
27
percentage points per year.
2002)
we view
The
Given the median investment
of 14.7% per year (1993-
rate
these effects as large.
employment.
right panel in the figure presents an analogous calculation for
The marginal
effects again are significant (for all cases
where
sales
growth
is
above two
percent per year). For a firm with sales growth often percent per year, an increase in the
zombie percentage
fi"om 0.1 to 0.2
percentage points per year.
would depress annual employment growth by 0.25
Since employment growth for this sample of firms was
approximately zero, the implied cumulative effect of the high level of zombies during the
late
1990s
is
big.
Given the depressed condition of the economy between 1993 and 2002
which benchmark
clear
have expected
to find
to use in
gauging the size of the
some firms with
sales
growth of 10
firms are quite rare in Japan over this period.
effects.
Normally,
it
is
not
we would
percent per year, but these
to 15
Nonetheless,
appears that there were
it
substantial distortions for the healthiest firms.
6.
Final
Remarks
Let us
First,
the
now
take stock and discuss the implications of our empirical findings.
mechanism we have highlighted compounds
was
all
that
a simple credit crunch.
was going
on, the
benchmark case we analyze.
zombie model, also shows
Thus,
if
analyze in the
economy would be expected
follows that the evidence
It
that a pure credit
characteristic
to
we
crunch explanation
of our mechanism
distortions that lower job creation
" There
we
a pure contraction in credit availability
(2000)) for the recent experience, while highly relevant,
One key
caused by a
Recall that the reduced form profit shock that
traditional credit crunch.
model subsumes
the problems
is
behave
like the
normal
presented to support the
(a la
Kitasaka and
Ogawa
is insufficient.'^
that
and industry productivity.
zombies create on-going
A straightforward extension
we have not tested that could be further used to distinguish these
zombie model explains why the fimis that do enter or expand need not have
high values of Tobin's Q - essentially because the zombie congestion costs lower their profitability. In
contrast, a standard credit crunch model would predict that these firms should be earning rents by virtue of
being able to operate against reduced competition. See Caballero and Hammour (2005) for a discussion of
the channels through which financial factors may depress restructuring during recessions.
are also other imphcations that
two models. For
instance, the
28
of the model would make long-run productivity growth endogenous. In
this case the
present value of the costs due to the suppression of restructuring generated
by continuing
forbearance with the zombies would greatly exceed calculation based only on the direct
costs of subsidies.
Japanese regulators
zombies
may have
failed to recognize the large costs
to continue operating during the episode.
given to Japanese banks in the
that they
late
no longer had an incentive
accrued had Japan returned
of allowing
For example, the capital injections
1990s did not recapitalize the banks sufficiently so
The forgone
to evergreen.
at that point to
benefits that
would have
having a normally functioning economy could
have been large enough to justify a very generous transition policy package
to the
displaced workers that would have been released if the zombies were shuttered.'
Finally our description of the Japanese experience
is
similar to the diagnosis that
has been used to describe the early phases of the transition of
economies
to
becoming market-oriented.
many former
socialist
In these economies the depressing effects
on
the private sector of the continued operation of state-owned enterprises (typically funded
by
state
be the
owned banks)
latest
restructuring,
is
often noted; discussions of the current situation in China
of these examples.
Also, note that the key to our
which may be also caused by
debtors rather than
by banks' behavior.
mechanism
is
would
lack of
legal bankruptcy procedures that protect
For example,
in the U.S. airline industry
it
is
routinely asserted that the industry has been plagued because unprofitable carriers go
bankrupt, yet they
fail to exit
the industry (see
suggest that the mechanism that
'*
The same reasoning
we have
Wessel and Carey (2005)). These cases
sketched
is
not unique to Japan.
'^
applies to the question of whether the lack of liquidations in the U.S. airline industry
raised or lowered the taxpayers' costs of rationalizing the industry.
" See Caballero (2006)
macroeconomics.
for a discussion
of different models and manifestations of sclerosis
in
29
References
Aheame, Alan
G.,
and Naoki Shinada, 2004, "Zombie Firms and Economic Stagnation
in
Japan," paper presented at the conference "Macro/Financial Issues and International
Economic
Relations: Policy Options for Japan and the United States" at University of
Michigan,
Ann
Arbor, MI, October 22-23, 2004.
Bemanke, Ben
Bohn, Henning, and Peter C. Reiss, 1988, "Alternative Non-Nested
S.,
Specification Tests of
Time
Series Investment Models," Jouinal
of Econometrics, 37(2),
March, pp. 293-326.
Caballero, Ricardo
J.,
2006, The Macroeconomics of Specificity and Restructuring, Yrjo
Jahnsson Lecture, forthcoming.
Caballero, Ricardo
J.
Specificity," Journal
Caballero, Ricardo
Development:
and
Mohamad
L.
Hammour,
1998, "The Macroeconomics of
ofPolitical Economy, 106(4), pp. 724-767.
J.
and
Mohamad
Institutions, Crises,
L.
Hammour, 2000,
"Creative Destruction and
and Restructuring," Annual World Bank Conference
on Development Economics 2000,^^.213-2^1
Caballero, Ricardo
Revisited:
J.
and
Mohamad
L.
Hammour, 2005, "The Cost of Recessions
A Reverse Liquidationist View," Review of Economic Studies, 72, 313-341.
Fukao, Mitsuhiro, 2000, Kin'yu Fukyo no Jissho Bunseki (Empirical Analyses of
Financial Recession), Tokyo: Nihon Keizai Shimbun-sha,
(in Japanese).
r
Genda, Yuji, Hiroshi Teruyama, Soichi Ohta, Ryo Kanbayashi, Mamiko Ishihara, Yuji
Senuma, Kazuhiro Sasaki, Kentaro Abe, Takayuki Kusajima, and Taku Morifuji, 2003,
"Koyo
Sdshutsu
to
Shitugyo
ni
kansuru
Jissho
Kenkyu"
(Job
Creation
and
30
Unemployment: Empirical Analysis) Keizai Kenkyu, No. 168, ESRI (Economic and
Social Research Institute), Cabinet Office (Japanese Government), (in Japanese).
Guner, Nezih, Ventura, Gustavo, and
Restrictions
Xu
Yi, 2004,
on Size", working paper, Pennsylvania
Hamao, Yasushi, Mei,
Japanese Stagnation:
"Macroeconomic Implications of
State University.
