You are on page 1of 10

Measuring the Effect of Restructuring on Corporate Performance: The Case of Management

Buyouts
Author(s): Scott B. Smart and Joel Waldfogel
Source: The Review of Economics and Statistics, Vol. 76, No. 3 (Aug., 1994), pp. 503-511
Published by: The MIT Press
Stable URL: http://www.jstor.org/stable/2109975
Accessed: 31/08/2008 21:55

Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at
http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless
you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you
may use content in the JSTOR archive only for your personal, non-commercial use.

Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at
http://www.jstor.org/action/showPublisher?publisherCode=mitpress.

Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed
page of such transmission.

JSTOR is a not-for-profit organization founded in 1995 to build trusted digital archives for scholarship. We work with the
scholarly community to preserve their work and the materials they rely upon, and to build a common research platform that
promotes the discovery and use of these resources. For more information about JSTOR, please contact support@jstor.org.

http://www.jstor.org
MEASURING THE EFFECT OF RESTRUCTURING
ON CORPORATE PERFORMANCE: THE CASE
OF MANAGEMENT BUYOUTS
ScottB. SmartandJoelWaldfogel*
Abstract-Recent researchhas attemptedto documentthat Mueller (1985) reaches similarconclusionsin his
the financialgains associatedwith takeovers,LBOs and other
types of restructuringare attributableto subsequentimprove- studyof marketshare data.
ments in operatingperformance.In this paperwe develop a Thoughtheir conclusionsdiverge,these studies
moregeneralframeworkfor measuringthe effectof corporate share the commongoal of measuringthe effect of
restructuringon performanceand applythe frameworkto a
sample of firms taken private by their management.We corporate restructuringon the performance of
demonstratethat the estimationapproachesemployedin the firms.Answeringthis question is complicatedbe-
literatureembodyrestrictionson the generalframeworkwhich cause the relevant benchmark against which
the data can reject. However, our best estimates provide
evidencethat MBOsimprovecorporateperformance,and the post-reorganizationperformanceshould be mea-
magnitudesof these improvementsare similar to existing sured is not preorganizationperformance.The
estimates. preferredbenchmarkincorporateschangesin ex-
pected performance for the restructuringfirms
I. Introduction prior to their decision to reorganize as well as
A GGROWING body of researchindicatesthat unanticipatedshocksto performanceunrelatedto
the restructuringdecision. It is difficultto find a
corporaterestructuringgenerates value for
stockholders, and recent empirical evidence control measure that convincinglyanswers the
points to improvementsin operatingperformance question, "what would have happenedto perfor-
as a primarysource of these gains. Kaplan(1989) mance at the restructuringfirmsin the absenceof
and Lichtenberg and Siegel (1989) study firms restructuring?"
taken private in management buyouts (MBOs) In most cases, researchersattempting to ad-
and find that both financial(sales, income, etc.) dress this issue have measuredthe expost perfor-
and real (factor productivity)performancemea- mance of restructuringfirms relative to the re-
sures improve after the buyout. Healy, Palepu sults achievedby companiesoperatingin the same
and Ruback (1990) also discover signs of asset industry.1The effect of restructuringis thus mea-
productivityimprovementsin a sampleof merged sured as a "differencein differences,"or as the
firms.The authorsof these studies conclude that change in performanceat the restructuringfirm,
minusthe simultaneouschangein performanceat
MBOs, mergers and other types of corporate
restructuringgenerate operating efficiencies by the control firm(s).The chief shortcomingof this
altering managerial incentives, reducing agency approachlies in its assumptionthat, if not for the
costs and improvingfactor productivitythrough restructuring,the reorganizingfirm would have
other means. In contrast, a number of other experiencedthe same change in performanceas
studies report results at odds with the view that its competitors.If individualfirmshave their own
corporate restructuringgenerates long-run im- dynamicperformancepatterns,it may be unrea-
provements in value. For example, Ravenscraft sonable to assume that except when organiza-
and Scherer (1987a and 1987b) detect no evi- tional structurechanges, the performancemea-
dence of improvementsin post-takeoveroperat- sures of firmsin the same industrymove together.
ing performancein a sample of merged firms. An importantreason to doubt that the change
in performance at non-restructuringfirms pro-
vides an appropriatecontrol measure is that re-
Received for publicationSeptember19, 1992. Revision ac-
cepted for publicationOctober18, 1993.
* IndianaUniversity;and Yale Universityand NationalBu-
1
reau of EconomicResearch,respectively. In additionto matchingfirmsaccordingto industry,some
We would like to thank Steve Kaplan for providingpost- studies have attempted to match firms according to size,
buyout data on MBO firms. We are grateful to Michael leverage and other financial characteristics.In contrast,
Jensen, John Shoven,ChrisJones and two anonymousrefer- Jarrell (1991) employs an expectations-basedmethodology
ees for helpfulcomments.Any errorsare our responsibility. similarto the approachwe propose.

