Asset-Based Lending

Assessing Securities Lending -Return Performance in a Portfolio Context

by Ben Atkins and Glenn Horner hile many institutional investors embrace paying a rebate rate on this cash collateral.1 Demand securities lending as an attractive tool to to borrow securities usually causes the rebate rate to enhance portfolio returns, others remain wary. fall below the risk-free rate (the line). For some secu- WMany perceive securities lending to be an eso- rities, this spread is quite substantial; the demand teric distraction—a tool limited by risky, immaterial spread represents the “specialness” of a security to returns. State Street addresses the latter view by borrowers. Some lenders are content with the grounding investors’ performance analysis on risk- demand spread; they simply invest the cash collateral adjusted returns. The data lead to two key conclu- in Treasury repo. Most lenders, however, seek addi- sions for investment managers and plan sponsors: tional returns by investing the collateral in high-qual- 1. Although securities lending returns are relative- ity money market instruments (collateral reinvest- ly small, superior risk-adjusted performance ment).2 In this way, they capture reinvestment highlights its value. returns by assuming a limited degree of credit and 2. Managers may optimize their lending program through a broader framework that integrates the Figure 1 risk-return performance of the underlying Disaggregation of Securities Lending Returns investments. Increasingly, investors focus on minimizing Reinvestment return “frictional” losses due to management fees, commis- sions, and inefficient trading. A well-structured lend- Reinvestment spread ing program represents an attractive tool to offset Risk-free rate some of these losses. Demand spread An assessment of securities lending requires disaggregating program returns into two Rebate rate components: 1) demand, or “below-the-line,” spread; and 2) reinvestment, or “above-the-line,” spread. Depicted in Figure 1, this framework derives from the lending transaction. Typically, a prime broker (the borrower) provides cash collateral to borrow a security; the lender compensates the borrower by © 2006 by RMA. Ben Atkins, CFA, FRM, is a manager in the Corporate Initiatives Group, General Electric Company, in Fairfield, Connecticut; formerly, he was Asset- Liability manager at State Street Bank. Glenn Horner, CFA, FRM, PRM, oversees Portfolio Strategy and Quantitative Modeling for State Street’s Securities Finance Division. 42 The RMA Journal May 2006 Assessing Securities Lending Risk-Return Performance in a Portfolio Context

Figure 2 Comparison of Risk Statistics

Standard Deviation Skewness VaR4 Worst Excess Return Securities Lending Demand spreads 0.006% 0.7 0.01% 0.00% Reinvestment: overnight5 0.004 0.0 0.01 -0.01 Reinvestment: money market5 0.02 0.4 0.04 -0.05

Bond Indexes Lehman Brothers U.S. 1.29% -0.6 2.53% -4.23% Lehman Brothers U.S. Credit 1.35 -0.4 2.24 -4.36 Lehman Brothers Global Aggregate 1.68 0.3 2.70 -3.74

Stock Indexes S&P 500 4.73% -0.5 8.48% -14..90% Russell 2000 6.28 -0.5 8.93 -19.84 MSCI EAFE 4.56 -0.5 8.09 -12.97 Data period: monthly returns from 1/1/98 through 7/31/2005 (n = 91) Source: State Street SL PerformanceAnalyzer®, LehmanLive.com, Bloomberg, State Street analysis

duration risk. Analyzing both of worst-return measures indi- fees, which can be 20-50 basis these return components reveals cate that securities lending points. On this scale, lending the relatively low levels of market resulted in minimal down- returns appear more substantial. risk in securities lending. side exposure compared to In a more rigorous approach, risk- We have studied 7.5 years of major benchmarks.6 adjusted return analysis quanti- lending data. During this period, This analysis highlights the large fies the risk-return trade-off. securities lending was consider- disparity between market risks in An analysis of risk-adjusted ably less risky than familiar mar- investment benchmarks and performance asks: Do returns ket indexes. Figure 2 compares securities lending. adequately compensate an risk data from State Street’s lend- A cynic may respond to investor for the level of portfolio ing program with several key Figure 2, by saying, “Yes, the risk? The addresses benchmark indexes.3 Several con- securities lending volatility is this question through a cost- clusions emerge: trivial, but so are the returns.” benefit approach: 1. The volatility of securities Some investment managers con- lending returns—the stan- clude that lending returns are dard deviation—was trivial immaterial compared to bench- Rp - rf compared to levels experi- mark returns, but the comparison S = ␴ enced in key benchmarks. is spurious. Active managers do p 2. The skew statistics show that not provide the benchmark extreme returns over the last return—they add value by seek- 7.5 years of lending tended ing alpha—incremental perform- With Rp the portfolio return, ␴ to be in excess of the mean, ance in excess of the rf the risk-free rate, and p the while extremes in bench- benchmark.7 Top managers deliv- standard deviation of portfolio mark returns tended to be er 100-400 basis points of alpha. returns; this ratio may be inter- unfavorable. Similarly, passive managers preted as the excess return pro- 3. The value-at-risk (VaR) and emphasize low and vided for each unit of volatility.8