Jianping, and Yexiao Xu, forthcoming, "Unique
An
Symptoms of
Equity Market Perspective," Journal of Money, Credit and
Banking.
Hoshi, Takeo, 2006, "Economics of the Living Dead", Japanese Economic Review, 57(1),
pp. 30-49.
wa Ryud5sei no Wana
Hoshi, Takeo, 2000, "Naze Nihon
is
the Japanese
Economy Unable
and Hiroshi Yoshikawa
eds.
to
kara Nogarerareainoka?
Get Out of a Liquidity Trap?)" in Mitsuhiro Fukao
Zero Kinri
to
Nihon Keizai (Zero
MIT Press:
K Kashyap, 2001,
and
Interest Rate
Japanese Economy), Tokyo Nihon Keizai Shimbunsha, pp. 233-266,
Hoshi, Takeo and Anil
(Why
the
(in Japanese).
Corporate Financing and Governance
in
Japan,
Cambridge.
Hoshi, Takeo and Anil
K
Kashyap, 2004, "Japan's Financial Crisis and Economic
Stagnation" Journal of Economic Perspectives, 18(1), Winter 2004, pp. 3-26.
Hoshi, Takeo and Anil
K Kashyap,
2005, "Solutions to Japan's Banking Problems:
Might Work and What Definitely Will
Weinstein (Eds.) Reviving Japan
MA: MIT
's
Fail," in Takatoshi Ito,
Economy: Problems and
Hugh
Patrick,
What
and David E.
Prescriptions. Cambridge,
Press, pp. 147- 195.
Hosono, Kaoru, and Masaya Sakuragawa, 2003, "Soft Budget Problems
Credit Market,"
Nagoya
in the
Japanese
City University Discussion Papers in Economics No. 345.
31
Ito,
Takatoshi and Yuri N. Sasaki, 2002,
"Impacts of the Basle Capital Standard on
Japanese Banks' Behavior," Journal of the Japanese and International Economies, 16, pp.
372-397.
Jerram, Richard, 2004, "Banks
Kim,
Se-Jik, 2004,
"Macro
still
not reflating",
ING
Financial Markets, June 2, 2004.
Effects of Corporate Restructuring in Japan", International
Monetary Fund, working paper.
Kitasaka, Shin-Ichi and
Kazuo Ogawa, 2000, "Bank Lending
and Macroeconomic Implications,"
in Crisis
and Change
in Japan: Its
the
in
Determinants
Japanese Financial
System, eds. Hoshi,T. and H.Patrick, Kluwer Academic Pubhshers, 2000, pp. 159- 199.
Miyagawa, Tsutomu,
Yukiko and Nobuyuki Harada, 2004, "The IT revolution and
Ito,
productivity growth in Japan", Journal of the Japanese
and
International Economies,
18(3), pp. 362-389.
Nakakuki, Masayuki, Akira Otani, and Shigenori Shiratsuka, 2004, "Distortions
Markets and Structural Adjustments
in the
in Factor
Economy", Bank of Japan Discussion Paper
No. 2004-E-4.
Nguyen, Hung
T.,
and Elbert A. Walker, 2006, A First Course
Boca Raton, FL: Chapman
in
Fuzzy Logic, 3rd
edition.
& Hall.
Nishimura, Kiyohiko, Takanobu Nakajima, and Kozo Kiyota, 2003, "Ushinawareta 1990-
nen
dai,
Nihon Sangyo
ni
Nani ga Okottanoka?" (What Happened
to the
Japanese
Industry in the Lost 1990s?), RIETI Discussion Paper 03-J-002.
Peek, Joe and Eric
S.
Rosengren, 2005, "Unnatural Selection: Perverse Incentives and the
Misallocation of Credit in Japan", American Economic Review, 95(4), pp.
1
144-1 166.
32
Restuccia,
Diego and Richard Rogerson, 2003, "Policy Distortions and Aggregate
Productivity with Heterogeneous Plants", working paper, Arizona State University.
Sakai, Koji, lichiro Uesugi, and
Tsutomu Watanabe, 2005, "Firm Age and
the Evolution
of Borrowing Costs: Evidence from Japanese Small Firms," RJETI Discussion Paper 05E-026.
Sakuragawa, Masaya, 2002, Kin'yu Kiki no Keizai Bunseki (Economic Analysis of
Financial Crisis), Tokyo: Univeristy of Tokyo Press, (in Japanese).
Sasaki, Yuri, 2004, "Japanese Banks'
Behavior under the Basle Capital Standards,
mimeo, Meiji Gakuin University.
Schaede, Ulrike, 2005, "The 'Middle Risk Gap' and Financial System Reform: Small
Firm Financing
in Japan,"
Monetaiy and Economic
Sekine, Toshitaka, Kobayashi, Keiichiro, and
Studies, 23 (1), pp. 149- 176.
Yumi
Saita,
The Case of Japanese Firms", Monetaiy and Economic
Smith, David
C, 2003, "Loans
to
2003, "Forbearance Lending:
Studies, 21(2), pp. 69-91.
Japanese Borrowers", Journal of the Japanese and
International Economies 17(3), pp. 283-304.
,
Tett, Gillian,
2003, Saving the Sun
:
A Wall
Street
Gamble
to
Rescue Japan from
Its
Trillion-Dollar Meltdown^ Harper-Collins Business.
Tett, Gillian
and David Ibison. 2001. "Tokyo 'May Have
to
Support Banks'." Financial
Times, September 14.
Wessel, David and Sarah Carey, 2005, "For U.S. Airlines, a Shakeout Runs Into Heavy
Turbulence", Wall Street Journal, September
19.
33
White, Halbert, 1980,
"A
Heteroscedasticity-Consistent Covariace Matrix Estimator and
a Direct Test for Heteroscedasticity," Econometi-ica, 48, 817-838.
34
Appendix
1
The
some
variable
R
plays a critical role in our analysis.
additional details
In this appendix
we
provide
on the construction of this variable and the other data used
in the
analysis.
In constructing
might pay
bank
to
loans,
borrow.
R
our goal
produce a plausible lower bound for what firms
For the portion of the
which accounts
40%
for about
interest
to
45%
payments coming from short term
of
total
lending in our sample,
straightforward because almost no loans are
believe that this
is
prime
we
rate (once
to
is
take into account
all
made
at rates
the origination and other fees).
we
below the
Thus,
we view
the use of the short term prime rate as relatively uncontroversial."^"
Ideally,
we would
find an equally conservative assumption for handling long-term
quite likely that interest
payment on a new long-term loan would be above
loans.