Copyright ? 1994 [ 503 ]


504 THE REVIEW OF ECONOMICS AND STATISTICS

structuringfirms are a non-random sample of ence in differences measure, as well as others.


firms.The reasonsfor restructuringmay depend, Each special case entails a set of testable restric-
at least in part, on past and (expected) future tions on the general framework.Using data on
performance.For example, suppose that firms' performance at firms undertakingmanagement
performancepatterns are positively autocorre- buyouts(MBOs),we presenttests that distinguish
lated and that restructuringsoccur when perfor- betweenvariousestimatorsfor the effect of MBOs
mance is low. Suppose,further,that at any point on performance,as well as our best estimates of
in time, the performancelevels of differentfirms the MBO effect.
in an industryare at different levels relative to The paper is organized as follows. Section II
their own historicalaverages.Then, independent describeshow the circumstancessurroundingcor-
of the restructuring,we expect performanceim- porate restructuringaffectthe measurementof its
provementat the restructuringfirm. Equallyim- effect on firm performance.Section III develops
portant,we do not necessarilyexpect the same a general empirical framework for measuring
improvement at the "control" firms. Conse- the effect of reorganizationon firm performance
quently,the differencein differencesmay provide and discusses,thespecial cases of this framework
a misleadingmeasure of the effect of restructur- that have been employedin existingstudies. Sec-
ing on performance. tion IV describes data and empiricalestimation
An accuratemeasure of the effect of restruc- of the componentsof the framework.Section V
turingon firm performanceis importantbecause presentsresults.A conclusionfollows.
it sheds light on organizationalefficiency.In addi-
tion, the influenceof corporatecontrol events on
II. The ReorganizationDecision
privatesector productivityis directlylinked with
a numberof publicpolicyconcerns.For example, The numberand size of corporationsundertak-
corporate restructuringaffects the tax revenue ing significantrestructuringactivitiesin the form
collected from corporations. When companies of mergers, share repurchases, and leveraged
privatize,they typicallyreplace equity with debt. buyouts grew during the last decade, motivating
While the cash flow previously paid to equity economiststo studythe effects of corporatereor-
holders faced corporateincome taxation,the in- ganization on various aspects of firm perfor-
terest paymentspaid to post-buyoutdebtholders mance. Jensen (1986) and others advance the
escape corporate-leveltaxation. Jensen, Kaplan, theory that some types of restructuringcreate
and Stiglin(1989)estimatethat the U.S. Treasury value by alteringthe incentivesof managersand
has actually gained from management buyouts owners in a way that enhances efficiencyand by
(MBOs)because the tax revenuesgeneratedfrom reducing agency costs. Shleifer and Summers
improvedcorporate cash flows have been large (1988) suggest that corporate takeovers merely
enough to offset corporate tax losses associa- transfervalue from employees and other stake-
ted with the increased usage of debt.2 Bulow, holders of firms to shareholdersby breakingim-
Summers,and Summers(1990) show that these plicit contracts.Similarly,Lowenstein(1985) dis-
gains to the Treasurydisappearwithout the im- cusses the possibility that managers who take
provementsin operatingperformance.Thus, it is their companiesprivatein leveragedbuyoutshave
necessaryto determinewhether increasedcorpo- private information about their firm's future
rate cash flows are attributableto restructuringor prospects. This informationallows managers to
to other factors. buy the companyfrom public shareholdersat a
This paperpresentsa frameworkfor estimating pricebelow thatwhichan informedinvestorwould
the effect of restructuringon performance.This pay. Thoughthis hypothesisimplies that financial
frameworkembraces as special cases the differ- performance improves after the buyout, the
source of the value increase is not related to the
2
There are other tax effects of these transactionsincluding buyoutitself.
extra capital gains taxes paid by selling shareholders,in-
creasedrevenuesfrom more efficientuse of capital,taxes on Whateverthe motivationfor corporatereorga-
debtor'sincomesand asset sales and so on. Some of these tax nization, theory providesreasons to suspect that
effects(e.g., the added capitalgains taxes) are not contingent these firmsare systematicallydifferentfromother
upon future operatingperformance.However,the net effect
on the Treasury'srevenuesis negativewithoutthe operating companies, and empiricalevidence supportsthe
performanceimprovements. theory. Neither the types of firmsthat choose to
RESTRUCTURING AND CORPORATE PERFORMANCE 505

