Wilshire Consulting Summary of Wilshire Consulting Publications March 31, 2010

Summary of Wilshire Consulting Publications Prepared for Tacoma Employees’ Retirement System First Quarter 2010

Page 1 Wilshire Consulting 2010 Wilshire Report on Corporate Pension Funding Levels April 1, 2010

2010 Wilshire Consulting Report on Corporate Pension Funding Levels

Julia K. Bonafede, CFA, President Steven J. Foresti, Managing Director Russell J. Walker, Vice President April 1, 2010

Copyright  2010, Wilshire Associates Incorporated Page 1 Wilshire Consulting 2010 Wilshire Report on Corporate Pension Funding Levels April 1, 2010

Summary of Findings o The financial health of corporate pension plans improved somewhat in 2009 from the substantial declines experienced in 2008. Defined benefit pension assets for S&P 500 Index companies increased by $108.8 billion, from $883.0 billion to $992.9 billion, while liabilities increased $21.9 billion, from $1,101.5 billion to $1,191.2 billion. As a result, the aggregate funding ratio (assets divided by liabilities) for all plans combined increased from 80.2% to 83.4% and the -$217.9 billion deficit at the beginning of the year shrank to a -$198.2 billion deficit. (Exhibit 1) o Ninety-two percent of corporate pension plans are under-funded, which is barely higher than the 91% reported for the previous year. The median (50th percentile) corporate funded ratio is 79.0%, which represents an improvement from 73.2% last year. (Exhibit 3) o Pension fund performance experienced a decisive turnaround from the significant losses of 2008. The median 2009 investment return was 16.2%, a sharp rebound from the -27.4% median return of 2008, following five years of rising median returns of 8.2% in 2007, 11.2% in 2006, 8.5% in 2005, 10.8% in 2004, and 17.1% in 2003. o Interest rates used to discount future benefits fell over 2009, contributing to the overall increase in pension liabilities for the year. The median discount rate fell from 6.25% to 6.00%, while total liabilities increased 8.1% for the year. (Exhibit 8 & Exhibit 9) o The combined pension expense for the S&P 500 Index companies in our study was $30.5 billion for 2009, up markedly from $16.4 billion a year ago. Regular annual pension expense accruals from employee service and interest expense on existing liabilities totaled $93.2 billion in 2008, 0.9% higher than the $92.3 billion a year ago. (Exhibit 11) o S&P 500 Index companies contributed $54.9 billion into their defined benefit plans in 2009, a notable increase from the $30.0 billion contributed in 2008. o Aggregate benefit payments from corporate pension plans increased somewhat during the past year. Benefit payments totaled $75.0 billion in 2009, compared to $71.4 billion during the previous year. o The distribution of pension liabilities and assets of S&P 500 Index companies is relatively concentrated among the largest plans. As of the end of fiscal year 2009, more than half of the total pension assets and liabilities were held by the 24 and 25 largest plans when ranked by asset and liability size, respectively. Conversely, the smallest 100 plans when ranked by asset and liability size made up just 2.4% and 2.7% of the total asset and liability pool, respectively. (Exhibit 6)

Copyright  2010, Wilshire Associates Incorporated Page 2 Wilshire Consulting 2010 Wilshire Report on Corporate Pension Funding Levels April 1, 2010

Financial Overview

The Data

This is Wilshire Consulting’s eighth study covering defined benefit plans sponsored by S&P 500 Index companies. Wilshire’s practice is to collect data on U.S. pensions from 10-K filings for companies in the S&P 500 Index at fiscal year-end. All data for fiscal years 2009 and 2008 are based on S&P 500 Index constituents as of year-end 2009 and, therefore, may differ slightly from the list of companies represented in earlier years.

Assets, Liabilities, and Funding Ratios

The financial health, as measured by the aggregate funding ratio, of corporate pension plans improved slightly in 2009 from the substantial decline in funding ratio of 2008, according to our latest survey of 308 companies in the S&P 500 Index that maintain defined benefit plans. Exhibit 1 shows the change in aggregated assets, liabilities, and surplus (assets minus liabilities) for the surveyed companies from 2000 to the most recent 2009 fiscal reporting year.

Exhibit 1 Corporate Pension Assets & Liabilities (in $billions)

$1,400 1,294 1,235 1,196 1,216 1,191 1,154 1,160 $1,200 1,102 1,069 1,031 1,199 1,200 998 1,112 993 $1,000 1,070 892 883 991 964 $800

$600

$400

$200 244

$0 95 34 17 -84 -$200 -123 -90 -177 -218 -198 -$400 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009

Assets Liabilities Surplus/(Deficit)

The aggregate pension deficit, represented by the difference between the market value of assets and liabilities, shrank by $19.7 billion from a deficit of -$217.9 billion at the end of 2008 to a deficit of -$198.2 billion at the end of 2009. At the same time, the aggregate funding ratio, equal to assets divided by liabilities, increased from 80% to 83%.

Copyright  2010, Wilshire Associates Incorporated Page 3 Wilshire Consulting 2010 Wilshire Report on Corporate Pension Funding Levels April 1, 2010

The aggregate figures in Exhibit 1 mask considerable differences among individual corporate plans. Exhibit 2 shows a histogram of funding ratios for the 308 corporate pension plans in our study. Exhibit 2 Distribution of 308 Corporate DB Pension Plans by Funding Ratio

87 86 100% Funded

59

35

13 11 4 5 3 3 2 0

Twenty-four of the 308 corporations, or 7.8%, have pension assets that equal or exceed liabilities. In comparison, 28 of the 308 corporations’ DB plans, or 9.1%, were fully-funded or running a surplus at year-end 2008. Note that Wilshire’s 2002 study, at the end of the 2000- 2002 bear market, found 11% of S&P 500 DB plans at fully-funded or surplus status.

Exhibit 3 displays graphically how the distribution of corporate pension funding ratios has changed during the past nine years. Four lines are charted, three corresponding to a percentile rank and one corresponding to the aggregate funding ratio. The 50th percentile or median, corporate funding ratio declined rapidly from 112% at the end of 2000 to 78% at the end of 2002, but climbed steadily to end 2007 at 97%. Fiscal year 2008 experienced a sharp reversal in fortunes however, with the funding ratio ending the year at 73%. The global equity markets started to rally strongly from their severe reversals in March 2009. As a result, the median corporate funding ratio ended 2009 at 79%, recovering some of the ground lost in 2008.

In 2000, the 125% aggregate funding ratio translated into a $244 billion surplus. The bear market coupled with falling interest rates over the following two years worsened the financial condition of corporate pension plans by $421 billion ($244 billion surplus in 2000 to -$177 billion deficit in 2002). The improving stock market during the subsequent five years improved the financial condition of corporate pension plans by $272 billion (-$177 billion deficit in 2002 to a $95 billion surplus in 2007), despite a declining interest rate environment for most of that period. In 2008, the $95 billion surplus of 2007 was completely wiped out and replaced by a year-end deficit of -$218 billion, contributing to the steep drop-off in funding ratio for the year.

Copyright  2010, Wilshire Associates Incorporated Page 4 Wilshire Consulting 2010 Wilshire Report on Corporate Pension Funding Levels April 1, 2010

Although 2009 proved to be a recovery year for equity markets, liabilities for these corporate DB plans rose along with assets (see below), and although funding ratios improved, they could not keep pace with global asset markets’ performance. Exhibit 3 Corporate Funding Ratios

140% 132% 130% 125% 120% 112% 110% 104%106% 108%107% 101% 100% 96% 101% 97% 97% 93% 96% 95% 89% 93% 90% 92% 83%90% 91% 87% 88% 82% 87% 86% 85% 83% 83% 81% 78% 80% 81% 79% 75% 74% 72% 73% 70% 70% 67% 64% 60% 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009

75th 50th Agg. Fund Ratio 25th

Exhibit 4 shows the combined assets, liabilities, and surplus for the 308 surveyed companies, broken down into Global Industry Classification Standards (GICS) sectors for the 2009 fiscal reporting year.

Exhibit 4 Corporate Pension Assets & Liabilities (in $billions) by GICS Sector

(19.0) 105.1 Consumer Discretionary 86.0 (16.0) Consumer Staples 83.8 67.8 (13.5) Energy 50.6 37.2 (5.3) 116.4 Financials 111.1 (13.8) Health Care 83.1 69.2 (62.7) 346.6 Industrials 283.9 (8.6) 83.9 Information Technology 75.3 (25.8) Materials 114.2 88.4 (9.5) Telecomm Services 96.9 87.4 (23.9) 110.5 Utilities 86.6

Surplus / (Deficit) Liabilities Assets

Copyright  2010, Wilshire Associates Incorporated Page 5 Wilshire Consulting 2010 Wilshire Report on Corporate Pension Funding Levels April 1, 2010

As was the case in 2008, all sector sub-groups ran an aggregate pension deficit at year-end 2009. Industrials, which comprise a hefty 28.6% of total assets and 29.1% of total liabilities among the S&P 500 Index companies studied, contributed 31.6% (-$62.7 billion) to the total -$198.2 billion deficit, with the Materials and Utilities sectors representing 13.0% (-$25.8 billion) and 12.1% (-$23.9 billion) of the total deficit respectively. Lockheed Martin Corporation’s -$10.7 billion plan deficit represents 17.1% of the Industrials sector’s aggregate deficit and 5.4% of the aggregate deficit of these 308 corporate plans as of year-end 2009. Ford Motor Company, the last Big Three US automaker still in the S&P 500, reported a year-end 2009 pension shortfall of -$6.2 billion, representing 32.6% of the Consumer Discretionary sector’s total deficit and 3.1% of the total pension deficit of all company plans in this sample.

Exhibit 5 summarizes the funding ratios for the surveyed companies in 2009 and 2008, broken down by their GICS sectors.

Exhibit 5 Corporate Pension Funding Ratios by GICS Sector

Consumer 84.3 Discretionary 81.9 78.8 Consumer Staples 80.9

60.6 Energy 73.4 90.6 Financials 95.4 73.0 Health Care 83.3 78.6 Industrials 81.9 Information 89.6 Technology 89.7 74.7 Materials 77.4 91.6 Telecomm Services 90.2 72.1 Utilities 78.4

Funding Ratio, 2008 Funding Ratio, 2009

The Financials sector had the highest funding ratio, at 95.4%, while the Energy sector had the lowest funding ratio at 73.4%. However, the Energy sector accounted for the smallest proportion of assets and liabilities in the surveyed companies, approximately 2.8% and 3.7%, respectively. Coming in second place, Telecomm Services posted a funding ratio of 90.2% followed up by the Information Technology sector’s 89.7% funding ratio. Overall however, eight of ten sectors experienced improved funding ratios for the year ending 2009.

The Concentration of Plan Assets and Liabilities

While the aggregated pool of S&P 500 Index defined benefit plans is only 83.4% funded this year, the concentration of assets and liabilities within this set of plans indicates that a relatively small sub-set of plans has an overwhelming impact on the entire pool of plans. Exhibit 6

Copyright  2010, Wilshire Associates Incorporated Page 6 Wilshire Consulting 2010 Wilshire Report on Corporate Pension Funding Levels April 1, 2010 outlines the concentration of plan assets and liabilities when ranked by size and provides funding ratio data for the subset of plans examined.

Exhibit 6 Concentration of 308 Corporate DB Pension Plans by Assets and Liabilities Assets Liabilities Median Funded Ratio Funded Ratio Range 25 Largest Plans 50.6% 49.2% 86.2% 67.5% to 111.3% 100 Smallest Plans 2.4% 2.68% 75.6% 45.5% to 170.1%

The largest 25 plans when ranked by assets and liabilities represent 50.6% and 49.2% of assets and liabilities, respectively, of the 308 total plans. The median funded ratios stand at 86.2% and 75.6% for the 25 largest and 100 smallest plans, respectively.

