1、ReceivedReceivedAcceptedAvailablemarvariableyate pensionJournal of Corporate Finance 15 (2009) 257271Contents lists available at ScienceDirectJournal of Corporate Financejournal homepage: rms affected the interests of creditors, stockholders, labor unions, and the Pension Benet Guaranty Corporation
2、. Inhindsight, the event provides an opportunity to study the workings of internal capital markets by observing the effect ofincremental nancial constraints that were not entirely related to the protability of future investment.The existence of internal capital markets appears to be generally accept
3、ed, but our understanding of how they function isincomplete. At stake are implications for the valuation of multi-segment rms and the relation between nancial leverage andcorporate investment policy. Much of the judgment on the efcacy of internal capital allocation rests on the empirical linkagebetw
4、een cash ow and both the quantityand quality of capital expenditures. Some studies suggest that the availability of internalThe term “perfect storm” was frequentlstock prices and bond yields on corporresources does indeed affect both the levelopportunities. Other research indicates thatinuences.We t
5、hank Toni Whited for making her GAUSS programseminar participants at Saint Louis University for helpful Corresponding author. Tel.: +1 314 977 8169.E-mail address: Aldersonslu.edu (M.J. Alderson).0929-1199/$ see front matter 2008 Elsevier B.V. Alldoi:10.1016/j.jcorpn.2008.12.002used duringthe 200002
6、 period to describethe impactof the simultaneous decline inplan sponsors. The shift from surplus to decit funded status experienced by“The Perfect Storm” not only impactsplans have been enjoying. Recentfunding contributions and PBGCthe balance sheet of pension plans, but it may ruin the contribution
7、 “holiday” manyket losses coupled with declining 30-year U.S. Treasury interest rates could triggerrate premiums.” Hewitt Investment Group commentary, October 2001.Internal capital marketsInvestment policyPension contributions1. IntroductionTobins q, and cash ow bothbefore and after the event. Wend
8、mixed evidence on the changein therelationbetweeninvestmentandq,which maybearesultofmeasurementerrorin q.Wend stronger evidence for the conclusion that after the pension storm,rms with underfundedpension plans directed more investment towards segments that produce higher cash ow. 2008 Elsevier B.V.
9、All rights reserved.JEL classication:G31G32Keywords:16 January 2008in revised form 2 December 200815 December 2008online 24 December 2008We examine capital expenditures in multi-segment rms before and after the “perfect storm”that affected pension plans between 2000 and 2002, when bond yields and st
10、ock prices bothfell precipitously. Our sample of rms went from having overfunded to underfunded pensionplans as a result of the storm. We examine the segment-level relation between investment,Were internal capital markets affected by the perfect pension storm?Michael J. Alderson, Brian L. BetkerSain
11、t Louis University, United Statesarticle info abstractArticle historyof corporate investment and method by which these funds are allocated to differentany apparent relation may be the result of measurement error and endogenousavailable and for other useful guidance, an anonymous referee, Bidisha Cha
12、krabarty, Tom Miller, andcomments.rights reserved.258 M.J. Alderson, B.L. Betker / Journal of Corporate Finance 15 (2009) 257271In other words, documenting an empirical relation between cash ow and capital expenditures is complicated by the fact thatcurrent cash ow could reect the future protability
13、 of investment as well as the funds available to undertake it. Cash owthereforeisntareliablemeasureofnancialresourcesbecauseitmightalsoreectthequalityofinvestmentopportunities.Recentresearch by Rauh (2006) exploits the independence of mandatory pension contributions and the quality of investmentoppo
14、rtunities to demonstrate that the nancial constraints created by mandatory pension contributions act to reduce the level ofcapital expenditures by sponsoring rms. In this paper, we examine multi-segment rms for evidence that increased pensionfunding obligations affect the allocation of investment be
15、tween high q and low q segments.The sample of rms studied here experienced a severe reversal of the funded status of their dened benet pension plans dueto the sharp decline in both equity values and bond yields between 1999 and 2002. For those rms, the simultaneous drop inpensionassetsandincreaseinp
16、ensionliabilitiescombinedtorapidlychangepreviouslyoverfundedplanstoseverelyunderfundedonesby denition, plans with dedicated assets less than the discounted value of projected benet payments. Extant minimumfunding requirements mean that the shift in funded status led to a signicant increase in contri