go private,nor the timing of the decision is ran- approaches examined is a special case of our
dom. Palepu(1986),Smart(1992),and othersfind theoretically preferable general approach, and
that firms involved in corporatecontrol transac- each embodies testable restrictionson the gen-
tions have characteristicsthat distinguishthem eral framework.
from other firms.3 If these characteristicsare The simplest possible measure of the effect of
correlated with future returns, then one must restructuringon firm performanceis simply the
account for these factors in order to obtain an raw change in performanceafter the buyout oc-
unconditional estimate of the effect of the re- curs:
structuringdecision on returns.
Firms that are experiencingunusuallylow cur- (1)
XtRk -XtR'
rent performancemay be more likely than other
firms to restructure. If this is the case, then where X is some measureof firmperformance,R
simply calculatingpre- and post-buyoutperfor- denotes a restructuringfirm,and subscriptst and
mance changes may overstate the effect of the t + k refer to pre- and post-restructuringtime
reorganization.In addition,part of the change in periods. We will refer to the raw changes as
performance at the restructuringfirms is also estimator(1). The problemwith this estimatoris
experiencedat other firms.For example,part of that it attributes the entire change in perfor-
the improvementat a firm that restructuresdur- mance to the reorganizationeven though some
ing an industryslump is experiencedby all firms portion of the change at the firm would have
in the industry and is not attributableto the occurred in the absence of any reorganization.
reorganization.4 For example,observablemeasuresof firmperfor-
Thus, there are two adjustmentsthat must be mance may be autocorrelated.If a firm'sperfor-
made to raw changes in performanceto isolate mance patternis cyclicaland restructuringoccurs
the effect of restructuring.Both the firm's ex- when the performancemeasure is below its his-
pected change in performanceand contempora- toricalmean, then some improvementis expected
neous shocks felt at the firm and elsewheremust and is not properlyattributableto the reorganiza-
be removed. Adjusting the change in perfor- tion.5 We can test whether performancechanges
mance at a restructuringfirmwith the change in are expected at restructuringfirms.
performance at non-restructuringfirm will not
yield a correct estimate unless both the control HYPOTHESIS A: E(XtR kIt) = XtR
and restructuringfirm have the same expected Prior to the restructuring, performance is ex-
performanceimprovement.The followingsection pected to remain at current levels.
lays out a frameworkfor measuringthe effect of Rejectionof the hypothesisthat no changesare
restructuringon performancethat is valid even if expectedindicatesthat the rawchangesestimator
control and restructuringfirms have divergent is an inadequatemeasureof the effect of restruc-
performanceexpectations. turing because significantperformancechanges
are expected absent restructuring.
III. An Empirical Framework A natural way to deal with the problem of
anticipatedperformancechanges is to adjustthe
This section introducesa series of approaches actual change in performancethat occurred for
to estimatingthe effect of restructuringon per- (an estimate of) the expected change that would
formance, including measures which have been have occurred without the buyout. Suppose we
advancedin the literature.Each of the estimation have a measure of expected performanceat the
restructuringfirm in period t + k, E(Xt/kIt).
3 See also Shivdasani(1993)and Morck,Shleiferand Vishny
(Note that this expectation is taken at time t,
(1989).
4Consider the analogyof an auto workerwho decides to
return to school to get an M.B.A. If the worker chooses 5Survivorship bias can generate the same type of effect.
graduateschool in part because of temporarilylow earnings, Considerthe populationof firmsexperiencingbelow average
then measuringthe effect of a graduatedegree on incomeby performanceat time t. Some of these firmswill survivein one
takingthe differencebetween pre- and post-M.B.A.earnings form or another,but others will cease to exist. Because any
will yield an upwardbiased estimate.Besides this individual- sample of firms with continuouslyobservableperformance
specific"low earningsshock,"wages of all auto workersmay data over some time interval will exclude those that go
have experiencedeither a positive or negative shock during bankrupt,we would expect to observeperformanceimprove-
the trainingperiod. ments amongthose firmsin the sampleinitiallydoing poorly.
506 THE REVIEW OF ECONOMICS AND STATISTICS

