Chapter 16 corporate risk management and pension asset allocation

Th e “pension put” hypothesis Sharpe, 1976; Treynor, 1977 implies that fi rms with underfunded pension plans should invest more in riskier assets such as equities to maximize the value o

16.4.2 Measures of Explanatory Variables 373

Th e t ax-based Black –Tepper h ypothesis i mplies t hat fi rms w ith o funded pension plans should invest in the most heavily taxed assets (such

ver-as bonds) to maximize their tax-savings because overfunded plans are less likely to default on their pension promises (Black, 1980; Tepper, 1981) Th e

“pension put” hypothesis (Sharpe, 1976; Treynor, 1977) implies that fi rms with underfunded pension plans should invest more in riskier assets (such

as equities) to maximize the value of the “put option” to default.†

However, the competing theoretical hypotheses lack consistent support from the small volume of empirical research on pension asset allocation (Friedman, 1983; Bodie et al., 1987; Peterson, 1996; Amir and Benartzi, 1999; Frank, 2002) Th e corporate asset allocation strategy over alternative asset investment categories (equities versus bonds) is not well understood based on fi ndings from prior empirical literature Bodie et al (1987) cau-tioned that further empirical research remains to be fi lled before a clear picture of these important corporate pension decisions can emerge

Th e objective of this chapter is to seek empirical regularities in the U.K

fi rms’ risk ma nagement of pension asset a llocation, a nd its ship w ith t heir pens ion f unding a nd pens ion-related acco unting po licy over an extended period of FRS 17 adoption (1998–2002) Th e U.K pension fund investment strategy has evolved substantially over the past decades

interrelation-Th e issue of the new U.K pension GAAP (FRS 17: ASB, 2000) in 2000 has transpired the risks of providing defi ned benefi t (DB) pensions to investors Pension trustees and managers started to attach greater importance to man-aging their pension risk exposures With the changing pension legislation

* In m id-July 2 008, F TSE 100 c ompanies re cognized £14 bi llion p ension d efi cits on t heir balance sheets.

† Pension Benefi t Guaranty Corporation insures U.S fi rms’ pension liabilities in full in the event of default Th e PBGC has a claim on 30% of the market value of the fi rms’ assets Th e PBGC’s insurance of pension benefi ts provide the fi rm a “put” option: it can shed its pension liabilities by giving the PBGC the assets in the scheme plus one-third of the fi rms’ assets.

and a g lobal convergence toward the fair-value-based pension accounting standards, changes in asset allocation strategy are taking place.

A new body of theoretical research concerning pension asset allocation has emerged over recent years (Sharpe, 1990; Leibowitz et al., 1994; Blake, 2003) Th ese studies developed asset/liability or surplus return models of portfolio diversifi cation taking into account not only asset returns and vari-ances but also changes in pension liabilities and their covariance with asset returns In light of these new theoretical insights on pension asset alloca-tion, this chapter develops and tests the hypotheses that U.K sponsors have managed t heir pension asset a llocation w ith an attempt to mitigate t heir exposure to the potential volatility risk of the fi nancial statements and/or cash fl ows during a period of accounting regulatory uncertainty

Controlling for the endogeneity among asset allocation, pension ing, and the expected rate of return assumption choices, the empirical evi-dence based on the cross-sectional data on a panel of 279 fi rm-years during the period from 1998 to 2002, suggests that the relationships among asset allocation, pension f unding, a nd related pension reporting choices a re, for most pa rt, consistent w ith t he corporate risk-management objective

fund-of hedging the cash contribution risks that stem from measuring pension assets and liabilities at a “fair value” basis

Th e rest of this chapter proceeds as follows: Section 16.2 describes the institutional background and critically reviews the prior literature Section 16.3 de velops r esearch h ypotheses Section 1 6.4 de scribes t he r esearch methodology, variable specifi cation, data, and descriptive statistics Results are presented in Section 16.5 and concluding remarks in Section 16.6.16.2 BACKGROUND AND PRIOR RESEARCH