17、butions to the pension fund, andwith it a new opportunity to examine the impact of large changes in nancial constraints on corporate investment activity.Financial constraints matter because they may decrease or enhance investment efciency. The analysis of segment-level datahas contributed signicantl
18、y to our understanding of the relation between nancial constraints and investment policy. Smallsample studies of cash ow shocks have provided one source of evidence. Lamont (1997) examines capital expenditures bydiversied oil-related companies and nds evidence that the 1986 oil market crash reduced
19、subsidies to underperforming non-oilsegments.Blanchardetal.(1994)ndthatcashwindfallsresultinincreasedinvestmentbythelowqsegmentsofdiversiedrms.Largesample studies have documented a tendency fordiversiedrms to direct funds tolowervalueduses through thecapitalallocation process. Shin and Stulz (1998)
20、show that more protable segments within diversied rms subsidize the investment ofothersegments,butnotalwaysthemostvaluableones.ScharfsteinandStein(2000)studythesensitivityofsegmentinvestmentinmulti-segment companies and nds that it is less sensitive to Tobins q in the cross-section than among single
21、-segment rms.Rajan et al. (2000) nd that diversied rms tend to favor low q investment opportunities more than single-segment entities.Restructuring activities can also affect nancial constraints, and there is evidence that restructuring events that tighten controlover internal capital market activit
22、y reduce the costs of managerial discretion. Gertneret al. (2002) examine investmentefciencyaround spinoffs andnd that among the spunoff divisions, capital expenditures are allocated to higherq segments after the event.Dittmar and Shivdasani (2003) report a similar nding for the remaining segments o
23、f a parent rm following the divestiture of asegment. Ahn and Denis (2004) consider both the parent and spunoff subsidiary and also nd that, post spinoff, there is asignicantincreaseinmeasuresofinvestmentefciency.However,olakandWhited(2007)re-examinedivestituresandspin-offsanddemonstratethattheimprov
24、ementsininvestmentefciencyfollowingthoseeventsarecorrelatedwith,butnotcausedby,therestructuring.This study bears the strongest relation to the work by Peyer and Shivdasani (2001) that examined the effect of leveragedrecapitalizations on the investment activity of 22 rms operating in more than one bu
25、siness segment. Using a sample of rms thatconductedaleveragedrecapbetween1982and1994,PeyerandShivdasanindthataftertherecapitalizationcapitalexpenditureswere allocated away from high q segments, and toward projects that provided the greatest short term cash ow. Their resultssuggest that the impact of
26、 high leverage on investment policy is not restricted to the classic form of underinvestment in whichrms entirely forgo value-creating projects; rather, the costs of high leverage can include a reduction in investment efciencycaused by the need to meet heavy payment obligations. We hereafter refer t
27、o these costs as “distortion costs.”Ourpaperexamineswhethersharplyhigherpensioncontributionsgeneratedthosesamedistortioncosts.Mandatoryincreasesinpensioncontributionsthatfollowareversalofpensionfundingstatusmaketheeventequivalentinmanyrespectstoaleveragedrecapitalization. Here we exploit that simila
28、rity to provide additional evidence on the relation between high leverage and thetendency to favor high cash ow over high value investment opportunities.Leveraged recapitalizations involve large and discrete increases in nancial constraints which make them ideal subjects in astudy of internal capita
29、l market functions. At the same time, they possess characteristics that suggest the need for an examinationof a broader, more random sample. First is the need to control for the endogeneity problem, meaning the possibility that thedistortion costs generated in the aftermath of a leveraged recapitali
30、zation are symptomatic of the same poor management qualitythat may have necessitated the leverage increase in the rst place. Peyer and Shivdasani (2001) nd that the rms in their sampleengagedinsystematicoverinvestmentpriortotheleveragedrecap.Theself-interestedbehaviorthatledtochronicoverinvestmentma
31、y have also generated the distortion costs that they document if “bad” managers sought to avoid the personal costs of nancialdistress byfavoring high cashowoverhigh value projects. The sample examinedbyPeyerand Shivdasani is also small (22rms),creating a natural interest in the study of a larger sam
32、ple.Weexaminermsthatexperiencedatighteningofnancialconstraintsbecauseofasharpreversalof thefundedstatusof theirpension plans. The sample of 125 multi-segment companies that we examine is larger in number and drawn from a broader arrayof industries than in prior studies, and is not dominated by rms t
33、hat chose to restructure their assets or their debt. The benetsassociated with these characteristics are partially offset by the potential effects of calendar time clustering and the reduced powerof tests involving a more gradual imposition of nancial constraints stemming from the smoothing characte