priorto the restructuringdecision, and it presup- The difference in surprises measure has the
poses that no organizationalchange will take following intuitive interpretation.Suppose that
place). Then the effect of reorganizationis only we have accountedfor every statisticallycontrol-
the part of the change in performancethat was lable aspect of the change in performance at
not expected prior to the buyout, or the restruc- restructuringand non-restructuringfirms pre-
turing surprise: dictable prior to the reorganization,except the
- E(XtR+klt). restructuringitself. What we have left is a set
Xtk (2)
of change-in-performancesurprises for the re-
Interpretingthis as a measure of the effect of structuring and non-restructuringfirms. The
restructuringrelaxes the assumptionof the first difference in surprises estimator is the average
approach that the expected change in perfor- difference in performancesurprisesbetween re-
mance in the absence of restructuringis zero. structuringand nonrestructuringfirms and may
While the restructuringsurpriseoffers an im- properly be interpreted as an estimate of the
provement over the raw change (1), it suffers effect of restructuringon firmperformance.
from the potential shortcomingthat it attributes
Separate from the progression of estimators
to the buyout any unforecastable change ex- above is a measureof the restructuringeffect that
perienced at the restructuringfirm even though has been used in the literature (Kaplan, 1989;
nonrestructuringfirms may experience the same Lichtenbergand Siegel, 1989, and others), the
performanceshock. That is, the surprise in ex- change in performanceat restructuringfirmsmi-
pression(2) will typicallyincludeboth firm-specific nus the change in performance for a control
'and more general shocks.6 If the restructuring sample of firms,
firm's industry does surprisinglywell following
the firm'sreorganization,then it would be incor- (Xt+k - XtR) - (Xt+k - Xt), (4)
rect to attributeall of expression(2) to restructur-
ing. We can test whether industryshocks under- or the difference in differences. A shortcoming of
mine the interpretationof estimator(2). this estimation approach is that it neglects the
possibility, discussed above, that some parts of
HYPOTHESISB: E(Xt+ k| t) = Xt+k (for controls) the changes in performanceat the restructuring
Thereare no performance shocks at controlfirms. and non-restructuring firms may be forecastable.
Rejection of hypothesis B indicates that esti- Our general estimator(3) reduces to expression
mator (2), while better than (1), still gives a (4) when the expected changes in performance
misleadingestimateof the effect of restructuring. are equal at restructuringand controlfirms.8We
Only the componentof the restructuringsurprise can test whetherthis conditionobtains,and hence
not experiencedby similarnonrestructuringfirms whether the differencein differencesmeasure is
is properlyattributedto the reorganization. appropriate,by examiningthe followinghypothe-
The possibilityof a performanceshock experi- sis:
enced by both restructuringand control firms
suggeststhat the effect of restructuringshouldbe HYPOTHESIS C: [E(Xt+ kIt) -X/R] =
measured as the performance surprise experi- [E(Xt+klt) - Xt]
at
enced restructuringfirms,less the performance Expected improvements at restructuringand con-
at
surprise comparablenon-restructuringfirms.7 trol firms are identical.
Our most general measure of the effect of a An example will help illustrate the difference
restructuringis then the restructuringsurprise among the estimators.Suppose that performance
minus the "controlsurprise,"or the differencein at XYZ Corp. is expected (in the absence of a
surprises: restructuring)to improveby 10% over the next
- - [Xt+k - E(Xt+klt)].
two years.9Suppose a buyout occurs and perfor-
[Xt+k E(Xt+klt)] mance improves15%. The restructuringsurprise
(3)
8An intuitivespecial case of this arises if no performance
6
For example,a petroleumfirm that went privatein 1990 improvementsare expectedat either type of firm.
would have experienceda performanceimprovementin 1991 9 In the empiricalsectionwe use the changein the ratio of
due to the Gulf Crisis,as did firmsremainingpublic. operatingincometo sales as our performancemeasure.Thus,
7Of course,findinga non-MBOcontrolfirmis not a trivial if the current income to sales ratio for XYZ is 0.10, an
matter,but we leave that concernfor the empiricalsectionof expectedimprovementof 10%impliesan expectedincometo
the paper. sales ratioof 0.11 in two years.
RESTRUCTURING AND CORPORATE PERFORMANCE 507
TABLE l.-FOUR ESTIMATES OF THE RESTRUCTURING EFFEcr comparisonsbetween our results and the results
Raw Change = 15 of previousstudies.
Restructuring Surprise = 15 - 10 = 5 The data on actual performanceof MBO firms
Difference in Differences = 15 - (1) = 14
Difference in Surprises = (15 - 10) - (1 - 5) = 9 (X/?) come from two sources. Pre-buyoutdata
are primarily from COMPUSTAT's Research File,
althoughdata unavailableon COMPUSTATwere
obtained from 10-k's directly. Post-buyoutdata
is 5%. Suppose that the performanceat Refer- for 48 MBOs are obtained from the data set
ence Corp. is a relevantcontrol. Performanceat developed by Kaplan (1989).11 These firms were
Reference Corp.was expected to improveby 5% subject to public disclosure requirementsafter
but actually increased by 1%. The control sur- their buyouts either because they continued to
prise is thus - 4%. What is the restructuring have public securities (debt or preferred stock),
effect? Table 1 shows restructuringeffect esti- or because they sought additional public fund-
mates arising from each of the four estimators. ing.12Up to three years of post-buyoutdata, not
The raw change is 15%, but the difference in includingthe fiscalyear of the reorganization,are
surprisesis 9%. In principle,all estimatorsexcept availablefor each firm. In what follows, the first,
the difference in surprisesgive misleading esti- second, and third years after the buyout are re-
mates of the true effect of restructuring.The ferred to as t + 1, t + 2, and t + 3.
appropriatenessof each estimator in practice is The data on actual performance at control
an empiricalissue whichwe examinebelow. firms(Xe) come from two basic sources,COMPU-
STAT and Value Line. We use three different
control samples. From COMPUSTAT,we con-
IV. Data structan annualtime series of the aggregateratio
Estimation of the effect of restructuringon of operating income to sales at the 3-digit SIC
performanceusing the frameworkabove requires level. These 3-digitindustriesare chosen to match
us to create empirical analogues to each of the the 3-digit industriesof the MBO firms. In any
components of expression (3). These are com- year, we include all firmsthat have data for both
prised of two types of information,performance variables(COMPUSTATdata items 12 and 13).
and expectedperformance,for two differenttypes From ValueLine, we obtain two differentcontrol
of firms, restructuringfirms and control firms. measures of performance.First, for each MBO
Expected performance measures are obtained firm, we choose up to five individualValueLine
from both time series methods and analysts'fore- firmsof similarsize in the same SIC industry.We
casts.10 aggregatethese firms'operatingincome and sales
The performance measure (X) used in this figuresand use the aggregateratio as our second
study is the ratio of operating income to sales. control measure of performance.Alternatively,
Performancechanges are reportedin percentage we associate each MBO firm with a ValueLine
terms. We use a scaled measureto allow compa- industrygrouping,and we use ValueLine's aggre-
rabilityacross firms and to control (at least par- gate industry-grouping data as a control measure
tially) for post-buyout divestitures.In addition, of performancefor that firm.13
the income to sales ratio may be loosely inter- For both control and restructuringfirms, we
preted as a measure of the efficiencywith which must also derivemeasuresof the expectedchange
firmsutilize a given amountof sales. We scale by in performanceat the time of the restructuring
sales rather than assets because the accounting
value of assets typicallychanges at the time of a See Kaplan(1989) for a detailed descriptionof his data.
12A few firmsdisclosedtheir financialdata when they were
buyout,makingpre- vs. post-buyoutcomparisons sold to existingpubliccompanies.
13
difficult.Finally,our measureof performancehas ValueLine industrycategoriescorrespondonly loosely to
SIC
been used elsewhere in the literature (Kaplan, different codes and group together firms that operate in vastly
marketson dramaticallydifferentscales. It is not
1989;Jarrell,1991),so we are able to make rough uncommonfor ValueLine to group two firms whose SIC
codes are dissimilareven at the two-digitlevel. We preferon
10
a priori grounds using firm-specificdata to construct our
We employboth types of forecastsbecause of conflicting control measuresbecausewe can matchcontrol and restruc-
evidenceon the qualityof time series and analysts'forecasts. turingfirmsmore closely on the basis of firmsize and indus-
See, for example,Brownet al. (1987) and O'Brien(1986). try.
508 THE REVIEW OF ECONOMICS AND STATISTICS