It is
prime
rate at the time the loan is originated.
the
Unfortunately, the available data on long-
term bank debt gives just the stock outstanding without information on the exact maturity
we assume
of the loans. So
that
each firm's long term loans have an average maturity of
2.5 years and with one-fifth having
been originated
each year for five years.
in
Five
years corresponds to the average maturity of bank loans in the dataset of Smith (2003).
This assumption implies that the right interest rate
years of the long-term prime rates.
last five
interest
an equally weighted average of the
is
Thus,
we
calculate the
payment on the long-term loans by multiplying
to the
non-bank financing, we know
of interest paying debt was bonds and about
the required
payment ignores the
interest
3%
for
for the
1
990s, this
is
paper.
low
40%
Our measure of
commercial paper.
limited importance of commercial paper financing and the
commercial paper
rates.
that during the 1990s, roughly
was commercial
payments
required
the outstanding long-term loans
of all maturities with the five year average of the long-term prime
Turning
minimum
Given the
interest rates
on
the
not likely to cause any serious problems for our
analysis.
^^
we
computed a required rate that imposed a mark up over the London Interbank
based on the average spreads reported in Smith (2003). This approach produced
similar results regarding the numbers of firms with negative interest rate gaps.
As
alternative
instead
Borrowing (LIBOR)
rate
35
:
For the remaining debt
possible.
Specifically,
we assume
we assume
that
that
it
was financed
bond financing
is
as advantageously as
done with CBs (which by
their
nature have lower yields) and that firms were always able to time the issues so that the
rate is the lowest within the last five years.
Implicitly, this
presumes
perfect foresight and refinance their bonds every time there
This assumption
rates.
corporate debt.
that all
other hand, the approach
firms that
banks.
would increase the
on the bonds they issued
we
a local trough in interest
is
imputation procedure will assume
By assuming
very low required
of our misclassifying credit worthy
risk
On
extreme low bond rates in the public market as zombies.
that enjoy
interests
this
at a zero interest rate.
on bonds, the approach reduces the
interest rates
pay
done
is
label as
have
almost surely going to understate the required payments on
For instance, from 1996 onwards
bond financing
companies
is
that the firms
risk
the
of failing to identify the zombies that
in the past.
Thus,
we can be
zombies must be getting very favorable
confident that any
interest rates
from
their
Put differently, by assuming access to such low bond financing rates our
scheme
classification
will pick out only the
most egregious zombies
that receive
massive
help from their banks.
Besides
this
alternative centered
we
baseline procedure
also
on estimating the maturity
just describe the calculation for long-term
structure of bonds in the
structure
of each firm each year. Here
bank borrowing.
We
t
7VBL;,
(5Z;,
)
and the long-term bank borrowing
that
be the amount of new long-term bank loans
We use the following equation to estimate
NBL, = max {5Z,„ - 5L,_, + 5Z1„_,
still
bank loans
outstanding
to
estimate the maturity
at
the end of
be 10 years.
If
NBL
t.
is
i
at the
comes due within
that the firm
i
1
end of accounting
year (BLl.^
).
Let
takes in during year
t.
NBL.^
,
0}
Let BP{n).^ denote the amount of long-term bank loans to firm
n and
we
same way.
We observe the total long-term bank borrowing for firm
year
One
explored several approaches.
We
assume the
available for
all
i
that
maximum
was given
in year
t-
maturity of long-term
years in the past 10 years,
we
can
estimate BP(n) recursively as follows.
36
5P(0)„.,
= min {NBL.,_, max {BL.,_,
5P(n),.,
=mm|A^54_,,max|54^, -Y,BP{k),_M
5P(9),,_,
=max|5L,.,
,
,
O}}
n-I
(n
=
l,2,--,8)
-J,mk\.A
>
If NBL.i_^_^ is not available for n
n*,
we
stop the iteration at «
= «* and assume
that
the remaining borrowings (if any) are uniformly distributed across different maturities.
Formally, this implies:
BP{0),_,
= min {NBL.,_, max {BL.,_,
5P(n),_,
= min I NBL,_„_, max \bL,^, -
BP{n).j_^
- max
,
O}}
,
The associated regression
For bonds,
minimum
,
we
_Q
.=0
10-«*
results are
shown
in
g 5P(^),_, U
,
(77
< «*)
> n*)
Table A-3
(that
we
discuss in
Appendix
2).
also adopted an extremely conservative approach that assumes the
required interest rate for bonds was zero for the entire sample period. This
approach guarantees that any firms with a negative
interest rate
unusually low interest rates on their bank borrowing.
this classification
shown
(«
scheme are shown
in
gap must be receiving
The regressions associated with
Table A-4 (and are almost identical
to those
in Table 2).
The data
for
prime bank loan rates are taken fi-om the Bank of Japan web
( http://www.boi.or.ip/en/stat/stat
are collected
f.htm ).
from various issues of Kin
The
'yu
subscribers' yields for convertible
for the regression analyses are
taken from the Nikkei Needs Corporate Financial Database.
when we
bonds
Nenpo (Aimual Report on Finance) published
by the Ministry of Finance. The remaining data we use
instance
site
refer to 1993 data they are
The data
are armual, so for
from a firm's balance sheet and income
statement for the accounting year that ended between January and
December of 1993.
37
Appendix 2
We
checked the robustness of the significance of the estimated
minimum
several ahemative measures of the required
Table A-1 repeats the regressions from Table
di)
=
50bp) and
(0,
First, the estimates
(dj, d2)
of
cp
=
difference can be explained
by
the fact that the industry
and probably
related, for the estimates
cp
is
of (11), the
accompanied by smaller standard
When we
(p
and
ti
falls.
statistical significance
Second,
of the estimates
We
Table A-2 shows the
employment growth equations
observations with
Xjt
are again negative
close to zero.
the estimated value of
and
9
falls
is
(p
in the
Xit
between
often higher
d]
investment and
statistically significant in
((3
almost
when we drop
all
the
in equation 11) rises substantially, while
dramatically and becomes insignificant.
dummy
x
sales
The
result for the
growth x industry zombie
not robust to this change in specification, either, suggesting the result
depends on the inclusion of the observations with
We considered several
so similar to those
2,
cases,
For the productivity proxy, however, the estimated
three-way interaction term (non-zombie
is
all
this specification is less robust
The estimates of
results.
gap between the zombies and non-zombies
percentage)
So
significant.'^'
Indeed, the size of the coefficient
the cases.
to the equations
also estimated the regressions dropping the observations with
entirely.
critically
errors, so that the t-statistics are similar.
measure of zombies.
to this alternative
We
are larger
crisp measures.