43 Figure 3 Analysis of RIsk-Adjusted Performance Mean Excess Return (annualized) Sharpe ratio Securities Lending Demand spreads 0.21% 10.1 Reinvestment: overnight 0.02 1.6 Reinvestment: money market 0.24 2.8

Bond Indexes Lehman Brothers U.S. Govt 2.47% 0.6 Lehman Brothers U.S. Credit 3.05 0.7 Lehman Brothers Global Aggregate 2.45 0.4

Stock Indexes S&P 500 1.19% 0.1 Russell 2000 3.75 0.2 MSCI EAFE -0.28 0.0 Data period: monthly returns from 1/1/98 through 7/31/2005 (n = 91) Source: State Street SL PerformanceAnalyzer®, LehmanLive.com, Bloomberg, State Street analysis

Figure 3 provides a key finding: 3. The money market reinvest- this finding should further reduce Lending returns were smaller than ment strategy drives superior concerns about securities lending market index returns, but lending risk-adjusted performance by . If, for example, an offered superior risk-adjusted per- capturing additional returns investor suffers losses from a formance over this time period. through modest exposure to Russell 2000 portfolio, lending Both sources of securities duration and . does not amplify these losses. lending returns—demand and On a stand-alone basis, the Over time, lending returns offer a reinvestment spreads—contribute demand and reinvestment spreads small dampening effect for overall to this conclusion. Taking each represent a compelling source of volatility. component in turn: low-risk, incremental returns. Construction of an efficient 1. The high Sharpe ratio for Securities lending also performs frontier offers a useful synthesis of below-the-line returns stems well when viewed within a broad- the benefits of securities lend- from borrowers’ stable demand er portfolio context. ing—low-risk incremental returns for securities. As one invest- Specifically, correlation analy- with low correlation to investors’ ment manager remarked, sis demonstrates that lending portfolios. Figure 5 shows how “The demand spread looks returns provide modest diversifi- lending returns have expanded like a free lunch.” cation by exhibiting zero correlation the efficient frontier.9 The lower 2. The overnight reinvestment to benchmarks. Figure 4 provides frontier depicts, for a given level strategy achieves a favorable the correlation coefficients for the of volatility, the maximum excess Sharpe ratio by earning a components of lending returns monthly return earned with the small, nearly fixed spread by with the returns from a composite optimal allocation across asset assuming incremental credit index of underlying lendable classes. Using typical on-loan uti- risk on the collateral. Speci- assets. A correlation less than 1.00 lization levels, the higher frontier fically, reverse Treasury repo leads to a reduction in the overall shows that lending increases sets the risk-free rate, and the portfolio’s volatility. This effect monthly returns by 0.5-1.5 basis 2-basis-point spread repre- increases as the correlation falls to ponits across a range of alloca- sents a weighted average of zero; correlations less than zero tions.10 Lending improves the repo spreads for lower grades indicate that returns tend to vary efficient frontier by consistently of collateral (e.g., agencies, inversely with the underlying adding incremental returns, while mortgage-backed securities). portfolio. The effect is small, but reducing the volatility of the over-

44 The RMA Journal May 2006 Figure 4 Correlations of Returns from Lending and Underlying Portfolios

Reinvest Spread— Demand Spread Reinvest Spread—Overnight Composite of Client Assets Short-term

Demand Spread 1.00

Reinvest Spread—Overnight (0.48) 1.00

Reinvest Spread— (0.12) 0.03 1.00 Short-term

Composite of Client Assets 0.05 (0.01) (0.10) 1.00

Note 1: Composite of client assets represents the asset-weighted returns from the Lehman Brothers US, Europe, and Asia bond indexes; MSCI North America, Europe, and Asia Pacific equity indexes Note 3: Correlation uses monthly returns from 1/1/98 through 7/31/2005 (n = 91, DF = 89) Source: State Street SL PerformanceAnalyzer®, LehmanLive.com, Bloomberg, State Street analysis