Exhibit 7 plots all of the 308 plans sampled in this study based on their 2009 year-end funded ratio and plan liability size. While a simple trend line appears to suggest there is some relationship between liability size and funding ratio, the relationship is less pronounced when the few largest outliers are removed from this scatter plot.

Exhibit 7 Liability Size and Funding Ratio

$50.0

$45.0

$40.0

$35.0 ) s n o i

l $30.0 l i b $ (

$25.0 s e i t i l

i $20.0 b a i

L $15.0

$10.0

$5.0

$0.0 0 15 30 45 60 75 90 105 120 135 150 Funding Ratio (%)

Exhibit 8 provides a combined accounting of S&P 500 Index corporate pension plans for the 2009 fiscal year.

Copyright  2010, Wilshire Associates Incorporated Page 7 Wilshire Consulting 2010 Wilshire Report on Corporate Pension Funding Levels April 1, 2010

Exhibit 8 Change in Assets & Liabilities for 2009 In Billions % of BOY Liabilities - Beginning of Year (BOY) $ 1,101.5 Service costs 24.9 2.3% Interest costs 68.3 6.2% Benefit payments (75.0) -6.8% Actuarial losses (gains) 57.5 5.2% Other 14.0 1.3% Liabilities - End of Year (EOY) $ 1,191.2 8.1%

Assets - Beginning of Year (BOY) $ 883.1 Contributions 54.9 6.2% Actual return on assets 117.8 13.3% Benefit payments (75.0) -8.5% Other 12.2 1.4% Assets - End of Year (EOY) $ 992.9 12.4% Note: 13.3 % actual return on assets is based on beginning of year asset value.

Pension Plan Liabilities

There are three recurring items that affect the growth in liabilities. The first item is service cost. This cost arises from employees earning additional benefits from another year of service. Service cost, which changes little from year to year, added $24.9 billion, or 2.3%, to aggregate pension liabilities in 2008. The second item is interest cost. Liabilities are determined by discounting expected future benefit payments. As each year passes, liabilities increase by the annualized interest cost because there is one less year to discount future benefits. This cost item should also remain predictable from year to year. Thirdly, liabilities are reduced by benefits paid during the year since they represent a payment against the company’s pension liability.

If these recurring items were the only changes, then corporate pension liabilities would have grown by just $18.2 billion, or 1.5%. Instead, liabilities increased by $89.7 billion, or 8.1%, in 2009. A portion of the difference lies within the “other” category. The “other” category refers to changes in liabilities, either negative or positive, that arises from the addition or subtraction of liabilities from unrecognized prior service cost, curtailments, or corporate acquisition activity. Large dollar figures in the “other” category are however, often the result of companies acquiring other pension plans, as part of a corporate merger or acquisition. When that happens, there also is a corresponding entry into the “other” category under assets. The primary contributors to this net amount during 2009 however, were more evenly distributed among the different sub-categories of “other.”

 Merck (+$5.2 billion attributable to its merger with Schering-Plough)  Pfizer (+$4.3 billion attributable to its acquisition of Wyeth)

Copyright  2010, Wilshire Associates Incorporated Page 8 Wilshire Consulting 2010 Wilshire Report on Corporate Pension Funding Levels April 1, 2010

 Centurytel (+$3.5 billion attributable to its acquisition of Embarq)  Dow Chemical (+$2.5 billion attributable to its acquisition of Rohm & Haas)  Bank of America (-$1.1 billion due to plan transfer to new 401(k) plan) The final item that affects liabilities in Exhibit 8 is “Actuarial losses (gains)” which refers to changes in liabilities, either negative or positive, that arise when actual actuarial losses (gains) occur due in part to changes in discount rates – in general, when discount rates rise, the present value of liabilities fall. From 2001 to 2005 companies reported large actuarial losses partially as a result of falling discount rates. In 2006 however, the median discount rate increased from 2005, marking a turn in direction from the previous five year trend. Median discount rates increased through year-end 2008, where the rate stood at 6.25%. However, 2009 saw discount rates fall again, leading to an aggregate actuarial loss and contributing to the overall increase in aggregate pension liabilities over the year.

Exhibit 9 FAS 87 Discount Rate by Percentile

8.00 8.00 7.75 7.75 7.50 7.50 7.25 7.25 7.25 6.98 7.00 7.00 7.03 6.87 6.90 6.75 6.75

6.50 6.50 6.50 6.56 6.25 6.25 6.34 6.40 6.25 6.15 6.25 6.00 6.19 6.20 6.00 6.00 6.00 6.00 6.00 5.90 5.75 5.96 5.95 5.75 5.62 5.75 5.80 5.75 5.60 5.60 5.51 5.46 5.50 5.50 5.26 5.21 5.00 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009

95th 75th 50th 25th 5th Moody's Aa Moody's Aaa

Exhibit 9 shows the distribution of discount rates used by S&P 500 Index companies from 2000 to 2009. The median, or 50th percentile, discount rate has recently risen from 5.62% in 2005 to 6.25% in 2008, but has fallen back to 6.0% as of year-end 2009.

Each year companies select an FAS 87 discount rate that approximates a settlement rate for their pension liabilities, taking into consideration the current rates of return on high-quality fixed income investments. The Moody’s Aa and Aaa corporate bond yields are generally thought to be fair estimates for pension liability settlement rates. The Moody’s Aa corporate bond yield had fallen from 7.48% at end-of-year 2000 to 5.50% at end-of-year 2005. As

Copyright  2010, Wilshire Associates Incorporated Page 9 Wilshire Consulting 2010 Wilshire Report on Corporate Pension Funding Levels April 1, 2010 interest rates fell, companies were forced to lower their discount rates, thereby increasing the accounting value of total pension liabilities. However since year end 2005, both Aa and Aaa corporate bond yields have trended up, ending the year 2008 at 6.50% and 5.82% respectively. Reflecting the overall normalization of credit spreads over 2009, these yields have fallen from their 2008 levels; the Aa and Aaa yields stood at 5.62% and 5.40% respectively as of 12/31/2009. The 22-basis point spread between Aaa and AA yields is closer to historical quality spread levels than the 68-b.p. spread seen at year-end 2008. Exhibit 9 plots these year-end corporate bond yields with the historical distribution of discount rates.

Pension Plan Assets

There are three recurring items that cause changes in plan assets. The first is corporate contributions to the pension plan, which totaled $54.9 billion or 6.2% of assets during 2009. In an ideal world where the contribution and financial reporting actuarial methods and assumptions are identical and all assumptions are perfectly accurate, contributions would be at a level equal to service costs, as companies pay for new benefits earned during the year. This would occur when assets equal liabilities and when assets earn a return equal to the discount rate used to value liabilities. If assets are below liabilities, or asset returns fall below the discount rate, then corporate contributions typically rise to make up for the short fall and vice versa. This reactive behavior is prone to overshoot error wherein corporate contributions may substantially exceed or fall short of service costs depending on the experience of the previous year(s). Below in Exhibit 10, service cost and employer contribution are charted against funding ratio from 2002 to 2009. Reflecting the impact of funding levels on plan contributions, it is clear that company contributions slowed as market returns boosted funding levels. Given the recent decline in the aggregate funding ratio for 2008, the companies in this study dramatically increased their aggregate contributions to their DB plans for calendar year 2009.

Exhibit 10 Employer Contributions and Funding Ratio

70,000 120% o i 60,000 t 100% a R

50,000 g

80% n s i

n 40,000 d o n i

l 60% u l F i

30,000 B

e t $

40% a

20,000 g e r

10,000 20% g g - 0% A 2002 2003 2004 2005 2006 2007 2008 2009

ServiceCost Employer Contributions Funding Ratio

Copyright  2010, Wilshire Associates Incorporated Page 10 Wilshire Consulting 2010 Wilshire Report on Corporate Pension Funding Levels April 1, 2010

In 2009, the $54.9 billion in company contributions exceeded the $24.9 billion in aggregate service costs. The chart reflects a general pattern of over-contributing relative to the level of service costs in years where funding ratios are below 100%, indicative of an effort to reach 100% funding, and under-contributing relative to the level of service costs when funding ratios exceed 100%. For example, Hershey Foods’ DB pension plan ran a surplus at year-end 2007 (funded ratio of 134.3%); its 2007 contributions to the pension totaled $16 million compared to a service cost of $43 million. As markets fell in 2008, Hershey increased its contributions to the plan to $33 million, compared to a service cost of $30 million; however, the funded ratio of the plan fell to 95.8%. 2009 found Hershey contributing $54 million to the plan, over twice the annual service cost of $26 million; combined with the 2009 global market rally, this helped Hershey’s DB plan achieve an improved funded ratio of 98.4% at year end.

The second item is the actual return on plan assets, which posted a $117.8 billion gain for 2009. The aggregate actual annual returns as a percentage of beginning of year assets for the eight years from 2001 to 2008 have been (in ascending chronological order): -7.8%, -8.9%, 17.3%, 11.7%, 8.5%. 13.0%, 10.3% and -20.4%. During the past year, a 8.1% rise in liabilities combined with a 12.4% increase in assets resulted in an increase in aggregate funding ratio from 80.2% to 83.4%.

The third recurring item affecting plan assets is benefit payments, which totaled $75.0 billion in 2009 and reduced assets and liabilities by the same amount. The final item is “Other” and largely represents unrecognized prior service cost, curtailments, or corporate acquisition activity adding the pension plans from companies that were acquired. As previously mentioned, the most notable contributors to this net amount during 2009 were Merck, Pfizer, CenturyTel, Dow Chemical and Bank of America, whose collective charge to the “other” category was $9.7 billion.

The Impact of Pension Expense (Income) on Corporate Earnings

Much has been written about the impact of pension expense (income) on corporate earnings as a result of the Pension Protection Act1 which created a more defined and stricter set of guidelines for public companies sponsoring DB plans. Exhibit 11 provides an aggregate accounting of pension expense (income) for calendar 2009 for the 308 S&P 500 Index companies in our study.

1 For more information on the Pension Protection Act please refer to Wilshire Consulting’s “Pension Protection Act of 2006” research note.

Copyright  2010, Wilshire Associates Incorporated Page 11 Wilshire Consulting 2010 Wilshire Report on Corporate Pension Funding Levels April 1, 2010

Exhibit 11 Calculation of 2009 Pension Expense (Income) In Billions Service Costs $ 24.9 Interest Costs 68.3 Expected Return on Plan Assets (135.5) Losses (Gains) 72.8 Pension Expense (Income) $ 30.5

There are four items that comprise pension expense (income). The first two are service costs and interest costs, which were described in Exhibit 8. There is some controversy surrounding the third item, the “expected return on plan assets.” Market value accounting might suggest that service costs and interest costs be offset by the actual return on assets, a $117.8 billion gain, in calculating pension expense (income). If this were so, pension income would equal -$24.6 billion in 2009. Instead, pension expense was $30.5 billion in 2009 because asset returns are smoothed. In other words, to reduce the short-term volatility of asset returns on corporate net income, the expected return on plan assets from year to year reflects amortized gains and losses from the current year and prior years.

The expected rate of return for pension assets has been coming down in recent years (see Exhibit 12). The median expected return was 9.50% at the end of 2000 and has fallen to 8.00% at the end of 2009. The expected return assumption is multiplied by the level of assets to arrive at a dollar value of expected investment earnings that is credited against service and interest costs. In 2009, companies collectively expected their assets to earn $135.5 billion, and it was this number that was used in the calculation of pension expense and corporate net income. However, pension plans incurred actual gains of $117.8 billion, which represents a -$17.7 billion net loss. Yet, differences between accounting earnings and market-based earnings cannot continue unabated. Once these differences exceed 10% of assets, they must be amortized over time and appear in the expense calculation as “Losses” or “Gains,” the fourth item in Exhibit 11.