34、ristics of pensionmathematics. We also consider the implications of measurement error in the q ratio for our results.We ndthat priortothereversalof fundedstatus,thecompanies in oursampledirectedmore fundstosegmentswith higherqratios.Followingthesharpdeclineinfundedstatus,therelationwasreversed,withm
35、orefundsbeingdirectedtolowerqsegments.However, virtually identical results are obtained from a control sample of dened benet sponsors that stayed in surplus over thesame time period. Since the control sample of surplus rms is presumably not subjected to the same incremental nancialconstraints, we co
36、nclude that the observed change in q-sensitivity more likely reects the effect of measurement error in q asdescribed by olak and Whited (2007) than the effect of additional distortion costs. We show that applying the Erickson andWhited (2002) procedure to control for the effect of measurement error
37、in the q ratio causes the reduction in q-sensitivity in thereversal sample to disappear.In contrast to the control sample of surplus rms, the companies that experienced a sharp reversal in funded status allocatedcapital expenditures to segments with higher cash ows after the event when compared to b
38、efore. More importantly, the sameresult is obtained using the EricksonWhited procedure. Thus while we cannot make any denite statement regarding changes in259M.J. Alderson, B.L. Betker / Journal of Corporate Finance 15 (2009) 257271the quality of investment, it does appear that enhanced nancial cons
39、traints lend greater importance to more immediate cashows.Section 2 describes the data and Section 3 presents the results to an analysis of funds transferred between segments. Section 4reports the results to the regression analysis that examines coefcient changes before and after the event. Section
40、5 analyzes asample of rms that stayed in pension surplus after the storm, and uses the Erickson and Whited (2002) GMM method to controlfor the effect of measurement error in q. Section 6 concludes the paper.2. Sample selection and sample characteristicsThe market value of dened benet pension plan as
41、sets fell sharply in relation to the present value of benet obligations fromearly 2000 through late 2002. This shift caused the funded status of many dened benet pension plans to swing from surplus(assets exceed liabilities) to decit during that period. With the shift in funding status came a signic
42、ant increase in contributionsby plan sponsors to their respective pension fundsa nancial event not unlike a leveraged recapitalization. Our objective is todetermine whether the nancial constraints imposed by that shift affected the investment policy of those rms. We thereforeexamine a sample of 125
43、companies that experienced the most severe shifts in their pension funded status from the end of 1999through the end of 2002.The process of selecting that sample begins with the collection of three variables for all non-nancial rms on COMPUSTAT attheend ofscalyears 1999 and 2002. Those variableswere
44、 thefair value of planassets (item 287),the projected benetobligation(item 286) and market capitalization, which is the product of scal year end stock price (item 199) and shares outstanding (item25). We chose the projected benet obligation as the measure of plan liabilities because it estimates the
45、 present value of futurebenets underthe assumption that thermwillcontinuetosponsortheplan.1Data foreachof thethree variableswas available inboth years for 1441 companies.The difference between plan assets and the projected benet obligation is the pension surplus or decit. A total of 745companies in
46、the set of 1441 were overfunded (plan assets exceed plan liabilities) at the end of 1999 and underfunded (planliabilities exceed plan assets) at the end of 2002. The pension surplus or decit is normalized by the stock market capitalization atbothvaluationdatestoproducethemeasurethatwerefertoasthefun
47、dingratio.FollowingFranzoniandMarin(2006),weemploystock market capitalization as the denominatorin the funding ratio in ordertogaugethepension decit in relation tothenancialstrength of the plan sponsor.In order to concentrate the analysis on rms with the most severe shifts inpension funding, we next
48、 eliminated all but the 231rms with pension decits at the end of 2002 thatexceed 10% of thestock market capitalization at that time.2The median fundingratio for those companies was 6.9% at the end of 1999 and 19.6% at the end of 2002. Finally, we retained only those rms thatreported multiple segment
49、s in every year of the sample period.As in prior studies, we use Tobins q as a measure of growth opportunities. Tobins q is dened as the market value of assetsdivided by the book value of assets, and the market value of assets is dened as the market value of common equity plus the bookvalue ofassets,minusthebookvalue ofequity,minusdeferredtaxes.We compute imputedsegmentq ratiosas themedianTobinsq of single s