Th e productive deployment of pension plan assets directly reduces costs

of f unding a D B pension plan (McGill a nd Gr ubs, 1989) Pr ior cal research also established that the asset allocation is the main deter-minant of the investment performance of a pension fund (Brinson et al., 1991; Ibbotson and Kaplan, 2000) Th us the decision to allocate plan assets among diff erent investment vehicles represents a critical pension manage-ment decision by employer sponsors

empiri-During the last four decades, U.K employer sponsors have invested the majority of their assets in equities (Blake et al., 1999) Davis (1991) observes that the U.K fi rms have maintained a substantially higher equity proportion than fi rms in the United States, Canada, Japan, and Germany However, co rporate st rategic a sset a llocations i n t he U nited K ingdom

evolved since recent years have seen rapid changes occurring in both the legislative and the accounting regulation for fi nal salary pensions in the United Kingdom.

Th e case of Boots group provides a perspective over the asset allocation decision undertaken by employer sponsors during a period of regulatory uncertainty In November 2001, the Boots group announced that its £2.3 bil-lion pension fund, one of the U.K.’s 50 largest funds with 72,000 members, had switched 100% of its pension assets from equity into long-dated high-quality bonds.* Duration-matching by investing in fi xed-income investment products, such as bonds, can eff ectively reduce the likelihood of the accumu-lated assets falling short of the long-term pension liabilities (Blake, 2003)

It is also noticeable that patterns of DB pension plan asset allocation

in the United Kingdom and United States have been relatively invariant over the last several years Th e average equity allocation for a typical U.K sponsor was 60% (Urwin, 2002) Recent U.S research suggests t hat t he actuarial smoothing in the valuation of pension assets and liabilities has contributed to the high equity allocations by sponsors (e.g., Gold, 2000; Coronado and Sharpe, 2003) Th e U.K fi nancial press also claimed that the “fair value” approach as espoused by FRS 17-style pension accounting standard would lead corporate sponsors to shift their asset allocation in favor of fi xed-income securities, in order to shield themselves against the potential volatility onto their fi nancial statements (e.g., Financial Times, March 20, 2004)

Very limited empirical research has focused on explaining corporate asset allocation decisions Friedman (1983) presents evidence suggesting that less-profi table companies with higher leverage and higher earnings variability tended to hold less in equities His results on asset allocation seem to suggest that fi rms ma nage t heir pension f und asset a llocations

to counterbalance the risks across fi rms that are stemming from product markets or fi nancial structure By contrast, Bodie et al (1987) fi nd that the proportion of assets allocated to equities is negatively related to the level of funding and positively related to the size of the company In other words, underfunded plans tend to hold more equities and less fi xed-income secu-rities Th e negative correlation be tween f unding a nd t he proportion of assets allocated to equities provides some support to the “pension put” hypothesis Th ei r fi ndings on asset allocations taken together suggest that

* Th e bonds are a close match for the maturity and the indexation of U.K pension liabilities, which has a weighted average maturity of 30 years and 25% are infl ation-linked.

fi rms do not manage their sponsored pension funds as if they are entirely separate entities from the sponsor (Friedman, 1983; Bodie et al., 1987).Nevertheless, U S.-based em pirical r esearch d uring t he 1 990s p ro-vides no evidence supporting a t ax arbitrage–based pension investment strategy Motivated by the inconsistency among theoretical and empiri-cal st udies, Frank (2002) reexamines t he ex tent to wh ich t axes a ff ect a

fi rm’s decision to allocate its DB plan’s assets between equity and bonds within a simultaneous system of equations, which attempts to capture the joint corporate capital structure and pension asset allocation decision in such arbitrage strategy Contrary to prior empirical research, Frank (2002)

fi nds evidence consistent with fi rms trading off tax benefi ts and nontax factors as described by Black (1980)