Copyright  2010, Wilshire Associates Incorporated Page 12 Wilshire Consulting 2010 Wilshire Report on Corporate Pension Funding Levels April 1, 2010

Exhibit 12 FAS 87 Long-Run Return on Assets by Percentile

10.50 10.37 10.00 10.0 9.75 9.75 n

r 9.50 u 9.30 9.23

t 9.25 e 9.00 9.00 9.00 9.00 R 9.00 9.00 9.00 9.00 l 9.0 8.75 a 8.60 8.75

u 9.00 8.50 8.50 n 8.25 8.50 8.50 8.50 n 8.50 8.50 8.50 8.25 8.20 A 8.00 7.86 8.25

d 8.0 8.00 8.00

e 8.00 8.00 8.00 7.88 7.54 7.75

m 7.47

u 7.23 s

s 7.00

A 7.0 6.81 6.80 6.50 6.50

6.0 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009

95th 75th 50th 25th 5th

Although the median expected return on plan assets assumption has fallen during the past nine years, from 9.50% in 2000 to 8.00% in 2009, many pension accounting critics believe that this assumption is still too high. Wilshire Consulting’s long-term forecast for the return on corporate pension assets is approximately 6.4%, based on the average asset allocation of corporate pension plans as noted in the companies’ 10-Ks. However, individual pension plan expected returns will vary considerably depending upon their unique asset allocation. It should be noted that Wilshire Consulting’s asset class return forecast is for the next ten years while the horizon for the assumed return in corporate financial statements is an unspecified long term period. Additionally, Wilshire Consulting’s assumed returns are for the asset classes with no consideration of potential value added from successful active management. Please see below, “Pension Plan Asset Allocation”.

In Exhibit 13, the difference between the median expected return on assets and actual return on assets, or Net ROA, is plotted against the aggregate funding ratio for each year from 2002 to 2009. From this perspective, funding ratios, which are measured on the right-hand axis, display a high sensitivity to net return on assets. Over a string of good years, funding ratios can grow at a nice steady pace; however, in years of large market disruption, funding ratios can give back gains of several years abruptly. Thus the pro-cyclical effect of returns driving funding ratios is linked back to corporate pension plan contributions rising and falling with the tide of funding ratios rather than with the more consistent increase in current service costs. In light of the funding targets set out by the Pension Protection Act, company contributions will have to increase substantially to achieve full funding by 2011.

If the deflating of the technology bubble at the turn of the century is viewed as an analogous event, positive market returns for several years forward will have to be experienced, five years in the case of the technology bubble, to grow out of the current funding deficit. Yet market

Copyright  2010, Wilshire Associates Incorporated Page 13 Wilshire Consulting 2010 Wilshire Report on Corporate Pension Funding Levels April 1, 2010 growth assumptions did not do all the work in the past as displayed in Exhibit 10; companies today will at minimum face steeper contribution schedules to meet full funding requirements under the targets of the Pension Protection Act.

Exhibit 13 Net Return on Assets and Funding Ratios

20.0% 120.0%

10.0% 110.0% o

0.0% 100.0% i t a A R

O g

R -10.0% 90.0%

n t i e d N n

-20.0% 80.0% u F

-30.0% 70.0%

-40.0% 60.0% 2002 2003 2004 2005 2006 2007 2008 2009

Net ROA Agg Funding Ratio

Pension Plan Asset Allocation

We now turn to the asset allocation of these 308 corporate pension plans. Exhibit 14 summarizes the average asset class exposures of these funds at year-end 2009:

Exhibit 14 Average Asset Allocation of 308 Corporate DB Pension Plans, Year-End 2009

Hedge Funds 0.8% Other (incl. Cash) 8.0% Private Equity 1.4%

Real Estate 1.7%

Total Equity Total Fixed 34.0% 54.1%

Copyright  2010, Wilshire Associates Incorporated Page 14 Wilshire Consulting 2010 Wilshire Report on Corporate Pension Funding Levels April 1, 2010

Portfolio expected return and risk are calculated by combining Wilshire’s assumptions for the major asset classes and each retirement system’s actual asset allocation. These long-term return and risk assumptions are illustrated in Exhibit 15:

Exhibit 15 Wilshire Consulting’s 2010 Asset Class Assumptions

Expected Return (%) Risk (%) US Equity 7.50 16.00 Non-US Equity 7.50 17.00 Global Equity 7.75 16.00 US Bonds 4.25 5.00 Non-US Bonds 3.90 4.00 Real Estate 6.50 15.00 PrivateEquity 10.00 26.00

Exhibit 16 contains summary statistics on asset allocation for these corporate DB plans. The median allocation2 of publicly-traded equity is 56.2%, compared to 33.6% for publicly-traded fixed income assets. However, as the lowest and highest columns suggest, there is considerable variability in allocations among individual systems. Using Wilshire’s asset class assumptions to calculate return projections, the median corporate DB plan in this sample has an expected return of 6.5% (compared to a 6.4% return calculated using average asset exposures, as cited above). This result is 1.5% less than the current median expected return on assets (as stated in the companies’ 10-Ks) of 8.0%, but 0.5% above the median plan long-term discount rate of 6.0%:

Exhibit 16 Summary Asset Allocation Statistics for 308 S&P 500 DB Plans

Lowest Median Highest TotalEquity 0.0% 56.2% 100.0% TotalFixed 0.0% 33.6% 100.0% RealEstate 0.0% 0.0% 22.0% PrivateEquity 0.0% 0.0% 47.0% Hedge Funds 0.0% 0.0% 56.0% Other(incl.Cash) -3.8% 4.0% 100.0%

Expected Returns 3.0% 6.5% 8.8% Expected Risk 1.3% 10.0% 17.9%

2 The “Median” column in Exhibit 16 represents the median for each asset class and therefore does not sum to 100%. The median expected return is based on the median fund return, not on the median asset mix.

Copyright  2010, Wilshire Associates Incorporated Page 15 Wilshire Consulting 2010 Wilshire Report on Corporate Pension Funding Levels April 1, 2010

The next exhibit presents the risk/return profile of these plans as of year-end 2009 using Wilshire-calculated projected long-term rates of return and standard deviation of returns for each plan. Two horizontal lines plot the median expected ROA and median long-term discount rate for the plans; the median projected plan return and risk are also plotted as a dotted horizontal and vertical line, respectively:

Exhibit 17 Projected Risk/Return Profile of 308 S&P 500 DB Plans, 2009

10.0

9.0 Median Expected Return on Assets from 10-Ks: 8.0%

8.0 Median Projected Return: 6.5% (annualized) ) 7.0 % (

n r u t

e 6.0 R

d e t c e

j 5.0 o r Median Long-Term DiscountRatefrom 10-Ks:6.0% P

m

r 4.0 e T - g n

o Median Projected Risk: 10.0% (annualized) 3.0 L

2.0

1.0

0.0 0.0 2.0 4.0 6.0 8.0 10.0 12.0 14.0 16.0 18.0 20.0 Annual Risk (%)

Performance Statistics Projected Using Wilshire Consulting 2010 AC Assumptions

Exhibit 18 addresses the relationship between asset allocation and funding for the 308 corporate plans in our study. The allocation to equity asset classes, a proxy for investment aggressiveness, is plotted on the vertical scale. The funding ratio is on the horizontal scale.

Copyright  2010, Wilshire Associates Incorporated Page 16 Wilshire Consulting 2010 Wilshire Report on Corporate Pension Funding Levels April 1, 2010

Exhibit 18 Asset Allocation & Funded Status of 308 S&P 500 DB Plans, 2009

120.0%

100% FundedStatus

100.0%

80.0% ) % (

e r u s

o 60.0% p

x R² = 0.002 E

y t i u q E

40.0% l a t o T

20.0%

0.0% 0.0% 20.0% 40.0% 60.0% 80.0% 100.0% 120.0% 140.0% 160.0% 180.0% Funded Ratio (%)

The vertical line at 100% funded ratio visually separates under- and overfunded plans. A casual glance at the plot suggests no discernable relationship between equity allocation and funded status; indeed, the r-squared of the two data series calculates to nearly zero (0.002), indicating a near-random relationship. The corporate DB pension plans in our study show a broad spectrum of asset allocations that appear to be unrelated to their funded status.3

3 We would like to thank Amy Hemphill, Jason Samansky, Christopher Barry, Andrew Chen, Claire Cohen, Tom Dunlap, Alex Ford, Jeremy Henningsen, Jerry Hsu, Sasha Johnson, Mark Lampa, Henry Lim, Vardges Markosyan, Taveen Miloyan, Andy Schroeck and Brian White for their diligence in the data collection for this report, as well as valuable assistance from Alex Browning in the preparation of this report.

Copyright  2010, Wilshire Associates Incorporated Page 17 Wilshire Consulting 2010 Wilshire Report on Corporate Pension Funding Levels April 1, 2010

Important Information

This material contains confidential and proprietary information of Wilshire Consulting, and is intended for the exclusive use of the person to whom it is provided. It may not be modified, sold or otherwise provided, in whole or in part, to any other person or entity without prior written permission from Wilshire Consulting.

This material is intended for informational purposes only and should not be construed as legal, accounting, tax, investment, or other professional advice. Past performance does not guarantee future returns. This material may include estimates, projections and other "forward-looking statements." Due to numerous factors, actual events may differ substantially from those presented.

The information contained herein has been obtained from sources believed to be reliable. Wilshire Consulting gives no representations or warranties as to the accuracy of such information, and accepts no responsibility or liability (including for indirect, consequential or incidental damages) for any error, omission or inaccuracy in such information and for results obtained from its use. Information and opinions are as of the date indicated, and are subject to change without notice.

Wilshire is a registered service mark of Wilshire Associates Incorporated, Santa Monica, California. All other tradenames, trademarks, and/or servicemarks are the property of their respective holders.

Copyright © 2010 Wilshire Associates Incorporated. All rights reserved. Information in this document is subject to change without notice. No part of this publication may be stored in a retrieval system, transmitted, or reproduced in any way, including but not limited to, photocopy, photograph, magnetic or other record, without the prior written permission of Wilshire Associates Incorporated, Santa Monica, CA. U.S.A. www.wilshire.com

Copyright  2010, Wilshire Associates Incorporated Page 18 Wilshire Consulting Risk-Focused Diversification February 11, 2010

Utilizing Leverage within Asset Allocation

Steven J. Foresti, Managing Director Michael E. Rush, CFA, Vice President Email: [email protected]

February 11, 2010

Copyright  2010, Wilshire Associates Incorporated Page 1 Wilshire Consulting Risk-Focused Diversification February 11, 2010

Introduction

The asset allocation process – determining how much of a portfolio to invest in equity, fixed income, etc – is well established as being the most important step in the investment process. Studies have shown that over 90% of the return variability among portfolios is attributable to this decision4. Until recently, most institutional investors were well served following the enduring theory of mean-variance optimization. After the 2001-2002 market environment, explicitly considering liabilities in the asset allocation decision increased in importance as the average plan’s funding ratio went from well funded to significantly underfunded. The spike in risk and convergence of correlation during 2007- 2008 has once again refocused attention on the asset allocation decision. As a means of improving diversification and efficiency, the concept of utilizing leverage within the asset allocation process is being considered by numerous institutions. As will be discussed, leverage can be employed to better balance risk by focusing greater assets on lower risk, higher Sharpe ratio asset classes and away from higher risk equities. The theory behind the idea is relatively clear however its implementation introduces new and complex risks to the management of an investment portfolio. Wilshire Consulting will review this strategy in the following research paper in greater detail by quantifying the risk and return implications and highlighting operational limitations. First, a discussion behind the theory and the issue it attempts to resolve is warranted.

Expanding the Efficient Frontier through Leverage

A common approach to asset allocation is to create a series of efficient portfolios based on risk and return expectations for a collection of investable asset classes. While this approach is structured to identify portfolios with optimal achievements of return at each level of risk, it is important to understand how an investor's return target can impact portfolio diversification. The further out investors reach for return, the fewer asset classes that are available to meet the return target and diversification must be sacrificed. Since most institutional investors target returns in the high single-digits, portfolios become dominated by relatively high risk asset classes like equities. Exhibit 1 displays Wilshire Consulting’s 2010 capital market assumptions for several representative asset classes versus an 8.0% return hurdle5 to demonstrate this point graphically.