firm's buyout. We derive expected changes in that allow the data to distinguishbetween raw
performancefor the MBO firms (E(XtRklt) - changes and restructuringsurprises(1 vs. 2), be-
X/R) in two ways. First, we estimate a dynamic tween restructuringsurprisesand differences in
performanceregressionon the MBO firms' an- surprises (2 vs. 3), and between differences in
nualperformancehistoryup to the last pre-buyout differencesand differencesin surprises(4 vs. 3).
year using the ratio of operatingincome to sales Table 2 presents Wilcoxonstatistics(in paren-
as the dependentvariable,and we use this regres- theses) for these three hypotheses,estimatedover
sion to generateforecastsfor post-buyoutperfor- the three different control groups' data. Figures
mance. This regression includes one lagged de- in columnA representthe median expected per-
pendentvariableand allowseach firm a different formance improvements at restructuringfirms
intercept. Our second measure of expected measured as a percentagechange in the operat-
change in performanceat the restructuringfirms ing income to sales ratio.14For example, using
is the ratio of operatingincome to sales predicted COMPUSTATdata we estimate that the median
by Value Line. Value Line publishes a multi-year expected change in the ratio of operatingincome
forecastcoveringthe second throughfourthyears to sales is 4.0%by the firstyear after the buyout,
following the buyout (years t + 2 to t + 4). We 5.3% by the second year, and 7.0% in the third.
obtainthe expectedchangein performanceas the ValueLine forecasts show even larger expected
last forecast of performance before the an- performanceimprovementsfor these firms,7.5%
nouncementof the buyout.Thus, these expected by year two and 17.0% by year three, and they
changes are the expectations of performance are significantlydifferentfromzero.15Becausewe
changes in the absence of a buyout. consistentlyreject hypothesisA, it is inappropri-
We derive expected changes in performance ate to use raw changes to measure the MBO
for control measures (E(Xf+klt) - XI) in three effect as they will give an upwardlybiased esti-
ways: (1) using autoregressionson SIC 3-digit mate of the true effect.
aggregate COMPUSTAThistory (in a fashion ColumnB containsmedianestimatesof control
analogousto the regressionson MBO history),(2) firm surpriseswhich should theoreticallybe sub-
by aggregating Value Line's performance fore- tracted from restructuringsurprisesto yield the
casts for the 5 individualfirmsthat we matchwith true MBO effect. Note that accordingto all of
each MBO firm, and (3) using ValueLine's fore- our firm-specificdata,controlfirmsperformworse
casts of industry aggregate performance.In all than expected, and significantlyso by year three.
three cases, the control forecastsare made given In other words, both COMPUSTATand Value
informationavailableprior to the buyout at their Line forecasts of the performanceof individual
matchedMBO firms. control firms prove to be overly optimistic, al-
thoughthe ValueLine data on aggregateindustry
V. EmpiricalResults performance fail to reject the hypothesis that
control surprises are zero. To the extent that
In this sectionwe estimatethe effect of restruc-
control surprisesare significantlybelow zero, the
turing on performance using data on MBOs.
restructuringsurpriseestimatesunderestimatethe
We test between the measures discussed in sec-
MBO effect because they neglect simultaneous
tion III, and we present our best estimatesof the
negativeperformancesurprisesat controlfirms.
MBO effect.
Column C presents median estimates of the
difference between restructuringfirms' expected
A. Choosing an Estimator
performanceimprovementsand that of control
In principle,the most general measure of the firms.A positive estimate indicatesthat expected
effect of restructuringis the best measure,since it improvementsat restructuringfirmsexceed those
relaxes as many assumptionsas possible. Yet, in at control firms, and therefore that the differ-
choosing empirically among estimators for the ences in differencesestimateswill exceed the true
effect of restructuring,what matters is whether
the restrictionsembodiedby a particularestima- 14 These changesare measuredrelativeto the year immedi-
tor indeed hold. In what follows, we test the ately priorto the buyout.
15
The bottom panel of columnA is blank because to test
restrictionsimplied by choices of various estima- this hypothesiswe need to measure expected performance
tors over others. In particular,we test hypotheses improvementsonly at restructuringfirms,not controlfirms.
RESTRUCTURING AND CORPORATE PERFORMANCE 509