Motivated by the release of U.S pension GAAP (SFAS 87), Amir and Benartzi (1999) investigate the possibility of a relation between fi rms’ pen-sion accounting and investment choices during a post-pension accounting regulatory change period (1988–1994) in the United States Th ey contend that SFAS 87 provides managers the opportunity to choose between recog-nition and disclosure In particular, they focus on the question of whether the recognition of add itional m inimum pension l iability i n accordance with SFAS 87 aff ects asset allocation decision Indeed, they fi nd evidence that companies closing to a recognition threshold will make an economic decision of allocating more plan assets into fi xed-income investments Uncovering the existence of such a relationship implies that SFAS 87 has potential economic consequences for fi rms, consistent with the corporate

fi nance perspective Th e adoption of a “ fair value” approach in pension accounting (FRS 17), consistent w ith t he corporate fi nance perspective, implies a short-term volatility mismatch between pension assets and lia-bilities It is possible that U.K fi rms may manage their pension asset allo-cation in a way so as to mitigate their pension risk exposures attributable

to changing pension accounting regulation

16.3 DEVELOPMENT OF HYPOTHESES

16.3.1 Financial Reporting Risk

Th e release of new pension accounting information (FRS 17) may a lter the nature or the perception of risks of employers’ pension exposure One common assumption was that the adoption of FRS 17 would expose U.K

fi rms to signifi cant ba lance sheet volatility Under FR S 17, t he pension defi cits or surpluses are required to be recognized on the corporate bal-ance sheets once they arise By contrast, the former U.K pension GAAP

(SSAP 24: ASC, 1988) allows any surpluses or defi cits to be spread over the employees’ future working time, typically 15 years.

If sponsoring fi rms were forced to recognize their past funding tices on a “fair value” basis onto the balance sheets, the desired position-ing of the fi rm’s consolidated pension balance sheet may not be attainable solely t hrough pens ion ben efi t r eductions, such a s p lan ter minations Consequently, plan sponsors are exposed to risks of having potentially volatile fi nancial statements Managers can mitigate such fi nancial report-ing risk by choosing a diff erent mix of equities and bonds Investing in bonds has the advantage of obtaining a high correlation between assets and liabilities, thus reducing such fi nancial reporting risks

prac-Following Bergstresser et al (2006), a sensitivity measure (PENRISK)

is constructed to capture the variation in fi rms’ pension exposure to the potential fi nancial reporting risk PENRISK is calculated as the natural logarithm of the ratio of the market value of pension assets to total net assets in a fi rm year.*

Hypothesis 1 ( H1): Ceteris p aribus, t he per centage o f a ssets

invested in equities decreases as fi rms’ ex posure to t he potential

fi nancial reporting risk increases

16.3.2 Contribution Volatility Risk

Risky assets, such a s equities, are characterized by their volatile returns Financial theory suggests that the higher risk of equity investment is awarded b y t he h igher r eturn i t g enerates ( Markowitz, 1 952; Sha rpe, 1964) However, if the volatile return on the pension assets translates into changes in the required cash contribution, then risky assets will translate into more risky required contributions to the pension plan Th e main con-cern of “cash fl ow risk” managers is to minimize the volatility of changes

in cash fl ows (Culp, 2001) Friedman (1983) fi nds some evidence that fi rms have incentives to time their pension contributions so as to smooth the reported earnings Prior to FRS 17, fi rms’ past pension funding practices are not reported on fi nancial statements Shareholders and investors may judge the fi rm’s performance by its reported earnings rather than by more comprehensive c ash fl ow measures Coronado et al (2008) show that investors fa iled to d istinguish between pension a nd operating earnings

* Bergstresser e t a l ( 2006) s uggest t hat s uch a s ensitivity me asure c ollapses t he i nfl uence

of outliers and brings the distribution of t he ratio closer to t hat of a nor mally distributed random variable.

and capitalize them similarly Th e adoption of FRS 17 implies that tors a re becoming m ore aware of t he r elative ma gnitude a nd po tential risks of DB pension provisioning (Klumpes and Li, 2004) Consequently, prudent va lue ma ximizing managers face strong incentive to hedge t he riskiness of their future cash contributions.