4 According to a study by Brinson, Singer and Beebower, Financial Analysts Journal 1991: “Determinants of Portfolio Performance II: An Update” 5 The average actuarial rate from Wilshire Consulting’s 2009 state and city & county funding studies is 8.0% and the expected return from Wilshire's 2009 corporate funding study is 8.20%.

Copyright  2010, Wilshire Associates Incorporated Page 2 Wilshire Consulting Risk-Focused Diversification February 11, 2010

Exhibit 1 The Impact of Return Targets on Diversification

12.00%

10.00% Private Markets

8.00% Return Hurdle GlobalEquity n r u

t 6.00% U.S. High Yield e

R PublicReal Assets 4.00% U.S. Core Fixed Income Cash 2.00%

0.00% 0.00% 5.00% 10.00% 15.00% 20.00% 25.00% 30.00% Risk

As can be seen above, only equity and equity-like investments approach this 8% expected return target. As such the resulting portfolio must either be dominated by equity investments or pursue aggressive levels of alpha to achieve the desired rate of return. The cherished and benevolent investment principle of diversification becomes sacrificed in terms of asset class concentration and, to a greater degree, contribution to risk. Exhibit 2 contains an efficient frontier comprised of four of the above asset classes6 with two portfolios highlighted, a low risk and high risk portfolio; Portfolios A and B, respectively. To the right of the chart are the allocations and risk contributions for each portfolio.

Exhibit 2 Low vs. High Risk Portfolio Exposures

14.00

12.00 Portfolio A Weight Risk Cntrb 10.00 Global Equity 19% 41.7% U.S. Core Fxd Inc 29% 11.4% ) U.S. High Yield 19% 23.1% %

( 8.00 Portfolio B n Public Real Asts 33% 23.7% r u t Portfolio A

e 6.00 Portfolio B R Weight Risk Cntrb 4.00 Global Equity 54% 75.7% U.S.CoreFxdInc 0% 0.0% 2.00 U.S. High Yield 32% 20.7% Public Real Asts 14% 3.6% 0.00 0.00 2.00 4.00 6.00 8.00 10.00 12.00 14.00 16.00 18.00 Risk (%)

6 The asset classes include are Global Equity, U.S. Core Fixed Income, U.S. High Yield and Public Real Assets.

Copyright  2010, Wilshire Associates Incorporated Page 3 Wilshire Consulting Risk-Focused Diversification February 11, 2010

Portfolio A is invested in all four available asset classes and the risk contribution from each is relatively balanced. However, based on Wilshire's 2010 assumptions, the portfolio's expected return is a modest 5.7%; likely lower than what most institutional investors require. Portfolio B has a much higher expected return of 7.0% but contains no core fixed income and derives three-quarters of its risk from equities. It is this undesirable concentration of risk against the 2008 market backdrop that has refocused some attention on a strategy for improving the diversification and efficiency of a portfolio while still targeting relatively high returns.

The theoretical concept of Risk-Focused Diversification (RFD) utilizes leverage to move along an improved efficient frontier where each step up the return spectrum sacrifices less in asset diversification than would otherwise be possible without leverage. Exhibit 3 displays such an opportunity set, represented by the orange line, and a third portfolio, Portfolio C, with the same expected risk as Portfolio B in Exhibit 2 but with a higher expected return. As was the case with Portfolio A (the low risk portfolio), Portfolio C employs all four of the available asset classes but also improves upon the concentrated contribution to risk that was evident in Portfolio B.

Exhibit 3 Leveraged Efficient Frontier

10.00

9.00 Portfolio C 8.00 Weight Risk Cntrb Portfolio C Global Equity 38% 47.0% 7.00 Portfolio B U.S.CoreFxdInc 44% 8.3%

) U.S. High Yield 34% 22.9% 6.00 %

( Public Real Asts 56% 21.0% Portfolio A n

r 5.00 Borrowing -72% 0.8% u t

e 4.00 R

3.00

2.00

1.00

0.00 0.00 2.00 4.00 6.00 8.00 10.00 12.00 14.00 16.00 18.00 20.00 Risk (%)

The orange line, or “tangent frontier,” begins at a point that implies an investor has all of their money in cash. Such an investor could then move along the tangent frontier as they begin to diversify, eventually arriving at the original efficient frontier. After passing along the curve, denoted by the dotted line in Exhibit 3, the tangent frontier starts to diverge from the original efficient frontier as leverage begins to be deployed. To accomplish this, an investor would need to “sell cash,” meaning borrow, and invest the

Copyright  2010, Wilshire Associates Incorporated Page 4 Wilshire Consulting Risk-Focused Diversification February 11, 2010 proceeds in their portfolio, thereby creating leverage.7 The investor would then need to pay a cash-like return to the counterparty that is providing the leverage. Throughout this analysis, we will assume that the borrowing rate will equal an expected LIBOR rate plus a financing rate of 0.25%. As is visible in the chart in Exhibit 3, Portfolio C is allocated in excess of 100% of assets owned but is more efficient than the previous high risk portfolio. The gray area between the curves denotes the efficiency improvement that is available through leverage. At lower expected risk levels, the difference is minimal but increases exponentially at higher risk levels as an inability to utilize leverage on the original efficient frontier becomes more costly as the portfolio forgoes diversification in favor of a concentrated exposure to higher returning equity assets. The following pie- charts summarize the contributions to risk for the three portfolios described above and illustrate the potential benefits of utilizing leverage in the asset (and risk) allocation mix.

Exhibit 4 Summary of Risk Contributions

PortfolioA PortfolioB PortfolioC

Global Equity U.S. Core Fixed Income U.S. High Yield Public Real Assets

As an investor stretches for greater return in moving from Portfolio A to B, the concentrated risk from equity and equity-like assets is dramatic. Notice, however, that Portfolio C from the leveraged frontier is able to essentially restore the balance of risk contributors from low-return Portfolio A at the enhanced 11% expected risk level of Portfolio B. This brief discussion demonstrates that utilizing leverage during the asset allocation process is a promising theoretical concept. However, new and complex issues arise during implementation. The following analysis will review these issues including cost, deciding on the amount of leverage and its affects within an asset-liability framework.

Re-Leveraging the Capital Markets

“Leveraging” an investment involves explicitly or implicitly borrowing cash to invest with assets owned. In the introduction, it was posited that the use of leverage may result in more efficient portfolios; due in part to improved diversification and greater investment in higher Sharpe Ratio asset classes. Exhibit 5 begins to simplify the above

7 In practice, the investor generally does not explicitly borrow cash, but instead achieves leverage through the embedded borrowing implicit in various derivative instruments that can be utilized to provide synthetic (i.e. notional) market exposure. These operational issues are discussed in the "Framework for Considering the Degree of Leverage" section.

Copyright  2010, Wilshire Associates Incorporated Page 5 Wilshire Consulting Risk-Focused Diversification February 11, 2010 discussion to the most commonly held asset classes – U.S. equity and fixed income. We will use Wilshire’s expected return and risk assumptions for a base case portfolio as a point of reference for the historical analysis that follows in the next section. However, rather than limit our focus to a single set of assumptions at a specific point in time, Wilshire’s asset class assumptions for the past three years8 are included in Exhibit 5. Note that our current 2010 assumption suite reveals an important point that is also evident in the historical analysis; that leverage will not necessarily result in a more efficient portfolio at all points in time. Nevertheless, as we demonstrate in the Appendix, an expanded asset-liability analysis that allows for greater diversification through a broader set of asset classes will present the potential benefit of introducing leverage under Wilshire’s 2010 asset class assumptions. Each set of assumptions is utilized in Exhibit 5 to create an initial portfolio that is invested in U.S. equity and core fixed income in a common 60/40 ratio, which we will refer to as a “Typical” portfolio throughout the remainder of this paper, and contains no leverage. A second portfolio is then created in each case that re-configures this “Typical” market exposure through leverage to help achieve a higher expected return at the same expected risk level. The 2008 and mid-year 2009 assumptions result in 61% and 54% leverage, respectively, and lead to allocations that heavily favor core fixed income. The return enhancement using the 2010 assumptions is slight to the point of making a leveraged strategy appear unimpressive.

Exhibit 5 Leverage Opportunities under various Market Outlooks 2008 ACA 2009 Mid Year ACA 2010 ACA Asset Allocation U.S. Equity 60% 44% 60% 47% 60% 58% U.S. Core Fxd Income 40% 117% 40% 107% 40% 54%

Borrowed Cash 0% -61% 0% -54% 0% -12% Leverage Ratio 1.00 1.61 1.00 1.54 1.00 1.12

Expected Return 7.19% 7.51% 6.74% 6.99% 6.45% 6.45% Expected Risk 10.36% 10.36% 10.36% 10.36% 10.36% 10.36%

Enhancement 0.32% 0.25% 0.01%

As suggested above, Exhibit 5 demonstrates that a leveraged strategy may not improve efficiency under all conditions. The results further reveal that changes to forward- looking market expectations from one year to the next may lead to an adjustment in the optimal amount of leverage utilized within an ongoing program. We will use Wilshire’s 2009 mid-year assumptions – the most recent set to reveal a meaningful benefit from a leveraged strategy – for the remainder of this section simply to identify asset allocations that can then be analyzed historically. Under this scenario, employing more of the asset

8 The mid-year 2009 assumptions were used for last year’s expectations as the beginning of the year assumptions were based on a point-in-time of extreme market dislocation.

Copyright  2010, Wilshire Associates Incorporated Page 6 Wilshire Consulting Risk-Focused Diversification February 11, 2010 class with a higher Sharpe ratio – core fixed income – allows for a higher expected return at the same level of expected risk. Exhibit 6 contains efficient frontiers utilizing only U.S. equity and fixed income, with and without leverage. Borrowing was limited at 50% for simplicity although at this moderate level the leveraged efficient frontier does shift upward across much of the risk spectrum. Anywhere the frontier shifts, an improvement in efficiency is expected (i.e. increase return or decrease risk).

Exhibit 6 Efficient Frontier including U.S. Equity and Core Fixed Income

12.00

10.00

8.00 n r u

t 6.00 e R

4.00

2.00

0.00 0.00 4.00 8.00 12.00 16.00 20.00 Risk UnleveragedEF Leveraged EF max 50% 60/40 Eqty/Core FI

Also in Exhibit 6 and represented by the red triangle is the “Typical” 60/40 portfolio. To improve the efficiency of this portfolio, an investor would look to increase return and move upward toward the leveraged frontier or decrease risk and move left from their current position. This 60/40 base case portfolio will next be used to quantify the expected efficiency improvements – based on mid-year 2009 assumptions – possible through leverage and the historical record of representative portfolios.

Expected and Historical Performance

The underlying mechanics of using leverage to reduce risk or enhance return in the previous section involved borrowing cash to invest in core fixed income and, at the same time, decreasing the amount invested in equities. It should be noted that leverage within asset allocation does not hinge solely on the core fixed income asset class. Rather, portfolios throughout this initial example contain only bonds and equity for the sake of keeping the presentation simple. As additional asset classes are included within a total portfolio, investors will benefit from both natural diversification and an increased effectiveness of leverage. Improved diversification will raise the initial efficient frontier, resulting in higher Sharpe ratios across the entire risk spectrum, so that more efficient portfolios are then being leveraged.