TABLE 2.-MEASURES OF THE DIFFERENCES BETWEEN ESTIMATORS


AND HYPOTHESIS TESTSa

Control N Ab Bc Cd

1. COMPUSTAT data used to forecast expected performance


t+ 1 35 4.0 -1.5 6.1
(2.19)' (-1.03) (1.92)9
t+ 2 31 5.3 -4.7 6.0
(2.19)' (-1.29) (1.96)'
t+ 3 14 7.0 - 28.7 -6.7
5.8 - 27.8 -11.7
(1.6) (-2.54)' ( -0.41)
2. Value Line data used to forecast expected performance
1. Individual firms
t+ 2 25 7.5 -7.9 1.0
(3.43)f ( - 2.52)' (0.96)
t+ 3 11 17.0 -18.3 4.6
(2.22)f ( 1.96)' (1.69)9
b. Industry aggregate datae
t+ 2 24 2.3 1.6
1.9 1.5
(0.77) (0.44)
t+ 3 10 2.3 4.0
1.9 1.6
(0.87) (0.76)
a
Wilcoxon statistics are in parentheses. Critical values for two-tailed tests are 1.96 and 1.65 for 95% and 90%
significance levels, respectively.
Ho: E(XR+klt) - = 0, or that the expected performance improvement at restructuring firms is zero.
Rejection indicates that estimator 2 provides an improvement over estimator 1.
Ho: XI+k - E(XI+klt) = 0, or that control surprises are zero. Rejection indicates that estimator 3 is more
appropriate than estimator 2.
d
Ho: [E(X/R+klt) - XR- [E(X,+klt) - XI] = 0, or that the control and restructuring performance measures
have the same expected improvements prior to the restructuring. Rejection indicates that estimator 3 provides an
improvement over estimator 4.
e Note that since there are an even number of firms in this panel and in year t + 3 of panel 1, there is no unique
median, so we report both medians. Since hypothesis B does not require industry adjustment, we do not repeat the
results from panel 2.
f Significant at 95%.
g Significant at 90%.

MBO effect. Our evidence on this hypothesisis effect and that the differencein surprisesestima-
mixed. COMPUSTATdata suggestthat there are tor offers an improvementover the differencein
significantdifferences between expected perfor- differences.
mance at restructuringand control firms in the
shortrun (the firstand second post-buyoutyears), B. Choosing an Estimate
but that these differencesdisappearby year three.
The Value Line firm-specificdata indicate that The tests reportedabove indicate that conven-
MBO firms have greater expected performance tional measuresof the effect of restructuringon
improvements,but they are significantat conven- performance embody restrictions that the data
tional levels only in the third year. The industry sometimes reject. The pattern of test results
aggregatedata also indicate higherexpected per- guides our choice of estimates, presented below.
formancefor MBO firms,but the differencesare Table 3 reportsmedian MBO effect estimates as
not significant.The erraticestimatesfor the latter well as Wilcoxon tests of the hypotheses that
years probablyreflect our small samples (fewer these effects are zero. Our finding of positive
than 15 firms by the final year). In summary,all expected improvementsat restructuringfirms-
but one of our estimates indicates that MBO and concomitantrejectionof hypothesisA-sug-
firmshave higherexpectedperformanceimprove- gests that the raw changes (1) gives larger esti-
ments than controlfirms.Because these estimates mates than the restructuringsurprises (2). Re-
are not consistentlysignificant,we conclude that sults in table 3 confirmthis. With the exceptionof
there is only weak evidence that the differencein the COMPUSTATestimate for the first post-
differences estimator overstates the true MBO buyoutyear, all of the restructuringsurpriseesti-
510 THE REVIEW OF ECONOMICS AND STATISTICS