Their estimated signs remain negative in
but most of the coefficients are no longer
'
this table.
However, part of the
add sales growth (and the associated interaction terms)
the significance of both
Tables
from
{dj,
similar to those in Table 2; in other words, the declines in the size of the
coefficients are
and Ai
and zombie indices.
zombie percentages
zombie measures than when we use the
the frizzy
to
ti's
using the fuzzy zombie indices with
2,
are smaller than those in Table 2.
ti
when we use
of
/'*
We draw three conclusions
(-25bp, 75bp).
and
interest rate
and
(p's
shown
Xjt
close to zero.
other alternatives that are not reported since the results are
in Tables 2, A-1,
and A-2.
In particular, the regressions in
A-1 and A-2 consider only the post 1993 period, when the zombie percentages
use a mixture of crisp and fiizzy assignments to separate the individual firm
and the industry zombie percentages. We performed a few experiments of this type for
employment growth and did not find any systematic ways in which the results were affected.
could
in principle
classifications
38
began
When we
to rise noticeably.
re-estimated the Table 2 regressions to include the
1980s in the sample, the estimates for investment and employment growth remain
unchanged, while those for productivity change. The estimated gap between the zombies
and non-zombies
becomes
rises substantially,
while the estimated value of
falls
(p
sharply and
insignificant.
We
also tried different definitions of non-zombies.
firm as a non-zombie only
if it is
Specifically,
we counted
a
not classified as a zombie in two or three consecutive
years. In both cases the esfimates of
cp
continued to be significantly negative for the
investment and employment regressions, and significantly positive for the productivity
regression, but for one exception
significant
even
at
20%
level.
this alternative definition
This alternative
indices as well.
the
maximum
We
where the estimate of
while
cp,
The estimates of tt were never
negative,
still
statistically significant
way of defining non-zombies can be
applied to the fiizzy zombie
did this by recoding the zombiness of each firm in each year to be
last
two
(or three) years.
Thus, to be classified as a non-zombie for sure, a company has to have z =
The estimates of
The regression
cp
with
of non-zombies, although the point estimates remained negative.
of the z calculated using the equation (1) over the
consecutive years.
was not
results did not differ
much from
for 2 (or 3)
those in Table A-2.
are statistically significant with expected signs in the regressions
without the sales growth.
With
sales growth, the estimates of
n
are not significant,
and
the estimates of cp often lose significance, although the point estimates remain negative.
Table A-3 shows the results using more detailed estimation of the maturity
structure for long-term borrowings
and bonds discussed
estimates on the simple interaction term (non-zombie
percentage) are similar to those in Table 2 in
are,
however, sometimes
10%.
three
Under
way
all
in
Appendix
dummy
1
.
The
coefficient
times industry zombie
the specifications.
The standard
larger, so that the estimates are statistically significant
this alternative
errors
only
at
assumption about the maturity structure, the results for the
interaction term (non-zombie
zombie percentage) disappear. The
dummy
times sales growth times industry
coefficient estimates
on the three way interaction
in
the last two columns are not significantly different from zero.
Finally, Table
minimum
A-4 shows
the regressions under alternative assumption that the
required interest rate on bonds
is
zero.
The
results are essentially the
same
as
39
those in Table 2 except for the
sales
last
column.
In the emplojmient change equation with
growth variables, the estimates of the interaction
significant, although the point estimates fall only
All in
main
text,
dilute the
all,
be
statistically
by small amounts.
the results of these robustness exercises are consistent with those in the
although
it is
apparent that the precisions of some of our estimates suffer as
measures of zombism and increase
and classification
terras cease to
their robustness to different
we
measurement
errors.
40
—
—
Figure
—
— —
Prevalence of Firms Receiving Subsidized Loans in Japan
1:
Raw
Percentage
40
35
30
25
(%) 20
15
10
5
—
I
— — —— — ——— — — — — — — —
1981
I
1
I
I
1983
I
\
1985
1
'
i
1987
1989
1
1
I
I
\
1991
1993
1995
I
I
\
I
I—
\
1997
1999
2001
1997
1999
2001
Asset-weighted Percentage
18
16
14
12
m
10
6
4
1981
1983
1985
1987
1989
1991
Note: Percentages calculated as described in the
1993
text,
1995
with di=d2=0 in equation
1.
Figure
2:
Membership Function
di
for a
Fuzzy Zombie Set
111
zooz
/
/»
V.
8661
>
t
666 V
8661
Z661
i
9661.
\.
S66t
nN
wet
\
A
\
£661.
\V
.2
u
3
2661
1561
066
i
B
5
o
J
9661
N<
s
0002
(
Z661
>
1002
''/
6661
^
«
a
o
f
0002
V
2002
//
lOOZ
«k
en
1
3)
cn
f65t
\
C661
2661
-Sv
u
t
V&SV
^1
0661
''V
/
6861
C/1
8B6t
u
9661
686
886
>v
SB6t
Z861
/
986
9861
S861
>>
N
N
S
N>
(-86
8861
id
to
"O
4
a
V
£861
286
2861.
/
^86L
I86t
5S
CD
a
tB6l
1
ss
es
L
1^
O O O
O O
TfO
ir>
'3-
o
o
o
o
o
o
o
1tn
ID
tt
cn
cm
C
CvJ
en
u
ZOOZ
%
;2
666 i
DJD
)
0)
8661.
)
o
«
/'
i66l
'i
9661
\
V
S
•^
/
<2
3
B
)
9661
^661.
£661
!n
2661
\
Ol
V66V
u
066
5861
o«
H
L.
*
«
8861
ZBSt
1
§
\
1
£661
I
2661
i
}
0661
•T3
1.661.
6861
8861
Z861
.
^
986t
l-l
9661.