Copyright  2010, Wilshire Associates Incorporated Page 7 Wilshire Consulting Risk-Focused Diversification February 11, 2010

The correct mix of assets and leverage needed to improve efficiency is found by identifying the points on the leveraged frontier where expected return and then expected risk are equal to the “Typical” 60/40 portfolio. Exhibit 7 details those portfolios along with the expected improvement in efficiency. Also in Exhibit 7, we have utilized historical data from 1926 through 2009 to measure how these same portfolios would have performed historically. U.S. Equity is represented by the S&P 500 Index and the U.S. Core Fixed Income asset class is represented by multiple bond indexes prior to 1975; post 1975 the asset class is represented by the Barclays Capital Aggregate Index. The cost of borrowing is a combination of the 3-month LIBOR return (from 12/1986 to 12/2009) and the return on U.S. Treasury Bills plus an average spread to LIBOR (before 12/1986). The 0.25% financing rate that we have been using is then added to that return stream to fully reflect the cost of the leverage.

Exhibit 7 Risk Reduction / Return Enhancement through Leverage Expected Performance Historical Performance 'Typical' Risk Return 'Typical' Risk Return Asset Class 60/40 Reduction Enhancmt. 60/40 Reduction Enhancmt.

U.S. Equity 60% 44% 47% 60% 44% 47% U.S. Core Fxd Income 40% 100% 107% 40% 100% 107%

Borrowed Cash 0% -44% -54% 0% -44% -54%

Return 6.74% 6.74% 6.99% 8.58% 8.45% 8.70% Risk 10.36% 9.68% 10.36% 12.10% 10.68% 11.43% Sharpe Ratio 0.43 0.46 0.46 0.39 0.43 0.42

Arithmetic: Risk Reduction -0.67% -1.42% Return Enhancement 0.25% 0.12%

The same asset mixes plus leverage that are expected to improve efficiency based on the asset class assumptions were used for historical observations, as well, to maintain a consistent asset allocation. The results between “Expected” and “Historical” are different, not surprisingly, and provide for some interesting observations. While the ‘Risk Reduction’ portfolio is expected to lower risk by 0.67% going forward, that same allocation would have improved risk by a much larger 1.42% historically. However, the historical return for the ‘Risk Reduction’ portfolio falls slightly from the "Typical" mix rather than holding constant, which is the case based on expectations.

With respect to the ‘Return Enhancement’ portfolio, the historical results fall short of what is expected moving forward. Based on the assumptions utilized, a return improvement of 0.25% is expected at the same risk level of the "Typical" portfolio. The historical return of the ‘Return Enhancement’ portfolio, though, only increased by 0.12%

Copyright  2010, Wilshire Associates Incorporated Page 8 Wilshire Consulting Risk-Focused Diversification February 11, 2010 above the "Typical" asset mix. However, the efficiency improvement, evidenced by the Sharpe ratio, is meaningful as the historical risk was lower for this portfolio. The dampened return enhancement is due to a combination of factors that made leverage “more expensive” historically. Exhibit 8 shows that the historical spreads between core fixed income and the borrowing rate and between equity and core fixed income are different than the expectations used, therefore decreasing the effectiveness of a leveraged portfolio strategy.

Exhibit 8 Historical Spreads versus Current Expectations

5.00%

4.50% 4.14%

d 4.00% 3.75% a e r 3.50% p S

n

r 3.00% u t

e 2.50% R

c i

t 2.00% e 1.50% m 1.50% 1.33% h t i r

A 1.00%

0.50%

0.00% CoreFxdInc-Borrowing Equity-CoreFxdInc

Expected Historical

The historical spread between core fixed income and the borrowing rate is smaller than the spread built into the above expectations. As the portfolios in Exhibit 7 involve financing a heavier allocation to fixed income, the smaller historical spread implies a decrease in the return from borrowing cash and investing in bonds, from a historical perspective. Additionally, the historical spread between equity and core fixed income is larger than that implied by the expectations, leading to an accelerated drop in portfolio return as money is moved out of equity, again from a historical perspective. As we noted in the “Re-Leveraging the Capital Markets” section above, return expectations can change from one year to the next. As a result, the implied spreads between asset classes can also change, leading to either favorable or unfavorable environments for leverage. The ‘Return Enhancement’ portfolio from Exhibit 7 was used to identify historical time periods where a leveraged strategy would and would not have worked. Exhibit 9 contains rolling performance observations for that portfolio versus the “Typical” 60/40 portfolio and shows both periods of positive and negative return and/or risk effects.

Copyright  2010, Wilshire Associates Incorporated Page 9 Wilshire Consulting Risk-Focused Diversification February 11, 2010

Exhibit 9 Rolling 3-Year Arithmetic Excess: Leveraged versus Unleveraged Portfolios

10.00%

8.00%

6.00%

4.00%

2.00%

0.00%

-2.00%

-4.00%

-6.00%

-8.00%

-10.00%

Excess Return Excess Risk

From a return perspective, the orange line in Exhibit 9 shows that a leveraged portfolio would have experienced both positive and negative periods of relative performance versus the “Typical” portfolio and that deviations are usually substantial – well over 1%. A closer look at the underlying data in Exhibit 9 reveals several scenarios where a leveraged portfolio is likely to trail the “Typical” allocation. The most straightforward market environment is an equity bull market, but not necessarily any up market. Monthly returns of the two portfolios in Exhibit 9 show that when the equity market was positive for the month and the leveraged portfolio outperformed, equities were up an average of 3.2%. During months when the leveraged portfolio trailed the “Typical” allocation and the equity market was positive, equities were up an average of 4.4% for the month. Therefore, and rather intuitively, the higher the equity return, the harder it was for the leveraged portfolio to keep pace.

It is important for investors to know that when the reverse of the above market scenario is true – when core fixed income is outperforming equities – it is still possible for these leveraged portfolios, heavily weighted towards fixed income, to trail. The year 1981 provides an excellent example of how a leveraged strategy may not perform as hoped. The conditions during the year are charted in Exhibit 10 along with what the Return Enhancement portfolio would have produced. The monthly returns are displayed for the S&P 500 Index, Barclays Aggregate Bond Index and the arithmetic difference between the leveraged and unleveraged portfolios. The market yields also are shown for the aggregate index and T-Bills.

Copyright  2010, Wilshire Associates Incorporated Page 10 Wilshire Consulting Risk-Focused Diversification February 11, 2010

Exhibit 10 Experience of a Leveraged Strategy 1981

10.00% 18.00%

8.00% 16.00%

6.00% 14.00% n

r 4.00% 12.00% d u l t e i e Y

R 2.00% 10.00%

t y l e k h 0.00% 8.00% r t a n o -2.00% 6.00% M M -4.00% 4.00%

-6.00% 2.00%

-8.00% 0.00%

Lev-UnL Return Equity Core Bonds BCAgg. Yield T-BillsYield

During 1981, there were straightforward occurrences of the leveraged strategy underperforming when the core fixed income market trailed the equity market, as in April and December. Those two months along with August also were months when both markets were down by large amounts and an investor would have been exposed to additional negative beta – while paying for the leverage financing. The major theme during the year, however, was the difference in yield between core fixed income and Treasury bills. There is typically a spread between the two but that either did not exist during the year or was minimal – especially considering the fact that the total leverage cost is higher than the Treasury yield. This type of interest rate environment represents a higher hurdle rate for the returns of leveraged fixed income instruments.

Another way to consider the environment is that the leverage was too expensive, i.e. the financing rate overwhelmed the return benefits of the leverage strategy. January and September provide examples of such occurrences; in both cases, core fixed income significantly outperformed equity, which posted large negative results, seemingly providing an edge for the risk-balanced leveraged strategy to outperform the equity- heavy 60/40 portfolio. However, since the return to core fixed income in each month fell short of the cash financing rate, both months actually produced a negative excess return for the leveraged portfolio versus the 60/40 case. In fact, it is worth noting that core fixed income outperformed the equity market during 1981 by over 1,000 basis points yet the leveraged strategy, with a relative overweight to bonds and underweight to stocks, would have trailed a “Typical” portfolio by a 400 basis point arithmetic difference.

One last aspect of performance concerning a leveraged strategy is limiting the downside when equity turns negative. Downside protection is an attractive aspect for investors heavily affected by risk. Exhibit 11 shows the negative performing periods for the two portfolios in Exhibit 9 and their subsequent recoveries. The exhibit is a “growth of a

Copyright  2010, Wilshire Associates Incorporated Page 11 Wilshire Consulting Risk-Focused Diversification February 11, 2010 dollar” chart that only follows downturns in the market so that each valley shows the cumulative sell-off and then recovery from the previous peak.

Exhibit 11 Downside Recovery

$1.00

$0.90

$0.80

$0.70

$0.60

$0.50

$0.40

$0.30

Unleveraged Leveraged

During most downturns, the leveraged portfolio (in green) is down less and recovers more quickly. However, there are occasions where the reverse is true; when both equity and core fixed income decline, equity far outperforms bonds or leverage gets very expensive. The most recent meaningful example is 1994 (circled in red) when the unleveraged portfolio returned -0.33% while the leveraged return was -5.01%. During the year, stocks outperformed bonds and the yield on cash (the cost of borrowing) spiked.

Framework for Considering the Degree of Leverage

The amount of leverage utilized throughout this report has simply been a result of meeting the risk and return profile of the "Typical" portfolio. One way for an investor to answer the question “How much leverage is too much?” is to consider potential cash flow issues in managing their exposure. Like with many other investment scenarios, limiting or removing risk often involves replacing it with another type of risk. Cash flow and liquidity are major risks to address and manage while employing leverage at the asset class level. Some form of derivative (i.e. future, swap) is necessary to achieve an exposure greater than the physical dollar outlay to invest and these contracts require regular settlement. For example, assume an investor buys a futures contract for $100. No cash changes hands to establish the investment as this is essentially an agreement to purchase a security (or basket of securities) for $100 at some time in the future, often 90 days. Instead of actually trading the security, the futures contract is typically settled for the difference between $100 and the price of the security at settlement. If the security’s price increases, the investor receives the difference, representing an investment gain. If

Copyright  2010, Wilshire Associates Incorporated Page 12 Wilshire Consulting Risk-Focused Diversification February 11, 2010 the price decreases, the investor must settle the contract by paying the difference. In practice, derivative cash flow settlements are cleared through margin and collateral accounts that often include daily marked-to-market settlement terms. During periods of extreme market volatility, such as the one experienced during September 2008 in the wake of the Lehman Brothers collapse, the required derivative cash flows can become quite disruptive to an institution's liquidity operations. The money for this payment must come from another part of an investor’s portfolio whether it be a cash account or through liquidating hard assets.

This derivative maintenance issue must play a role in an investor’s decision of how much leverage to utilize. There is no one answer, however Wilshire believes that a framework for considering cash needs will help investors make the right decision for their specific liquidity needs. Exhibit 12 outlines the process and presents various scenarios to consider. In the exhibit, portfolio leverage of 20% is considered on a total fund of $250 million with different levels of risk underlying the derivative exposures to asset classes with annual risk of 5% and 17%.

Exhibit 12 Potential Outflow Scenarios Total Portfolio 250$ million Leverage 20%

EventinSD 1 2 3 Event Probability 15.9% 2.3% 0.1%

Outflow at Different Risk Levels (mil) 5% 1.25$ 2.50$ 3.75$ 17% 4.25$ 8.50$ 12.75$

Stepping through one example, assume the leveraged asset class has an expected annual risk of 17% and that a negative two-standard deviation event occurs during the quarter. The negative event is used to imply a relatively substantial loss and the resulting settlement amount is $8.5 million. The investor would subsequently need to either withdraw the amount from cash reserves, if available, or sell an equal amount of hard assets to settle the futures contract. While the probability of such an event is relatively low at 2.3% it is important to realize that, over a long enough period, two and even three- standard deviation events are certain to occur. When they do, an investor needs to be comfortable with the prospect of having or creating the necessary liquidity.

An important nuance is embedded in the above discussion. Settling a futures contract with a cash outflow does not necessarily equate to a relative loss versus a traditional approach, regardless of the settlement amount. In fact, even if the investor loses that quarter’s equity “bet” and loses money in their bond portfolio, the broad, alternative asset allocation must be considered to measure the effectiveness of the leveraged strategy. In

Copyright  2010, Wilshire Associates Incorporated Page 13 Wilshire Consulting Risk-Focused Diversification February 11, 2010 other words, had the investor not utilized leverage, what would have been their asset allocation and resulting return?