inves-By matching pension assets with liabilities, that is, allocating plan assets into bonds, sponsoring fi rms can eff ectively reduce the volatility of their pension contributions, thus achieve to hedge their cash contribution risks

If the incentive is strong for sponsors to minimize the volatility of pension contributions, then it would be observed that fi rms with both extremely overfunded and underfunded plans invest in bonds because such extreme overfunding a nd u nderfunding a ff ord le ss fl exibility to adjust t he t im-ing of pension contributions than do fi rms with moderate funding levels Pension plans with greater defi cits are required to make defi cit-reduction contributions and those with greater surpluses have to conform to the tax regulations.* By contrast, pension contributions are fairly predictable for moderate funding levels, but less predictable when funding levels become more extreme Th e contribution risk hypothesis thus predicts a nonlin-earity relationship between variations in funding level and pension asset allocation, and provides an alternative risk-management explanation for the confl icting results from prior studies on the eff ect of pension funding

on asset allocation.† Th is discussion leads to the second hypothesis:

Hypothesis 2 ( H2): Ceteris p aribus, t he per centage o f a ssets

invested in equities increases as the funding level increases up to a specifi c point, then decreases as the funding level increases beyond this point

16.4 RESEARCH DESIGN

16.4.1 Empirical Model

Accounting researchers have recognized the importance of analyzing the potential endogeneity in the choices made by fi rms along diff erent dimen-sions (e.g., Beatty et al., 1995; D’Souza, 1998) Treating endogenous vari-ables as exogenous, or excluding relevant choice variables, leads to biased

* U.K corporate sponsors of defi ned benefi t pension plans with a f unded status in excess of 105% were subject to taxation at a rate of 35%.

† Amir a nd Benartzi (1999) fi nd some empirical evidence on s uch a non linear relationship between pension funding and asset allocation However, their study does not control for the potential endogeneity between pension funding and asset allocation.

and i nconsistent pa rameter e stimates Pr ior pens ion r esearch ha s n ot taken into account the simultaneity of pension asset allocation, funding, and r elated fi nancial r eporting ch oices.* C ausality i s t herefore u nclear, and t he same cross-sectional results can be r ationalized by a va riety of explanations.

It is possible that the corporate pension asset allocation policy is ditional upon fi rms’ long-term pension funding and their discretion over the l ong-term ex pected r ate o f r eturn o n pens ion i nvestments ( here-inaft er “ ERR”) Pr ior U S.-based r esearch ha s dem onstrated t hat fi rm management ha s attempted to eng age i n t he s moothing a nd spreading

con-of pension costs over time (Picconi, 2006) Corporate sponsors can cise the discretion in the level of ERR assumptions to change the pattern and the magnitude of pension liabilities, thus mitigating their contribu-tion volatility risk Th ey could also increase the allocation of equities in their pension fund investment portfolio to justify the high level of the ERR assumptions Klumpes et al (2009) provide evidence that corporate pen-sion termination decision is indeed not independent of their discretionary ERR choices Consequently, a simultaneous equations model is employed

exer-to control for the simultaneity and identify the impact of pension ing risks on the pension asset allocation decision

report-Th e hypotheses concerning corporate pension asset allocation decision are tested by employing a simultaneous model with three equations: asset allocation, funding, and pension actuarial assumption choices It is based

on the assumption that sponsors can adjust their asset allocation, ing, and related pension reporting choices simultaneously Specifi cally, the system of simultaneous equations is specifi ed as follows:

* Mitchell and Smith (1994) employs simultaneous equations model approach to i nvestigate pension funding in the U.S public sector Th ey attempt to control the simultaneity between the required per worker annual contribution (REQ), actual pension plan funding in the pub- lic sector (ACT), and average worker compensation package (AVEPAY).