I
frsei
\
\\
o
Z661
\N
S661
\
DJD
w
0002
6661
8661
«a
SI
tooz
0002
V
s
o
zooz
\
A
lOOZ
\
9861
9861
S861
'53
^\
t'Bet
A
£861
Z96L
«
1.861
5^
C
1861
o
o
o
o^
CD
O
V
u
a
lO
^
CO
CM
o
o
o
o
o
o
*in
CD
^ o
CO
CM
T-
CO
zooz
r
u
S
200Z
,
'/
VOOZ
PM
7
8661
i
/
aS
«
9661
{
8661.
Z66I.
(
\
2661.
O
U
b
%
1
0661
teei.
6861.
*v
£661
i^
1/1
td
,
••
'S
T3
1661.
0661
\
cs
6861
O
N
8851
Z861
< >
996 >
3
o
o
d
o
2661
CQ
8861
«61.
;•
a,
f66l
X.
£651
'\
"3
9561
9661
A
f661.
E
6661
*.
S66t
1\:
>
6661
Z661
OOOZ
(
«
cn
lOOZ
(
0002
1
f
en
en
u
i-i
(n
.
£861.
Z86I.
1
986
9861
9851
N
N
3
*
f86l
to
f86l.
8851
£861
1
ss
o
o
o
o
o
o
».^
CO
(D
m o
CN
.y >
2861
1.86
ss
o o o
o
o
o
in
CD
^ CO CM »- o
2951
L96I.
o
:%
-%
)
\
f
-\
/
•
'
c
o
1
s
%
3
'
A
c
o
.<
'%
7
\
%.
\
%
'-B
s
)
\
%
1
(A
)
u
>
ij
1
u
o
U
*
Ss
J
Co
'
J
<
u
"31-
?f2
c >
5
c
5?
?
5?
£
o
o o
T(M
\
;
c>
h
gs
5?
sS
sS
o o
o o
-^
CN
%,
ro
u
Cli
\
-%
ja
1
-%
oi
COD
SS
u
u
%
:
,
O)
c
w
3
O
13
S)
TS
J2
#
%
:\
\
r
\
\
<u
Oh
-.1
1
«
\
%
%
A,
%
3
'
C
n
%
.\
0)
o
1-
1
.2
DC
4-*
>
o
CO
O
l-H
1
C >
^
1-
c
V,3
>
c
c>
>
o4
o
t
1
D
v?
-v
o
c3
3
&
c>
^t
t-
^
O
v-
t 3
oJ
c)
&
ss
;
r >
C3
s
%
\
'%
f
/
)
\
*
\
;
f
9)
<fl
%
-%
-\
1
%
u.
<
*
Q^
s
f
'\
1
1
c3
^
o?
C3
C3
•J
5?
O
T-
S
C;
\
%
[••
,
5
<2
GO
-4-»
u
<u
a.
*
3
E
^
-\
\ -\
1 -\
o
N
o
.£5
I/)
o
3
&^
c>
^^
^P
O^
-^
O^
VP
»N.
^
O
O O^
O
O
<CO
CM
00
-4—*
o
CD
S?
f3
ra
en
i3
s
1
>
o
<*i
i^
u
U
O
I
0\
e
o
•a
u
CNJ
c
o
•C
V
>
it)
If)
U
01
u
3
Q
DXI
o
£i
o
.a
i-s
e
"O
c
S
«
{/I
.—
IS
B
O
.a
f5
SI
an
c
R
U
p
(o/„)
O
--
33EJ3AE 00-96
'X6 :9}S-i uopDitjjsap
_
OJ
O
O
o
3§EJ3AB £6
qof
ui
3Seq3
o
d
o
o
CNJ
CO
(%) 3SeJ3AE 00-96
16-16
•'>?^-'
OJ
uopeaja qof
o
o
IT)
3SbJ3AB
ui
aScq^
in
o
^
o
O)
C)J)
2
u
>
C3
(N
O
r<1
ON
O
u
?s
vn
'-^
t
01
;>
ca
<N
On
^^
oo
I
u
3
O
H
os
X
(D
^
:^
K
V
I
t3
^
?^
o
oo
?^
•o
^
^
?N
(N
^
O
0002-0661 H^A^oaSdJl
J^
<N
'^
^
:
o
O
1
I
IS
(/2
1
/
*^,
:
:
4
1
t
/
t
/
*
^.
=
=
.
E
1
o
fO
,
^Z*.
=
1
J2
1
1%
?<
;
Q
.a
1
1
1
o
o
fo-
=
1
1/3
1
Po.
\
J
Pi
t^O
E
r
f
b\
4
f
f
r
%.
—
—
Oto
JF
u
«
O
s:
a
o
j
v^''
* '9^
s
o
c>
cj
f
tn
1
^
1
1
'
1
1
1
p
o
C3
-^
CO
cnj
cz>
CD
ir>
o
q
o
q
q
o
ci
r^
q
o
ipMOjQ juauiXoidiug jEnuuy
V
0X1
cs
•**
a
u
pM
J
:
B
o
ii
^.
>-.
;a
3
1
i
^^
O
1
H
%.
1
H<
v-
%.
==
<£?.
'at
/
f
1
'
//
o
^ %.
--.to
a
/
r
H
im
f
/
»
HV
%.
'--
/
//
,
o
-.-.
/
/
/
1
(?
"
#
/
1
°o
::
/
/
//
1
o
-
/
*
//
1
W %.
'-'
/
^
1
n%
----
a
/
#
t
==
JS
*
M
SI
73
//
1
O
::
/
/
1
f
/
y
f
/
^/
--±12^
/
/
f
CO
o
o
psjjg luauflssAU] jEnuuv
)
Table 1
Search Results For News Articles Regarding Restructured Companies
Total Hits for January 1990 through
Of which,
related to private sector
May 2004
companies
in
1,196
Japan
Clear description of the content of "financial assistance"
,085
20
(excludes duplicate articles on the same case)
New
19
loans
36
(^^IJM:^)
Interest concessions
Purchase of new shares
(
0f tt 5
1
^ §:
t~t
Debt-Equity swaps
Debt forgiveness
26
interest
("Corporation" and "Reconstruction"));
AND
If ft)).