Which Risk Level/Asset Class Would Employ Derivatives

It is worth noting that the asset class that is to be increased relative to a traditional portfolio is not necessarily where an investor must have derivative exposure. In other words, an investor that wishes to increase their market exposure to core fixed income could utilize equity market derivatives for a portion of their equity exposure and then use the available cash to purchase bonds. In determining what derivatives to use, there are two major factors to consider; which are the least expensive options and what is the expected resulting risk of the leveraged asset classes that will drive cash flow volatility. Throughout this analysis, we have been assuming a financing rate of 0.25% above LIBOR. In practice, the cost could be higher or lower, clearly affecting the results of the overall leverage program.9 If the cash requirement is a major issue to an investor, for example if a substantial amount of illiquid investments such as Venture Capital or Infrastructure are held in the overall portfolio, the lower risk asset classes may be more appealing. The lower level of expected risk imposes less potential disruption on an investor's liquidity position.

Leverage within a Surplus Framework

A surplus framework can be defined as analyzing the risk and return of assets in excess of an accounting liability. It is worth considering how a leverage strategy might affect asset allocation considerations as the typical liability of a pension plan is highly correlated with the lower risk, fixed income asset classes that are often favored when utilizing leverage. First, a proxy is needed for the “return” in the accounting liability that is then included as a negative asset within a portfolio optimization process. We will utilize the Barclays Capital Long-Term Government/Credit Index to represent the liability return, along with the corresponding risk and return expectations. Exhibit 13 displays the expected efficient frontiers, or surplus frontiers, that result from the optimization process. The expected return of the accounting liability is assumed to be 5.25% with an expected risk of 10.00%. The assumed correlation to U.S. Equity, U.S. Core Fixed Income and cash/borrowing is 0.31, 0.94 and 0.10, respectively. The asset class assumptions utilized in Exhibit 13 are Wilshire’s current 2010 forecasts and demonstrate that, contrary to the unflattering results we saw earlier for a simple equity/fixed income scenario in an asset- only framework, the potential benefits of utilizing leverage within an asset-liability construct are quite dramatic.

9 Relatively costly derivatives include those for fixed income credit while Treasuries, U.S. Equity and Non- U.S. Equity are typically less expensive.

Copyright  2010, Wilshire Associates Incorporated Page 14 Wilshire Consulting Risk-Focused Diversification February 11, 2010

Exhibit 13 Surplus Frontier: Assets in Excess of an Accounting Liability

10.00

8.00

6.00

n 4.00 r u t e

R 2.00

0.00

-2.00

-4.00 0.00 2.00 4.00 6.00 8.00 10.00 12.00 14.00 16.00 18.00 Risk Unleveraged EF Leveraged EF max 100% Leveraged EF max 50% 60/40 Stock/Bonds

In a surplus framework, a negative return signifies a shrinking surplus (or growing deficit) where a positive return is expected to increase the surplus. It is evident from the distance between the efficient frontiers that expected risk reduction/return enhancement is more dramatic across surplus frontiers. The main cause of these results is the high correlation between the accounting liability proxy and core fixed income. While the correlation is a positive 0.94, the liability is a negative asset so that as the liability grows (the “return” on the liability goes up), the return to core fixed income is also very likely to increase. As we have discussed in our previous research describing Wilshire's Commitment-Driven Investing approach, there are significant issues with conducting an asset allocation study within a surplus framework however these results further suggest that there is strong theoretical support for utilizing leverage within asset allocation. The case study provided in the Appendix more fully develops the surplus frontier topic, adding other asset classes and including operational issues.

Final Issues & Considerations

There are a number of operational issues that an institutional investor must address before implementing a leveraged asset allocation strategy. Understanding a program’s results would involve attributing relative performance to active management, identifying any tactical asset allocation decisions and assessing mechanical factors such as leverage costs. Investors attempting such a strategy in-house may need to dedicate resources specifically to maintaining the program, i.e. “rolling” the derivatives forward. For most institutions, implementation of a leveraged strategy would likely require the retention of a beta overlay manager to execute and maintain the desired leveraged systematic exposures or an allocation of capital to one or more off-the-shelf investment products which employ embedded leverage to achieve asset class risk balance. Each implementation approach

Copyright  2010, Wilshire Associates Incorporated Page 15 Wilshire Consulting Risk-Focused Diversification February 11, 2010 has advantages and disadvantages versus the alternatives, including considerations of incremental management costs. Wilshire Consulting can assist in discussing the potential trade-offs between implementation types. Also, should these strategies realize greater acceptance in the marketplace, capacity limits are an issue that the entire market could eventually face (i.e. what are the total dollars that the market could support?).

This paper by Wilshire Consulting detailed how leverage should theoretically improve both risk (as a decrease) and return (as an increase) as assets are diversified away from equity-heavy portfolios, which also improves the distribution of risk sources. Historical data show that a leveraged portfolio’s ability to enhance return has been somewhat limited – in a simplistic portfolio – although leverage would have decreased risk rather substantially. It is very important for investors to recognize that total leverage costs are variable and can have a dramatic, negative effect on a leveraged strategy during certain periods of market stress. Great consideration and understanding must be applied before an investor chooses such a strategy so that they have the appropriate expectations as their portfolio navigates different market conditions.

Copyright  2010, Wilshire Associates Incorporated Page 16 Wilshire Consulting Risk-Focused Diversification February 11, 2010

Appendix: Case Study

Adopting Risk-Focused Diversification, and permitting the use of leverage within the asset allocation process, represents a meaningful departure from a traditional approach to investing. The basic issues and considerations were introduced in this paper’s full text. A brief but complex scenario is introduced here to illustrate what implementation in a live portfolio might involve. The basic framework introduced in the paper’s ‘Surplus’ section is more fully developed below to include the following:

 Variable financing costs for equity and fixed income derivatives  Liquidity constraints, including required margin, based on the expected volatility of different asset classes  Alternative asset classes such as Real Assets and Private Markets  Liability assumptions based on very long term, corporate bond expectations  Underfunded pension status, 90% funded ratio

This example includes two sources of leverage – equity and fixed income derivatives, but can easily be expanded to include other sources of synthetic market exposure. The cost of equity financing is lower because of the depth and breadth of the market versus fixed income. However, we impose a larger liquidity constraint on equity financing to account for the greater volatility of stocks relative to bonds. As such, a higher proportion of total funds are assumed to be held in cash to comply with margin requirements. The basic assumptions are as follows:

Equity Fixed Income Financing Financing Total Cost 3.50% 3.75% Volatility 2.25% 2.00% Buffer held in Cash 20% 10%

The following additional constraints were introduced to limit the scope of the analysis.

 Less liquid assets – Real Assets and Private Markets – were limited to 20% of the overall portfolio in total.  Only equity and fixed income derivatives can be utilized (i.e. we did not allow synthetic exposure to any of the other asset classes).  Total leverage cannot exceed 50% of assets.

The results are charted below and include three efficient frontiers. Frontier 1 (EF 1) contains no leverage. Frontier 2 (EF 2) allows leverage but does not impose the liquidity buffer constraints. Finally, Frontier 3 (EF 3) reflects the entire scope of the case study.

Copyright  2010, Wilshire Associates Incorporated Page 17 Wilshire Consulting Risk-Focused Diversification February 11, 2010

The surplus optimization results are dramatic and even portfolios with single-digit expected risk are forecasted to produce a positive surplus return – indicating that the portfolio would improve upon the plan’s underfunded status.

The resulting portfolio compositions highlight the potential benefits of utilizing risk diversification leverage. We make several observations when comparing the three frontiers.

 The leveraged frontiers (EF 2 & EF 3) show dramatic efficiency improvements over the unleveraged frontier (EF 1). The leveraged frontiers are able to achieve positive expected surplus returns as they approach the 8% risk level, while the unleveraged frontier does not reach a positive surplus return until it approaches 12% surplus risk.

 Leveraged portfolios without a liquidity buffer constraint (EF 2) utilize the equity markets as much as possible to create leverage – due to the lower assumed cost. For example, at the 11% risk level on frontier EF 2, where leverage is utilized at its maximum 50%, roughly two-thirds of leverage is deployed through equity derivatives.

 As can be seen from EF 3’s portfolio mixes below, when introducing a liquidity buffer, fixed income derivatives are favored – with very little synthetic equity securities utilized – due to the lower, 10% buffer. In fact, at approximately the 11% risk level, where nearly two-thirds of EF 2’s leverage was achieved via

Copyright  2010, Wilshire Associates Incorporated Page 18 Wilshire Consulting Risk-Focused Diversification February 11, 2010

equity derivatives, EF 3 accesses its entire leveraged position through synthetic fixed income exposure.

 The distance between frontiers EF 2 and EF 3 provides interesting insight into the cost of adhering to the liquidity constraint at various risk levels (i.e. the cost of maintaining reasonable liquidity).

Efficient Frontier 3 Sample Portfolios AssetClass Portfolio1 Portfolio2 Portfolio3 Portfolio4 Portfolio5

Global Stock 1% 8% 15% 34% 64% LT Core Bond 119% 120% 110% 91% 61% Real Assets 0% 15% 7% 0% 0% Private Markets 0% 1% 13% 20% 20% Cash 2% 6% 6% 5% 5%

Equity Financing 0% -8% -5% 0% 0% Bond Financing -22% -42% -45% -50% -50%

Surplus Return -1.47% -0.51% 0.33% 1.03% 1.48% Surplus Risk 5.83% 6.33% 8.29% 11.40% 15.57%

In summary, leverage within the asset allocation process is an encouraging concept in theory and, as real world constraints are added, the strategy continues to show promise. The case study presented above demonstrates that the required tools and modeling capabilities exist to accommodate practical considerations such as variable derivative costs and liquidity thresholds. As such, leverage can be utilized to maintain the expected return on a portfolio while reducing the portfolios expected risk and dependence on high- volatility asset classes.

Copyright  2010, Wilshire Associates Incorporated Page 19 Wilshire Consulting Risk-Focused Diversification February 11, 2010

Important Information

This material contains confidential and proprietary information of Wilshire Consulting, and is intended for the exclusive use of the person to whom it is provided. It may not be modified, sold or otherwise provided, in whole or in part, to any other person or entity without prior written permission from Wilshire Consulting.

The information contained herein has been obtained from sources believed to be reliable. Wilshire Consulting gives no representations or warranties as to the accuracy of such information, and accepts no responsibility or liability (including for indirect, consequential or incidental damages) for any error, omission or inaccuracy in such information and for results obtained from its use. Information and opinions are as of the date indicated and are subject to change without notice.

This material is intended for informational purposes only and should not be construed as legal, accounting, tax, investment, or other professional advice.

Statements concerning financial market trends are based on current market conditions, which will fluctuate. There is no guarantee that these suggestions will work under all market conditions, and each investor should evaluate their ability to invest for the long- term, especially during periods of downturn in the market.

Copyright  2010, Wilshire Associates Incorporated Page 20 Wilshire Consulting Benchmarking Private Investments: Assessing Performance despite Index Imprecision Thursday, March 4, 2010

Wilshire Consulting Alexander C. Browning, Senior Associate

Benchmarking Private Investments: Assessing Performance despite Index Imprecision

Benchmarking a portfolio of private market investments is not a clear cut task. As a result private market investment returns frequently display considerable deviation from their chosen or created benchmarks. The origin of the measurement problem begins with pricing but quickly broadens to the availability of data and portfolio composition. Given these imperfections, institutional investors have long created solutions, or heuristics, which provide acceptable beta/performance measurements over long time periods but fail in shorter time periods. To the justification of this method, there are no simple out-of-the- box solutions but, there are a number of ways existing benchmarks can be used and modified fruitfully to provide different parts of the full picture. In the following research note, Wilshire Consulting examines several common benchmark mismatches and methods of piecing together the available information to create an accurate picture of investment performance.