FUNDSQ is the squared value of FUND

PENRISK i s t he sens itivity m easure o f fi rms’ pe nsion ex posure t o reporting risk

X1it, X2it, a nd X 3it a re a v ector o f predetermined co ntrol va riables i n

three respective equations

γt represents dummy variables for years 1998–2002

Equations 16.1 through 16.3 model pension asset allocations, funding, and ERR reporting choices, respectively ε1t, ε2t, and ε3t are error terms

16.4.2 Measures of Explanatory Variables

Multiple p roxies a re de veloped t o co ntrol f or t he co mpeting t cal ex planations o n t he co rporate pens ion a sset a llocation dec ision o f Equation 16.1 Th e “pension put” hypothesis (Sharpe, 1976; Treynor, 1977) implies that fi rms for which the pension “put” option is more valuable (i.e., more in the money) will hold more of the most risky assets, presumably equities, and vice versa Th e “put” option is likely to be more valuable for underfunded plans, for unprofi table companies or fi rms with higher vari-ability in their cash fl ows, or fi rms with more debt

heoreti-Consistent w ith p rior l iterature ( Friedman, 1 983; B odie e t a l., 1 987; Frank, 2 002), t hree p roxies u sed t o m easure t he va lue o f pens ion “ put option” to sponsoring fi rms include leverage (LEV), profi tability (PROF), and fi rm risk (STDCF) LEV is calculated as the long-term debt divided

by total tangible assets Higher leverage implies less debt covenant slack and higher probability of fi nancial distress Th us the “put” option is more valuable to highly levered fi rms PROF is calculated as the mean return on shareholders equity over the preceding 10 years, which is a proxy for long-term profi tability Less-profi table fi rms a re less l ikely to f ulfi ll the fi xed payments of retirees’ benefi ts and thus more likely to invest more in equi-ties to maximize the value of the “put” option to default (Sharpe, 1976) STDCF is calculated as the standard deviation of operating cash fl ows over the preceding 10 years defl ated by the book va lue of equity Firms with

* FUND me asures t he r atio of t he p ension pl an’s tot al a ssets to it s tot al prom ised b enefi t obligations.

higher va riability in t heir cash fl ows are likely to invest more in equity

to maximize the “pension put.” Friedman (1983) fi nds a negative relation between fi rm risk, measured as income variability, and the percentage of assets invested in equities He proposes a “risk off setting story” to interpret his fi nding Th e a rgument is t hat risky fi rms tend to off set t he risks by investing in less-risky assets in their pension plans, such as bonds

Th e “ Black–Tepper” h ypothesis p redicts t hat fi rms sh ould o verfund their pensions because tax arbitrage enables fi rms to earn a tax-free rate of return on investments TAXST is included as a proxy for fi rm’s average tax rate TAXST is calculated as the total reported taxes minus the change in deferred taxes over the preceding 10 years defl ated by beginning year total assets Firms with higher tax rate would gain more by investing the assets

in fi xed-income securities

It is important we also control for various economic determinants of pension asset allocation posited by prior research Prior U.S.-based stud-ies fi nd the plan demographics infl uences asset allocations (e.g., Amir and Benartzi, 1999; Friedman, 1983) Th e maturity of t he pension l iabilities may be an important determinant aff ecting the asset allocation decision

by U.K sponsors in the current economic environment when most of the funds g radually mature a nd dema nd nontrivial fi xed benefi t payments Firms with mature pensions may wish to invest more assets in bonds so

as to achieve better asset/liability matching, thus reduce the likelihood of the assets falling short of obligations PRET, measured as the percentage of vested member over the total number of vested and non-vested members,

is included as a control for the maturity of pension plan

It is also argued that the U.K pension fund management is partially driven by the herding behavior (Klumpes and Whittington, 2003) Th is suggests that U.K pension funds oft en benchmark their investment per-formance against other funds’ performance and relevant market indices