Source: Nikkei Telecom 21.
inz^^^hM^)
payments
11
(^O^^fA^lli^)
AND ("Management Reconstruction Plan" OR
actual phrases were :&Si^S AND (ll'S'S^lf M OR
Notes: Search words: "Financial assistance"
(d^^
44
(ttl^^j
Moratorium on loan principle
Moratorium on
29
Table 2
Impact of Zombie Firms on the Investment, Employment and Productivity of Non-Zombies
Using Baseline Zombie Estimates
Dependent Variable
Sample
Zombie
1993-
1993-
1993-
1993-
1993-
2002
2002
2002
2002
2002
0.0137
3.3842
0.2465
0.0162
(0.0024)
(0.0196)
(0.0084)
(0.0025)
0.0256
0.00109
0.0139
0.0241
0.0009
(0.0056)
(0.001751)
(0.0135)
(0.0058)
(0.0017)
-0.1370
-0.0454
-0.3418
-0.0987
-0.0283
(0.0376)
(0.0116)
(0.0922)
(0.0364)
(0.0108)
*
-0.0885
-0.0232
0.2183
-0.0678
-0.0163
Zombie%
(0.0330)
(0.0102)
(0.0756)
(0.0297)
(0.0088)
Non-Zombie
Industry
%
Sales growth
Non-Zombie
Growth
Industry
ALogE
I/K
0.2390
Non-Zombie
Industry
Log Sales
- % Log E
- 73 Log K
(0.0084)
Constant
Dummy
ALog E
I/K
*
Sales
Zombie%
*
Sales Growth
Non-Zombie * Sales
Growth * Industry
Zombi e%
0.0537
R'
The sample
0.0895
0.3599
0.1152
0.1078
(0.0318)
(0.0097)
0.1436
0.0160
(0.0376)
(0.0116)
1.1002
0.1674
(0.1402)
(0.0427)
-0.5823
-0.0912
(0.1733)
(0.0535)
0.1083
0.1700
consists of between 1,844 and 2,506 publicly traded firms (depending on the year).
Each
estimated after trimming the top and bottom 2.5% of observations (based on the dependent
regression
is
variable).
White (1980) standard
errors are reported in parentheses under each coefficient estimate.
Any firm with actual interest payments
below the hypothetical minimum is considered a zombie and any firm where this is not true is a nonzombie (di=d2=0 in equation (1)). Two digit industry classifications are used throughout. The industry
percentages for zombies are based on the share of total industry assets residing in zombie firms. Sales
growth is the log difference of each firm's sales. I/K is the ratio of investment in depreciable assets to
beginning of period stock of depreciable assets (measured at book value). E is the total number of full
time employees. K is the book value of depreciable assets.
Industry and year
dummies
are also included in each regression.
Table 3
Impact
of
Zombie
Firms on Non-Zombies
Cumulative
A. Cumulative investment losses (1993-2002) of the median non-zombie firm in the high
zombies industries
Wholesale
Industry
Retail
Construction
Real
Services
Estate
Actual Average I/K:
0.1184
0.1871
0.1373
0.0920
0.2215
0.1206
0.0525
0.0833
0.0793
0.0842
0.0963
0.0399
0.0503
0.1117
0.1408
1993-2002
Cumulative Lost I/K
Case
1
Cumulative Lost I/K
Case 2
"Actual Average I/K: 1993-2002" shows the actual average investment rate (I/K) of the median
non-zombie firm in the industry for 1993-2002. "Cumulative Lost I/K Case 1" shows the total
amount of investment (I/K) of the typical non-zombie that was depressed during the period
compared with the hypothetical case where the asset weighted zombie index had stayed at its
average level for 1981-1992. "Cumulative Lost I/K Case 2" shows the total amount of
investment (I/K) of the typical non-zombie that was depressed during the period compared with
the hypothetical case where the asset weighted zombie index of the industry was the same as that
of manufacturing in each year firom 1993 to 2002. The coefficient estimates fi-om the regression
in the column 2 of Table 2 were used for the calculation.
B. Cumulative
employment change (1993-2002) of the median non-zombie firm
in the high
Industry
zombies industries
Wholesale
Retail
Construction
Real Estate
Services
Average Actual
-0.0136
0.0015
-0.0043
0.0062
0.0134
0.0381
0.0190
0.0285
0.0301
0.0381
0.0303
0.0144
0.0172
0.0427
0.0641
Employment growth:
1993-2002
Cumulative
lost
employment
Cumulative
—
Case
1
lost
employment ~ Case 2
"Average Actual Employment Grovi1:h: 1993-2002" shows the actual average aimual rate of
change in the employment at the median non-zombie in the industry for 1993-2002.
"Cumulative lost employment Case 1" shows the total rate of new hiring at the typical nonzombie that was depressed during this period compared with the hypothetical case where the
asset weighted zombie index had stayed at its average level for 1981-1992. "Cumulative lost
employment Case 2" shows the total rate of new hiring at the typical non- zombie that was
depressed during the period compared with the hypothetical case where the asset weighted
zombie index of the industry was the same as that of manufacturing in each year firom 1993 to
2002. The coefficient estimates from the regression in the column 3 of Table 2 were used for the
calculation.
,
,
1
(
—
1
^
H
(
t^ t-^ ro MS r^ on" ON t^
ON o
m OS o
(N ^ o On 00
<N t^ m in o in
O
r^
^^
'^
ON
o o
CN o (N
oo
o O
o O
o O o
o — o
o O o o ro
d d. d d. d d. d d^ d d^ d d. d d. d d^
to
r-
T—
o
r^
t
.—
in
1
'
'
(
t—
r-j
00 CN
^
w
<
<
d
'
b
00
o
J
<
—
o
— (N t^
* ^ r00
^ oo cn" on" o f^
(N t— .— ^ Os oo i^ oo o o in m ^o oo o
o
CN ^ 00 00 (N
oO
o o
o oO
o o
O oo
o o (N
O
o
o o o
d d^ d d. d d d d d d d d d d d d
/
o~
^m ^
f^
(N f^ in o t^ oT VO ^ t^
^ o"
CO
VO in m ON
^
^
CO * CN
m o o (N
<N o o o
O o o —1 o in ^
o t^
d d^ d d. d d^ d d d d^ d d. d
^o CN on"
>o OO ro
On
(N r^ rsi
m
en
o o
(N o p
dd d
„
m
CN
N
3
.3
d
U O
o
^ ^
m
d^ d d^
ON oo
'^ oo
u-1
t~-
m
o
t^
C3N
<~n
^ ^o
m
m
^
m
m
o
o oo o O
d d 3 d d^ d d^ d
^.^^
S
o
r-~
'"^
t~^-i
W)
Os Ss 00 on" •* in'
r-i
oo~
t~~
r^ r^ roo >o rin MD C3N ro oo ON
;3: (N
t^ CN
<N
^
00
o
^^
o d
d
rON
o
d
-O
U
I
a
o
s
^"'^
r- in
oo oo
o
in
rsi
o"
u
s
-a
o
u
^^^
u
<n
r-
r^-
•-s
S
^
1
M^
rt
en
o
,
1
a N
N
a
<u
ja
s
o
.5
i«
W
d
d^
m
o" ro ^ o
^ ON in
'^ o ^
m
r- (N rn
t^
(N o
O Oo d
O
d^ d d^
d d d^
ro
VD
o"
nI-
r?