Benchmark Rules of the Road The CFA Institute has articulated a set of optimal benchmark construction guidelines. According to the CFA Institute, a benchmark should be:

1. Unambiguous: the names and weights of holdings within the benchmark must be known 2. Investable: the option is available to forego active management and simply invest in the benchmark portfolio 3. Measurable: the benchmark’s return can be calculated on a reasonably frequent basis 4. Appropriate: the benchmark represents the manager’s style and biases 5. Reflective of the manager’s current investment options: the manager must have up-to-date knowledge of the holdings that comprise the benchmark 6. Specified in advance: the benchmark must be constructed prior to the start of the evaluation period.

While private market investment benchmarks may technically abide by some of the CFA benchmark construction guidelines, in application, they frequently violate most if not all of them. In general, the following issues are commonplace in private investment benchmark indices.

1. Frequently the names and weights of index investments are held confidential

Copyright  2010, Wilshire Associates Incorporated Page 1 Wilshire Consulting Benchmarking Private Investments: Assessing Performance despite Index Imprecision Thursday, March 4, 2010

2. The number of underlying investments and illiquidity practically, if not completely, prohibit the re-creation of an investible benchmark. 3. Performance measurement is complicated by the variety of performance calculations presented, i.e. IRR, TVPI, DPI, cash on cash return multiple, etc. 4. Internal valuation processes are unique rather than uniform. 5. Benchmarks may aggregate multiple investment styles, vintage years, and suffer from survivorship bias.

Despite these short-comings, existing benchmarks may still provide a useful representation of the opportunity set and beta of the asset class if chosen correctly and adjusted appropriately.

When, selecting a benchmark, the most important issue to consider is whether the benchmark appropriately matches a manager’s, or pool of managers’, chosen investment opportunity set. Wilshire generally advocates the use of the broadest market benchmarks available since market segments that are omitted from a broad market index can introduce an opportunity cost to managers benchmarked against them. While some investors may choose to employ benchmarks for specific reasons other than to capture the broad market, it is important to select a benchmark that reflects the overall structure of the portfolio, is free from unnecessary biases, and is widely understood by both the investors and investment managers.

The Mainstream Approach: Index-Plus A common solution to the private investment benchmark problem is taking a public market index and adding a premium to its return. Private equity is a prime example where this solution is common. An investor might take the Wilshire 5000 Total Market IndexSM and add a 3% annual return premium as their private equity benchmark. The addition of the 3% premium is a short-cut to accounting for the return premia associated with illiquidity, leverage, and business risk. The advantages of this method are that the returns are from a continuously priced market, represent equity risk, are easy to modify, and can represent a reasonable proxy over long time periods. The large flaw to this methodology is its application as a performance benchmark in shorter time periods and for its inability to separately account for leverage factor differences which in part contribute to its long- run usefulness.

Leverage Mis-Match While the rule of thumb index-plus methodology may exhibit some long-term accuracy, a major problem is improperly capturing the penalty or reward that leverage generates in short-term performance. Though the use of leverage is neither good nor bad but, a reflection of a calculated risk/return decision, its use can reflect an important potential point of departure from a benchmark. When private market investments are performing well and asset values are growing, a portfolio with positive net leverage versus the benchmark is rewarded for employing leverage. Conversely, when markets are moving

Copyright  2010, Wilshire Associates Incorporated Page 2 Wilshire Consulting Benchmarking Private Investments: Assessing Performance despite Index Imprecision Thursday, March 4, 2010 down, the leverage accelerates the decline in the value of assets and the portfolio’s performance is penalized.

Because of the additive nature of the index-plus benchmarking approach, this increase in volatility tends to go somewhat unremarked in up markets but stands out as manager under-performance in down markets. Assuming the level of systematic leverage is consciously made, the full attribution of a return shortfall versus an unlevered index-plus benchmark to manager under-performance is incorrect10. For example, in a year where the Wilshire 5000 declines by -15%, the index-plus method would indicate a return of - 12%; but this violates the logic that leverage on the downside increases portfolio losses. Thus while the index-plus rule of thumb works well over long time periods, its application does not recognize leverage’s contribution to return and as a result renders it especially imprecise in down markets and shorter time periods.

Valuation Lag and Smoothing An important valuation difference between private and public markets lies in the time it takes to compile and translate financial statement information into valuations. In general, the process can take three months or more following the period which generated the performance. The time lag this creates not only produces an issue for comparison to public market investments in the greater portfolio but comparison to public market indexes used in creating performance benchmarks, adding further imprecision to the index-plus approach.

Private market investments, unlike public market investments, are also typically appraised internally on a quarterly/annual basis and done so by formula. Appraisers usually price on comparable metrics which are backward looking. As a result, appraisal based valuations and the indexes that are built from them can suffer from significant transaction and volume biases.

For example, one large transaction at a variance from its previously appraised value can materially impact appraisals of other investments and index returns overall. This significant transaction bias can also mask declines in other relevant metrics such as operating income. In the case of transaction volume bias, which is especially relevant in times where pricing is volatile and the number of transactions is few, greater reliance must be placed on past valuations that reflect stale pricing. For example, recently many large private market deals were cancelled by the prospective sellers; only the ones with a satisfactory price to the sellers were completed. Therefore, appraisers were forced to use a biased sample (i.e., the prices may have been biased up) relative to what might have been observed in a continuous marketplace. In other words, a few transactions with biased pricing were being used to mark-to-market a large amount of unmarketable assets. The overall effect of these pricing biases mutes a lot of the valuation volatility that naturally occurs in comparably priced public market investments. While it can be argued

10 Knowledge of the underlying components of both the portfolio and the index is crucial. When evaluating leverage, one must be cognizant of both the leverage employed in the capital structure of an asset as well as at the portfolio level.

Copyright  2010, Wilshire Associates Incorporated Page 3 Wilshire Consulting Benchmarking Private Investments: Assessing Performance despite Index Imprecision Thursday, March 4, 2010 that the additional price volatility of public equity markets reflects noise from the process of information assimilation, these valuations do reflect the market clearing price. As a result, appraisal-based valuations create smoother time series of asset values and generate return for measure of risk statistics that look superior to public markets.

Alternative Approaches: Which Beta to Use and Why? Finding a good representation of the market, requires considerable thought as to what investment universe is to be measured. Broad market indexes are valuable for representing the opportunity set in total but perhaps not for measuring the performance of more granular active structure decisions. The measurement of structure decisions typically results in the use of multiple benchmarks, one to measure the structural bias decision and another to measure the level of active management employed in the curtailed opportunity set. The ability to break down structural element decisions within a private markets portfolio is considerably more difficult than in public markets because of the underlying valuation and liquidity features of private market investments in general. Therefore the question becomes an exercise in identifying which features one can isolate and then apply to get the desired benchmark information.

We have highlighted that non-market valuation, financial reporting lags, leverage, and the match of underlying investments as central culprits to the benchmarking challenge. Below, some common benchmark creation and modification techniques are briefly detailed with a focus on their prominent strengths and weaknesses. In the following section, these techniques are evaluated in the context of an established private real estate portfolio and then followed by asset class specific recommendations in Appendix A.

Utilize a representative public market benchmark, apply lag, and adjust for leverage  Strengths: This approach takes advantage of the continuous pricing information available through a transaction-based public market proxy. The adjustment for lagged reporting schedules attempts to standardize timing differences, while the leverage adjustment seeks to normalize market exposures to the degree of financial risk being deployed. These strengths are particularly valuable over the short-run, as the combination of public market pricing with timing and leverage adjustments leads to a benchmark that is better able to capture short-term market trends and volatility. Over the long-term, the approach provides an opportunity cost metric against comparable public market investments.

 Weaknesses: This method’s reliance on a public market proxy's daily volatility in pricing may reflect distortions within public market investment trading. In the short-run, momentum of market direction may overstate returns on both the downside and upside. Over the long-run, the method has the potential to compound valuation differences between public and private market liquidity premiums. Business decisions made in the context of private market investments may differ from comparable public market investments as a result of a longer time horizon within which to realize investment gains.

Copyright  2010, Wilshire Associates Incorporated Page 4 Wilshire Consulting Benchmarking Private Investments: Assessing Performance despite Index Imprecision Thursday, March 4, 2010

Utilize an existing private market benchmark without adjustment  Strengths: This approach uses a comparable set of underlying investments and a standardized valuation schedule and thus attempts to give the “real” market value for the private market segment. The transaction prices in the market reflect valuations at which similar investors have completed transactions and thus do not suffer from the daily pricing volatility of a public market proxy.

 Weaknesses: The number of transactions, liquidity of the market, and similarity between investments are assumed to be “comparable.” The valuations based on these three assumptions are extremely sensitive to the “comparable” condition and as such any variance from that condition will result in biased appraisal estimates. Additionally, leverage differences will magnify up and downside performance in the short-run and over longer time periods may also magnify underlying structural differences between the benchmark and the portfolio. Thus biases for transaction volume, significant transactions, and structural differences are present.

Take Private Benchmark and re-lever  Strengths: This approach looks to normalize market exposures to the degree of financial risk being deployed between a private investment benchmark and a portfolio. The resulting benchmark thus reflects similar investments, investors, and helps to normalize the risk contribution of leverage.

 Weaknesses: After applying a leverage adjustment, the issues which emanate from appraisal based indexes still exist. Structural differences in sector exposure also are not corrected.

Case Study: Evaluating a Private Real Estate Program Recently, while examining a well established private real estate portfolio, a benchmark issue became evident; the multi-manager program was severely underperforming its benchmark. A more thorough examination of different time periods however revealed that the portfolio displayed both under and over performance relative to the benchmark. This situation generated a few fundamental questions.

 Is the chosen benchmark representative?  Is there a portfolio construction issue?  Is the manager research process flawed?

Before trying to answer the questions posed above, the existing portfolio had to be summarized in a way to make meaningful comparisons. The portfolio was mostly composed of direct private real estate funds diversified by property type and geography, and represented exposure to office, retail, apartment, industrial, and hotel. Depending on the strategy employed, core, value-add, and opportunistic, the levels of leverage varied and historical leverage figures and geographic exposures at the portfolio level were not

Copyright  2010, Wilshire Associates Incorporated Page 5 Wilshire Consulting Benchmarking Private Investments: Assessing Performance despite Index Imprecision Thursday, March 4, 2010 available. Sector data for the Private Real Estate Portfolio were available but only for the most recent seven years of quarterly data.

The chosen benchmark for the portfolio was the NCREIF Property Index (NPI). The NPI can best be summarized as a large total return composite index of individual private commercial real estate properties acquired for investment purposes largely on behalf of tax-exempt institutional investors and as such, held in a fiduciary environment. Returns to the index are reported quarterly and on a non-leveraged basis.

Given these characteristics, the first logical comparison was to measure the portfolio’s performance against the published NPI and a sector-adjusted NPI for which seven years of quarterly sector data was available11.

Exhibit 1: Private Real Estate Portfolio versus NPI Returns

20

10

0

-10

-20

-30 12/02 12/03 12/04 12/05 12/06 12/07 12/08

Sector Adjusted NCREIF NCREIF PrivateRE Portfolio

Source: Wilshire Compass

In general, without adjustment, the NPI appears to do a pretty good job of representing a line of central tendency for the Private RE Portfolio and helps to verify that the appropriate beta has been selected. With the sector adjustment, the improvement over the unadjusted NPI is hard to see and under further quantitative analysis of questionable improvement. Intuitively this result might follow from the similarity between the Private Real Estate Portfolio’s investment opportunity set and that which reflects the broader market opportunity set. The remaining difference between the Private Real Estate Portfolio and the NPI can thus be reasonably attributed to geography and use of leverage with the final residual being attributed to manager skill.