Th e co ntemporaneous ac tual r ate o f r eturn o n pens ion a sset po rtfolio (MKRTN) is included to control for this argument MKRTN is calculated

as the actual rate of return on a weighted market portfolio with an lent asset mix

equiva-Th e pension funding regression (Equation 16.2) adopts the empirical framework employed by Francis and Reiter (1987) to explain corporate pension funding strategy, while controlling for the endogeneity among asset allocation (%EQUITY) and ERR assumption choices Th e “fi nancial slack” eff ect has emphasized t he pension f und’s usefulness as a so urce

of corporate liquidity or as a store of temporarily excess corporate funds

(Myers and Majluf, 1984) Such slack could be kept in the form of either liquid assets, unused debt capacity, or pension assets Th e fi nancial slack hypothesis predicts that the fi rms should overfund its pensions to build excess assets, which can accumulate at the pretax rate and be used when

fi rms require funds to fi nance positive NPV projects Th e proxy for the rate of undertaking new investments (RUNI), measured as the sum total

of capital expenditure plus R&D expenditure divided by total assets, trols for t he “fi nancial slack” hypothesis of high-level funding Sharpe (1976) argues that “pension put” is of greatest value to underfunded pen-sion plans, therefore risky fi rms should underfund their pensions to max-imize the “put option” value STDCF is included as a proxy to control for the fi rm risk for the “pension put option” incentive of low-level funding

con-Th e ERR regression (Equation 16.3) explains cross-sectional variations

in reported ER R levels In add ition, Equation 16.3 i ncludes t he level of funding as an additional endogenous variable Bodie et al (1987) fi nd a negative association between the size of the pension plan and the percent-age of bonds allocated in the pension portfolio So a general control vari-able is included, plan size (LNSIZE), in all three equations.*

16.4.3 Sample and Data

Th e main constraint on the sample size is the availability of detailed sion asset composition.† Th e proprietary asset a llocation data are hand-collected f rom t he p rofessional p ublication “ Pension F und a nd Th ei r Advisers” book (PFTA 1988–2004) Th e sample period consists of period over 1998–2002 During this period, U.K sponsors were subject to both legislative-imposed minimum funding solvency restrictions and diff eren-tial pension accounting regulatory requirements To be included in the sample, the sponsor fi rst had t o be a p ublicly listed FTSE 350 fi rm that sponsors a t l east o ne D B pens ion sch eme w ith co mplete pens ion a sset allocation data available Second, to increase the power of empirical tests,

pen-fi rms with more than 5% of their pension assets as “unclassifi ed” are deleted.‡Finally, fi rms with missing data required for analysis are deleted

* A Hausman specifi cation test is performed on t he equation system specifi ed to check the existence of endogeneity (Hausman, 1978) Th e result indicates that exogeneity of asset allo- cation decision (% EQUITY) can be rejected at 5% level.

† Frank (2002) also noted the obstacle to investigating DB investment policy is obtaining the asset allocation data necessary to compute percentage of bonds invested by pension funds.

‡ Amir a nd B enartzi (1999) a nd Fr ank (2002) d eleted fi rms with 5% of the assets that are

“unclassifi ed” in their studies.

Following t he abo ve c riteria, t he co mplete a sset a llocation d ata c an only be obtained between 60 and 70 fi rms per year Aft er eliminating out-liers, the fi nal sample comprises 279 fi rm-year observations All the data used for this study is collected from the fi nancial statements of the sample

fi rms, Datastream and “PFTA” book

xed-an increase of nearly 4% allocated to bonds Overall the sample sion asset a llocation displays a sl ow trend toward a g reater percentage

pen-of bonds among the overall pension asset composition during the study period (1998–2002)

Panel A o f Table 1 6.2 p resents t he c ross-sectional d istribution o f

%EQUITY by year %EQUITY va ries signifi cantly across t he fi rms with

a st andard de viation o f 1 3.66 f or t he poo led s ample T o ex amine t he TABLE 16.1 Distribution of Pension Asset Composition by Year

Notes: Th is table presents the descriptive statistics on the distribution of pension asset

com-position of the sample U.K fi rms Th e asset investment categories are the U.K equity, the overseas equity, the U.K fi xed interest, the overseas fi xed interest, index-linked bonds, property, and cash.