'vf
'
^
<—'-^
o
in
^
^
r~-
,—1
o"
^ ^
m
t^
m
(N
VO
r-
r^ Co" ON
m
(N (N (N
^ r— o
oO
o O
o O
o o
O
d d d d dd d
'"'
r-H
r-^
r-'"^
'
w^-'
'
.H
5
tS
«
UJ
^-
^
r—
'i^f
>vf
^
>•
d^
o
m
^
rn
d
r-i
1
a o
H
^-1
^
ON
o
d
o
00 >o
r^ oo oo
00 (N (N
00 in ^o
r^
.—1
"^
T—t
^
M
o
J
(On
'sl-
'"^
m ^
o
O O o do o
d d d^ d^ d d^
'O
m
o
-o
r<-i
(>i
"-D
d
o
ON
o
p
d
00
ON
oo
p
d
00
o
H-1
<
os' in o
^
VO o
m
00
CN
^^
in o
o
o
o
o o o o oo oo
o
d d^ d d d d. d d,
^^^
O^l
o"
B
u
m
r1
^
O
in
'""*
r<->
o"
3
.2
>
c
ID
Pi
Q
tfl
^
^
s
—
,
s
OJ
d-
o
3
C/5
.s
-o 'o
r-j
i^^
^
S
(D
s
o
^
m
o
d
o
G" '^ ^ ^ in
^
oo o VO
^
o
'^ o m o o ro in CN
CN o o o (N O o o
d d d d d d d d^
-^
C3
O
u
N
a
o
N
>.
t/2
r2
=i
Z Q
3
CD
g
o
1o N
N
1
en
c 3
o
1—
3
.;5
en
"5
en
.^
u
-3
00
S =^
3 o
VO
t-
m
in
o
d
^
aj
*
-73
ON
in
o o
m o
o
o O o o
d d. d d^ d d^ d d^
in
S:^
o
d
oo o" r^ oo in oo rn
oo ON CN ^o ON vo OO ON
(N r^ in CN
'^
CN
(N
r.
N
1^
ON
OO
'—'
^
^^,^
B
o
*""•
oo
o
in
aj
^
s
00
C/3
_o
13
*
OJ
3
!S
^
S
o
^
*
*
u
J3
^D
U5
_U 3 13
"5 Td 00
*
1
a
o
2
en
_0J
00
S
'S
~"5
°
<+-!
3>
IS
s
o
2
N
S O N o 1 O
o a
o a
N
N
c
o
"2 'Sb
en
o u
3
13
CI
t—
u
DO Q,
XI T3
I
1
(
1
*
I
H
1
'
11
1
—
1
--^ r^ ^ ^ o \o ON t^ r^
5^ m CN CN T— VO m
m * NO
CO
m o
fN
CN
o
'^ (N CN (N in CN
O
o
o o o O »— p O o o — T— NO
—
oo Oo d
o d d d d d d d d d d d d^ d d.
in
r-
(N OO CO
rs)
m
CN
•>
w
'(
i
t^
m
MD
t~^
1
<
1
'
'
d
'
'
'
1
1
tlfl
o
hJ
<
o
in
,
m
m
o
o
,
O^
a
CNl
'
—
o
in
p"
u
3
o
Or
W
in'
o
^
ON
OO
1^
r~~
in
rn
DX)
r<-i
1
Sn
a ^
i
C/D
oj
!-•-
S
00
'"^
'""'
'
1-H
T-H
_::
NO
00
t-^
o
d
r<-)
xj-
d^
^
on"
^ (N r- ^ o ^
ON ON in <N '^ r- ro (N
^
OO VO o o NO rm
^
m
o —1 o -^ O m NO
d d d^ d d. ^ d^ d d^
o"
c "
d
ON
On
c^t"
o
O
d
O
J
^
<N ^
m OO
^
\o OO (N m m
m
^
— o ON
o o o m
o
d
d^ d d^
d. d d^
OO
(N
m
in
m
d
—
.
ON
-n1-
'
ao
G
O
^-t
J
J^ —
o
"^
^
t~^ r00 -o
(N
^^
OO
NO
a>
t«
21
a<
13
S^
00 in
m ^ f?
ON
m
o
o
^
o
(N o O o
o o
d
d d^ d d^ d^ d
r-1
1—1
1
o
.t *-S
"O
'
H-t
o"
o
m
'— o —1
d
d^
d.
r<-i
^
d
o
Z
in
OO CN
CN
a^
r'-)
1
I
'""'
in Os CN
(N
in
N
CN t-^ ^ G" r- S^ 1^
m OO m
•^ o"
o ^ t^
CN M2 o OO
CN T— o CN
o
o
o o o o CN
O
o o oo d
d d d d d d d_ d d^ d
m t~^ ^
on" CN o? 00 o" CN o" ON f-^ On in ro
'— ^ m
en ^ (N o
OO CN
r^ '^ t^ ^ r- in
^ o
t^ (N m r^
o
O
o
o O o o o o o ON m m
o d d d d d d d d d^ d d^ d d^ d
in
r^
S
o
m
o
o
d
(N
1
in
(T)
p"
rn
a»
cd
f?
3
G a
^^
^^ in
^
ON
OO
m
in OO T— CN
^
o
^
o^ ^ OO
O O o (N o O o
d d d^ d d. d d^
>o o^ ro
:^
r- <N
1
,
C/J
(N
in
NO
CO
d
-^ td -&
T3 -G
CN in m t^
OO
m^
<N 1— CN 'O sf OO O)
ro
CN
O
o
o
oo o o oo Oo
d
d d^ d d d, d d^
irT
E ^
1-^
r~^
C3N
r<l
m
W
CN
1
'
r^
o
d
'
O
J
<
C -=!