11 The NPI and sector-adjusted NPI were lagged one quarter as a result of rolling correlation tests showing a better fit for 1yr, 3yr, and 5yr periods. The sector adjusted NPI was created by re-weighting the NPI’s exposure to industrial, hotel, office, residential, and retail sectors to match the sector exposures in the private real estate portfolio on a quarterly basis.

Copyright  2010, Wilshire Associates Incorporated Page 6 Wilshire Consulting Benchmarking Private Investments: Assessing Performance despite Index Imprecision Thursday, March 4, 2010

A time series of geographic data for the portfolio, however, was not available, as multiple underlying managers and properties changed over the portfolio’s life. While real estate is quintessentially a local investment, given the diversified nature and the large size of the program, it is plausible that the overall geographic allocation, much like the sector allocation, might not deviate substantially from the overall market opportunity set represented by the NPI. In either case however, the information required to separately attribute return to geography and manager skill was not available and, therefore, is absent from this analysis.

With regard to leverage, the aggregate level of leverage was also not available. But, given the long history of the portfolio, a leverage proxy was estimated by calculating a three year rolling beta to the NPI and multiplying the level of implied borrowing by the historical yield on mortgage backed securities; for quarters where beta was less than one, no cost of leverage was charged to the portfolio12. Below in Exhibit 2, the returns to the NPI, the Private Real Estate Portfolio, and leverage adjusted NPI are displayed.

Exhibit 2: Private RE Portfolio versus NPI and Adjusted NPI Benchmarks

20 10 %

n 0 r u t

e -10 R

l a t -20 o T -30 -40 12/02 12/03 12/04 12/05 12/06 12/07 12/08 NCREIF Beta Adj.NCREIF Portfolio

Source: Wilshire Compass

Both the NPI and Leverage-adjusted NPI seem to capture the trend of the portfolio, once again affirming the general beta selection. Calculation of R-squared for the portfolio versus the leverage-adjusted benchmark yields 0.77, somewhat higher than the 0.72 R- squared for the portfolio against NCREIF 13 . However, when examining benchmark performance in volatile markets such as those experienced over the last five years, the Leverage-adjusted NPI appears to capture the upside and downside volatility more

12 For mature real estate programs, where the underlying allocations to core, value-add and opportunistic real estate are well established, a static beta may provide a simple yet effective alternative approach to the rolling beta approach. 13 R-squared is generally interpreted as a measure of the explanatory power of a data set over another; that is, the change of the leverage-adjusted NCREIF benchmark returns explains 77% of the change in the portfolio returns. The extent to which R-squared differs from 1.0 reflects the effect of active management and/or factors not captured in the benchmark.

Copyright  2010, Wilshire Associates Incorporated Page 7 Wilshire Consulting Benchmarking Private Investments: Assessing Performance despite Index Imprecision Thursday, March 4, 2010 accurately than the unadjusted NPI. Below in Exhibit 3, the standard deviation and returns for the entire portfolio history are displayed for the three measures.

Exhibit 3: Returns and Standard June 30, 1981 to June 30, 2009 Risk Total Return Private RE Portfolio 9.92 7.87 NPI Leverage Adj. 10.98 5.25 NPI 4.39 8.07

Over the portfolio history, the NPI appears to provide a good estimate of expected return to the portfolio but underestimates the risk. Alternatively, the NPI Leverage Adj. series provides a reasonable measure of long-term risk but underestimates the expected long- term return, suggesting that manager implementation skill may have played a role in outperforming the benchmark. Thus, while neither benchmark provides an entirely accurate picture of performance absent variance contributed by manager skill, each benchmark contributes separate pieces of valuable information to total performance and risk evaluation.

Conclusion

No one size fits all; trying to capture market exposure while measuring the performance of managers and active structure and leverage bets is not a simple task without overcomplicating the process of having a well-understood and easy to calculate benchmark. The exercise of creating custom benchmarks can yield diminishing returns when their purpose becomes a data mining exercise; only thoughtful consideration of measurement versus tailoring will yield a meaningful benchmark.

Above, three general methodologies are outlined with associated short and long-term strengths and weaknesses. The case study that followed examined a private real estate portfolio and showed how, in practice, various elements of the outlined approaches can be applied to better understand long- and short-term performance. Thus, while the methodologies discussed can serve as effective benchmarking strategies, none of them, much like benchmarks themselves, can be used as off-the-shelf solutions. Use of any benchmark creation strategy requires thorough understanding of the underlying portfolio investments and the attributes which can and cannot be measured.

During the last year when benchmarks of the index-plus flavor indicated vast underperformance for private investment managers, leverage adjusted benchmarks did a superior job of capturing the downside volatility and, subsequently, assessing manager results. Over long periods of time however, the index-plus methodology serves as an adequate return expectation. For performance measurement purposes, the case study appears to point in the direction of separate benchmarks for return and risk. With some

Copyright  2010, Wilshire Associates Incorporated Page 8 Wilshire Consulting Benchmarking Private Investments: Assessing Performance despite Index Imprecision Thursday, March 4, 2010 attention to convenience however, at the very least the case study highlights the need to broaden the scope of benchmark related discussions when making ultimate decisions regarding asset class exposure and manager performance.

Copyright  2010, Wilshire Associates Incorporated Page 9 Wilshire Consulting Benchmarking Private Investments: Assessing Performance despite Index Imprecision Thursday, March 4, 2010

Appendix A: Asset Class Specific Recommendations

Asset Class – Real Estate

The following recommendations are applicable at the real estate composite level and are intended for plan sponsors with a dedicated private real estate portfolio.

1. Create a Real Estate Composite

2. Show the Real Estate composite excluding what could be considered non-private real estate (i.e. publicly traded real estate securities, commercial mortgage-backed securities, private debt investments). This is consistent with GIPS standards.

3. For short-term analysis employ a sector and leverage adjustment to the NCREIF Property Index (NPI14). For the sector adjustment re-weight the sector component returns of the benchmark to the portfolio weights. For the leverage adjustment calculate as: (1+leverage factor)*NPI return – (periodic cost of leverage*leverage factor)

4. For medium term analysis use a sector adjusted NPI

5. For longer term analysis use sector adjusted NPI plus a risk premium to reflect addition of value to active management and the long-term compounding of leverage.

6. For comparable public market pricing, utilize a cash flow weighted public market investable proxy (e.g. Dow Jones Global Real Estate Securities Index) and normalize for the amount of leverage employed in the public market proxy versus the portfolio.

D/E = Debt to Equity of Market Proxy

d/e = Leverage in Private Real Estate Portfolio

Net Leverage (NL) = (D/E – d/e)

(CD15) = Cost of Debt/periodicity

Leverage Adjusted Return = (1+NL) x Market Proxy – NL x CD

14 The property returns within the index are reported on an unlevered basis. 15 Cost of debt is best represented by a yield series and not a return series.

Copyright  2010, Wilshire Associates Incorporated Page 10 Wilshire Consulting Benchmarking Private Investments: Assessing Performance despite Index Imprecision Thursday, March 4, 2010

Discussion:

Many real estate portfolios consist of a combination of financial (e.g. CMBS, mezzanine, etc) and tangible private real estate investments (e.g. direct, closed and open funds.) The Global Investment Performance Standards (“GIPS”) explicitly states that the former financial strategies are not considered Real Estate and Wilshire concurs. As such, financial strategies should be shown separately from the Real Estate Composite (e.g. real estate ex-alternatives composite).

Wilshire believes the NPI or the NPI plus a premium, for longer time periods, is the most representative for portfolios composed of a mix of closed-end funds or a mix of core, value added, and opportunistic strategies. For core private real estate in closed and open- end funds however, the NCREIF Open End Diversified Core Index (ODCE) that tracks 26 open-end core real estate funds can serve as a peer universe benchmark. While the use of the NCREIF ODCE index can provide a meaningful representation of the core opportunity set, it is not representative of a diversified strategy portfolio.

The use of cash-flow weighted or leveraged adjusted public market proxies can serve as a useful secondary benchmark and can illustrate the public market opportunity cost as well as serve as a liquid market indicator. This comparison can be useful for evaluating the longer term investment decision.

Asset Class – Private Equity

Wilshire believes the following benchmarking practices for private equity investments are appropriate and consistent with industry “best practices”.

Recommendations:

1. Create a private equity composite which can be sorted by vintage year, sector, and fund size.

2. Utilize private investment benchmarks, for example Thomson Venture Economics, to evaluate the portfolio at a composite, vintage year, sector, and fund size level against reporting peer universe16.

3. Utilize a cash flow weighted public market investable proxy (e.g. Wilshire 5000 Index) and adjust for leverage17 for opportunity cost versus leverage adjusted public market exposure.

16 VentureXpert and Cambridge also offer “time weighted” versions of their private equity benchmarks, which are useful at the total fund level. The weaknesses are the same as for the vintage IRRs as well as the fact that the index calculation for a given quarter is constantly changing as new data is added. Therefore, there is no “official” quarterly return. The user must pick their own cut-off date and thus reconciliation of long term numbers becomes problematic. 17 The levering method described above in Private Real Estate section can serve as basic methodology however, given the more complex capital structure of buy-outs, a more rigorous method can be adopted from a forecasting

Copyright  2010, Wilshire Associates Incorporated Page 11 Wilshire Consulting Benchmarking Private Investments: Assessing Performance despite Index Imprecision Thursday, March 4, 2010

Discussion:

Utilizing multiple benchmarks in the analysis of a private equity portfolio allows for comparison at the composite, sector, and vintage year levels. Common benchmarks utilized for private equity portfolios include:

1. Public market proxies with or without a return premium are utilized as a longer term benchmark as it reflects the public equity alternative. They also are convenient in the early stages of private equity programs if unallocated capital is allocated to public equity.

2. Cash-flow weighted Wilshire 5000 Index: The Investable Wilshire 5000 Index is constructed by applying the cash flow streams (i.e., contributions, distributions, fees, and expenses) applicable to a private equity investment and calculating an IRR based upon a hypothetical terminal cashflow value created by compounding those investments by the return of the Wilshire 5000 Index. This comparison is useful for evaluating the longer term investment decision.

3. Thomson VentureXpert: The index tracks over 2,300 funds globally and data is compiled by vintage year and fund type. Thomson Venture Economics uses cashflow schedules and financial reports to calculate internal rates of return and realization ratios covering venture, buyouts, funds, private equity, firms, executives, portfolio companies and limited partners.

4. Cambridge Associates: This index tracks 800 U.S. private equity funds and 1,287 U.S. venture capital funds from the years 1981 through the present by vintage year and fund type. Cambridge Associates includes this data to create a pooled sample of IRR, Distribution to Paid-In Capital (DPI), Residual Value to Paid-In Capital (RVPI), and Total Value to Paid-In Capital (TVPI) values as of September 30, 2009.

methodology described in: Wilshire Associates, Inc. (1997). Private Equity Investing: Part 2 – Generating Asset Class Assumptions.: Foresti and Toth.

Copyright  2010, Wilshire Associates Incorporated Page 12 Wilshire Consulting Benchmarking Private Investments: Assessing Performance despite Index Imprecision Thursday, March 4, 2010

Important Information

This material contains confidential and proprietary information of Wilshire Consulting, and is intended for the exclusive use of the person to whom it is provided. It may not be modified, sold or otherwise provided, in whole or in part, to any other person or entity without prior written permission from Wilshire Consulting.

The information contained herein has been obtained from sources believed to be reliable. Wilshire Consulting gives no representations or warranties as to the accuracy of such information, and accepts no responsibility or liability (including for indirect, consequential or incidental damages) for any error, omission or inaccuracy in such information and for results obtained from its use. Information and opinions are as of the date indicated and are subject to change without notice.

This material is intended for informational purposes only and should not be construed as legal, accounting, tax, investment, or other professional advice.

Statements concerning financial market trends are based on current market conditions, which will fluctuate. There is no guarantee that these suggestions will work under all market conditions, and each investor should evaluate their ability to invest for the long- term, especially during periods of downturn in the market.

Copyright  2010, Wilshire Associates Incorporated Page 13