TABLE 3.-MEDIAN ESTIMATES OF THE MBO EFFECr


USING ALTERNATIVE ESTIMATORSa

Control N lb 2C 3d 4e

1. COMPUSTAT data used to forecast expected performance


t+ 1 35 4.1 5.7 6.8 6.7
(1.97)9 (0.95) (1.39) (1.97)9
t+ 2 31 11.9 7.4 11.5 11.8
(2.62)9 (1.87)h (2.00)9 (2.36)9
t+ 3 14 22.8 18.1 33.6 26.9
19.4 10.2 28.7 22.7
(2.42)9 (1.66)h (2.86)9 (2.17)9

2. Value Line data used to forecast expected performance


a. Individual firms
t+ 2 25 18.0 7.7 21.4 15.7
(2.65)9 (0.98) (1.98)g (2.19)g
t+ 3 11 23.9 1.4 33.9 33.1
(2.31)9 (0.80) (2.13)h (2.84)9
b. Industry aggregate dataf
t+ 2 24 13.1 12.0
12.4 7.5
(0.71) (1.31)
t+ 3 10 21.0 28.7
4.6 23.1
(1.38) (1.17)
a Wilcoxon statistics are in parentheses. Critical values for two-tailed tests are 1.96 and 1.65 for 95% and
90% significance levels, respectively.
bRaw changes: X~t+k -t. All measures are reported as percentage change in the ratio of operating income to
sales.
c Restructuring surprise: Xt+k - E(XR+k It).
d Difference in surprises: [XjR+k- E(X!Rklt)] - [Xt+k - E(Xt+klt)].
Difference in differences: [XI k -Xt ]-[Xt+k - X].
Note that since there are an even number of firms in this panel and in year t + 3 of panel 1, there is no unique
median, so we report both medians. Columns 1 and 2 are blank for this panel because these estimators are based
only on restructuring firm data.
g Significant at 95%.
h
Significant at 90%.

mates are smaller and less significantthan the casts, respectively),while the associated median
medianrawchanges.For example,the rawchange differencein surprisesare 11.5%and 21.4%.
in the operating income to sales ratio for the Finally,we compareour theoreticallypreferred
median firm is 11.9%for the sample of 31 firms set of estimates,the differencein surprises,to the
with COMPUSTATdata in the second post- differencein differencesapproachcommonlyused
buyoutyear. Adjustingthese figuresfor time se- in the literature.The estimates differ when ex-
ries forecasts of expected performanceyields an pectationsof future performanceat restructuring
increase in the ratio of just 7.4%. While perfor- and control firms diverge. The positive and sig-
mance improvementspersist even after removing nificant performanceimprovementsexpected at
expected improvementsat restructuringfirms,we MBO firms (documented above) are somewhat
rejectthe hypothesisthat these improvementsare offset by the expected changesin performanceat
zero with less confidencethan for raw changes. control firms. As a result, the medians of the
The result of hypothesis B, that control firms difference in differences and difference in sur-
experienced negative performanceshocks, indi- prises are in most cases quite similar,even though
cates that the difference in surprises estimates we sometimes reject hypothesis C, that the ex-
should be larger than the restructuringsurprise pected changes at restructuringand controlfirms
estimateswhich ignore these shocks.Again table are identical.
3 is consistentwith our earlier findings.In every The reader may find it puzzlingthat although
instance, median differencesin surprisesexceed in some cases we reject hypothesisC, the median
restructuringsurprisesand are more significant. estimates of (3) and (4) are virtually identical.
For example, the median restructuringsurprises The answer relates to the distributionof differ-
in the second post-buyout year are 7.4% and ences in expected performancechanges for both
7.7% (using COMPUSTAT and Value Line fore- types of firms in this sample. The adjustments
RESTRUCTURING AND CORPORATE PERFORMANCE 511