O o
c«
S
ON OO OO
p"
01)
.2
=
V u
fe
o"
OO
m
OO
o
d
m CN m
OO
O o
o o
o oo p
d d d d dd d p
d
in
—
m
t^
m
O o
1
t^ on" ^^ o"
t^ OS
r^
ro
(N
CN
o
SI
1
u
M
a.
;:3
c3
^
--»
T3
'S
_3
"o
a
o
c
0X1
p^
en
Q
I/]
O
a
o
jj
G
o.
CN
o
m
o
d
^
^u
*
G
<u
^
o
d
'""'
y—t
<u
>
1^
in
m ON On CM m
o
oO m
o o o
p
d d d d^ d d. d p
d
p"
c3
C^'
'""'
OO
on"
OO ON
m
3
'"^
'•^
Co"
OO
m o on"
^
'* o 5^
o
o o o o 1— o
d d d
d d^ d d^ d ^
r.
c
O
O
s
o
N
N
o a
3 No
S
o
>s
C/i
G r^ p
O 3
Z Q G
N
JH
2
W)
t/l
G ^
O
c«
_aj
"13
c/0
^
!£ 43 *
*
*
u -G a
a
o o 3 % o
a 2 N a a 2 N
o o
o O
OJ
J3
3
N
G
O
en
_aj
"c3
00
c/l
0)
G 13
-a 00
G *
k-
G
O
2
CO
^
00
C/l
G
G
S e
t:5
"ti
<"
*—
00
Table A-3
Impact of Zombie Firms on the Investment, Employment and Productivity of Non-Zombies
Using Estimated Maturity Structure for Long-term Borrowings and Bonds
Dependent Variable
I/K
ALogE
3.3919
0.2528
0.0180
(0.0090)
(0.0026)
(0.0210)
(0.0088)
(0.0026)
0.0125
-0.0007
0.0133
0.0144
0.0005
(0.0062)
(0.0021)
(0.0147)
(0.0060)
(0.0019)
-0.0668
-0.0388
-0.3601
-0.0168
-0.0224
(0.0520)
(0.0163)
(0.1190)
(0.0493)
(0.0162)
*
-0.0867
-0.0321
0.2285
-0.0784
-0.0288
Zombie%
(0.0505)
(0.0155)
(0.1122)
(0.0473)
(0.0150)
0.1952
0.1316
(0.0561)
(0.0214)
0.0382
-0.0132
(0.0630)
(0.0248)
Dummy
Zombie
Non-Zombie
%
Sales growth
Non-Zombie
Growth
Industry
I/K
0.0169
Non-Zombie
Industry
Log Sales
- % Log E
- '/3LogK
0.2496
Constant
Industry
ALogE
* Sales
Zombie%
*
Growth
Non-Zombie * Sales
Growth * Industry
Sales
0.6669
-0.0068
(0.4490)
(0.1458)
0.4628
0.2068
(0.4983)
(0.2086)
0.1075
0.1704
Zombi e%
0.0521
R'
The sample
0.3614
0.0897
consists of between 1,844 and 2,506 publicly traded firms (depending on the year).
Each
estimated after trimming the top and bottom 2.5% of observations (based on the dependent
regression
is
variable).
White (1980) standard errors are reported
in
parentheses under each coefficient estimate.
Any firm with actual interest payments
below the hypothetical minimum is considered a zombie and any firm where this is not true is a nonzombie (di=d2=0 in equation (1)). Two digit industry classifications are used throughout. The industry
percentages for zombies are based on the share of total industry assets residing in zombie firms. Sales
growth is the log difference of each firm's sales. I/K is the ratio of investment in depreciable assets to
beginning of period stock of depreciable assets (measured at book value). E is the total number of full
time employees. K is the book value of depreciable assets.
Industry and year
dummies
are also included in each regression.
Table A-4
Impact of Zombie Firms on the Investment, Employment and Productivity of Non-Zombies
Assuming Zero for the Minimum Required Interest Rate on Bonds
Dependent Variable
I/K
ALogE
Log Sales
- % Log E
Log K
ALogE
I/K
'/3
0.2382
0.0131
3.3834
0.2464
0.0158
(0.0083)
(0.0024)
(0.0195)
(0.0082)
(0.0023)
0.0237
0.0007
0.0129
0.0223
0.0006
(0.0056)
(0.0017)
(0.0133)
(0.0055)
(0.0017)
-0.1879
-0.0533
-0.3915
-0.1452
-0.0338
Constant
Non-Zombie
Dummy
Industry
Zombie
(0.0394)
(0.0123)
(0.0941)
(0.0384)
(0.0120)
*
-0.0793
-0.0213
0.2283
-0.0575
-0.0145
Zombie%
(0.0336)
(0.0104)
(0.0764)
(0.0320)
(0.0101)
Non-Zombie
Industry
%
Sales growth
Non-Zombie
Growth
Industry
* Sales
Zombie%
Sales Growth
Non-Zombie *
Growth *
*
Sales
0.1240
0.1104
(0.0495)
(0.0214)
0.1394
0.0144
(0.0593)
(0.0239)
1.0730
0.1561
(0.3132)
(0.1191)
-0.5706
-0.0835
(0.1154)
(0.1489)
0.1084
0.1699
Industry Zombie%
0.0543
R'
The sample
0.0896
0.3599
consists of between 1,844 and 2,506 publicly traded firms (depending
on the
year).
Each
estimated after trimming the top and bottom 2.5% of observations (based on the dependent
regression
is
variable).
White (1980) standard en^ors are reported
dummies
in parentheses
under each coefficient estimate.
Any
firm with actual interest payments
below the hypothetical minimum is considered a zombie and any fimi where this is not tme is a nonzombie (di=d2=0 in equation (1)). Two digit industry classifications are used throughout. The industry
percentages for zombies are based on the share of total industry assets residing in zombie firms. Sales
growth is the log difference of each firm's sales. I/K is the ratio of investment in depreciable assets to
beginning of period stock of depreciable assets (measured at book value). E is the total number of full
time employees. K is the book value of depreciable assets.
Industry and year
are also included in each regression.
^
Date Due
Lib-26-67
„MIT LIBRARIES
3 9080 02617 9686