(which distinguishbetween estimators(3 and (4)) The results of our hypothesis tests guide our
are close to zero for firms near the median for selection of MBO effect estimatorsand estimates.
estimator (3) and more substantialin the tails. Even though we find some evidence that our
Hence, the mediansare virtuallyunaffectedwhile estimatorprovidesan improvementover the con-
the significancelevels drop somewhat. ventional approach, our best estimates of the
In summary,our best estimate of the MBO MBO effect lend added supportto the view that
effect leads us to conclude that there are large MBOs enhance operatingperformance.
positiveincreasesin the ratioof operatingincome
to sales after the buyout. This measure of per-
formance improves by 6.8% between the last REFERENCES
pre-buyoutyear and the first post-buyoutyear. Brown, L., et al., "SecurityAnalyst SuperiorityRelative to
Performanceimprovementsby the second post- UnivariateTime-SeriesModels in ForecastingQuar-
buyout year range from 11.5% to 21.4%. Our terly Earnings,"Journal of Accounting and Economics
9 (Apr. 1987),61-87.
estimatesof performancein the thirdpost-buyout Bulow, J., L. Summers,and V. Summers,"Distinguishing
year indicateimprovementsof up to 20%to 30%. Debt fromEquityin the JunkBond Era,"in J. Shoven
The median ratio of operatingincome to sales in and J. Waldfogel(eds.), Debt, Taxes, and Corporate
Restructuring (Washington,D.C.: Brookings Institu-
the last pre-buyoutyear is 10.6%. Therefore a tion, 1990).
10%improvementin performancecorrespondsto Healy, P., K. Palepu, and R. Ruback,"Does CorporatePer-
a post-buyoutoperatingincome to sales ratio of formanceImproveAfter Mergers?"NBER Working
Paper #3348, May 1990.
11.7%.Given a medianpre-buyoutsales figureof Jarrell,S., "Do TakeoversGenerateValue? Evidenceon the
$570 million, this improvementtranslates to an CapitalMarket'sAbilityto Assess Takeovers,"mimeo,
increasein operatingincome of about $6 million. SouthernMethodistUniversity,Feb. 1991.
Jensen, M., "Agency Costs of Free Cash Flow, Corporate
If MBOs improveperformanceby, say, 30% over Finance, and Takeovers,"American Economic Review
longer horizons, and if these improvementsper- 76 (May 1986),323-329.
sist, then the presentvalue of the firm'sincreased Jensen, M., S. Kaplan,and L. Stiglin, "Effectsof LBOs on
Tax Revenuesof the U.S. Treasury,"Tax Notes (Feb.
operating income would be $180 million (dis- 6, 1989),727-733.
counted at 10%). Kaplan,S., "The Effectsof ManagementBuyoutson Operat-
ing Performanceand Value,"Journal of Financial Eco-
nomics 24 (1989).
VI. Conclusion Lichtenberg,F., and D. Siegel, "The Effects of Leveraged
Buyoutson Productivityand Related Aspects of Firm
Recent developments have focused attention Behavior,"NBER WorkingPaperNo. 3022,June 1989.
on the difficult question of whether corporate Lowenstein,L., "ManagementBuyouts,"Columbia Law Re-
restructuringaffects performance.The principal view 85 (1985), 730-784.
Morck,R., A. Shleifer and R. Vishny,"AlternativeMecha-
obstacle to measuringthe effect of restructuring nismsfor CorporateControl,"American Economic Re-
on performance is accounting for what would view 79 (4) (Sept. 1989).
have happenedto performanceif the reorganiza- Mueller, D., "Mergersand MarketShare,"this REVIEW 67
(May 1985).
tion had not occurred.Existingstudies have gen- O'Brien,P., "Analyst'sForecastsas EarningsExpectations,"
erally measuredthe effect as the change in per- Journal of Accounting and Economics 10 (1988),53-83.
formanceat the restructuringfirm,less the change Palepu, K., "PredictingTakeoverTargets:A Methodological
and EmpiricalAnalysis,"Journal of Accounting and
in performanceat a control firm. Economics 8 (1986),3-37.
We have shown in this paper that this conven- Ravenscraft,D., and F. Scherer, "Life after Takeovers,"
tional measurementapproach is one of various Journal of Industrial Economics (Dec. 1987a).
, Mergers, Selloffs and Economic Efficiency (Washing-
special cases of a more general measurement ton, D.C.: BrookingsInstitution,1987b).
framework. Moreover, the restrictions on the Shivdasani,Anil, "BoardComposition,OwnershipStructure,
general frameworkrequiredto justify the use of and Hostile Takeovers,"Journal of Accounting and
Economics 16 (1993), 167-198.
each special case are testable. Using data on Shleifer, A., and L. Summers,"Breach of Trust in Hostile
MBOs, we test and sometimes reject these re- Takeovers,"in A. Auerbach(ed.), Corporate Takeovers:
strictions,suggestingthat more flexibleestimators Causes and Consequences (Chicago: University of
ChicagoPress, 1988).
of the MBO effect may be preferable to more Smart,S., "Large Shareholdersand Contests for Corporate
commonlyused estimators. Control,"mimeo,IndianaUniversity,1992.

You might also like