all portfolio. Graphically, the adjusted performance. opposite strategy of the Overnight Sharpe ratio is the slope of the Figure 6 outlines several column. Investors with high risk line from the origin to the risk- approaches to developing a col- tolerance lend securities to return point on the graph. lateral reinvestment strategy. A finance purchases to replicate the Lending steepens this line by program that restricts collateral exposures of the underlying port- shifting the endpoint up and to reinvestment to overnight repo folio. funds use the strate- the left. opts to avoid the risks of an gy in Pure Leverage to maximize How should investors respond asset-liability mismatch. The pro- returns by embracing the addi- to evidence that securities lending gram captures the value inherent tional risk inherent in leveraging favorably impacts their efficient in the demand spread, but it sac- the underlying portfolio. Although frontier? The analysis should rifices reinvestment returns it increases expected returns, the influence the response to two (compensation for reduced liq- Pure Leverage approach does not overarching questions: uidity and increased duration and improve overall risk-adjusted per- credit risk). The overnight strate- formance beyond the value of the First: Should I establish a securities gy often appeals to investors with demand spread.11 Pure Leverage lending program? higher risk aversion and owners also raises concerns around liquid- of securities with large demand ity risk. An example of liquidity How should I structure my collateral spreads (e.g., small-cap or non- risk in a Pure Leverage strategy reinvestment strategy? U.S. equities). The approach, would be Long Term Capital however, fails to recognize that 1) Management in 1998. The The risk-return trade-off is the large Sharpe ratio implies increased risk leads most plan not a legitimate basis for answer- that incremental returns can be sponsors to reject its use. ing “No” to the first question. An gained with a limited increase in Most investors select a com- analysis of securities lending risk and 2) the low correlation promise strategy, which views market risk reveals levels signifi- with benchmark portfolios limits securities lending as a source of cantly smaller than benchmark the overall impact of increased incremental income (the Money risk measures. Although its reinvestment risk. Market column of Figure 6). returns are smaller, securities Figure 6 depicts a spectrum Reinvestment in money market lending—with Sharpe ratios in of risk in structuring the collateral assets leads to returns in excess of excess of market benchmarks— reinvestment strategy. The Pure the overnight strategy, and—in provides extraordinary risk- Leverage column represents the contrast to Pure Leverage—the

46 The RMA Journal May 2006 Assessing Securities Lending Risk-Return Performance in a Portfolio Context

Money Market model Figure 5 increases the overall Impact of Lending on the Efficient Frontier risk-adjusted perform- ance to a greater 0.35% degree. Nonetheless, With Lending it remains inefficient: 0.30% Russell 2000 Portfolio optimization Without Lending requires that investors 0.25% equalize the Sharpe Lehman Credit ratio across their sub- 0.20% portfolios. Lehman Government Misalignment in 0.15% Sharpe ratios implies Lehman Global Aggregate that an investor could 0.10% attain the same expect- S&P 500

ed return with less Monthly Return Excess volatility (or, with the 0.05% same level of risk, high- er expected returns). 0.00% This line of reasoning, MSCI EAFE -0.05% for example, highlights 1.0% 2.0% 3.0% 4.0% 5.0% 6.0% 7.0% the inconsistency of Monthly Return Volatility owning a portfolio of Data period: monthly returns from 1/1/98 through 7/31/2005 BB-A rated, five-year Source: State Street SL PerformanceAnalyzer®, LehmanLive.com, Bloomberg, State Street analysis bonds, while restricting the collateral reinvest- vestment portfolios (the Enhanced Alternatively, since lending returns ment to 90-day A1/P1 commercial Money Market column of Figure are comparatively small, an paper. The investor would benefit 6), while shifting the underlying investor may reasonably conclude: by assuming more risk in the rein- portfolio to higher-quality issues. “Well, I’m comfortable with the Figure 6 Collateral Reinvestment Risk-Return Profile

Risk Profile Overnight Money Market Enhanced Money Market Pure Leverage

Focus ● Minimum risk ● Income ● Risk-adjusted performance ● Total return

● Extracts demand spread from ● Reinvestment extends/ ● Short duration (1-4 mo.) ● Attributes “specials” Relative value drives duration complements underlying ● Limits collateral reinvestment ● Superior credit (A1/P1) and credit positioning to overnight repo portfolio

● ● Lending should be optimized ● Must segregate strategies for within the overall portfolio Lending is simply a tool to ● Demand spread “free lunch” lending and underlying leverage underlying invest- Key Assumptions too good to pass up investments strategy ments ● Excessive conservatism ● Leverage undesirable ● Every loan must have positive ● Client objectives support spread every day creates risk-return opportunity increased risk cost

● Prime brokers for hedge Entities Lending agents for beneficial owners funds

47 Assessing Securities Lending Risk-Return Performance in a Portfolio Context risk-return profile of my high-yield control over the management 5 The analysis breaks the collateral reinvestment portfolios into two strategies: overnight, where bond portfolio, so I should accept a process may lead to less risk toler- investments are limited to overnight repo trades; bit more duration and credit risk in ance in the lending program. In and money market, where investments include repo, commercial paper, and asset-backed securities. my lending program.” addition, the strategy represents a The Enhanced Money departure from “the standard 6 Of course, it is inappropriate to imply that large losses are impossible in securities lending. The Market strategy seeks to address model.” In the same way that “no industry has experienced several chastening the inefficiencies associated with one ever got fired for hiring instances, including duration-based losses in 1994 and default losses in 2003. “mental accounting.” John IBM,” a broad shift to this model 7 Christopher Wright (2005) discusses current Nofsinger (2002) describes this will await a period of success by thinking on portfolio alpha. concept of behavioral finance: more innovative plan managers 8 Frank Reilly and Keith Brown (2003) provide a Imagine that your brain uses a mental and boards. cogent discussion of the Sharpe ratio and other accounting system similar to a file cabi- Many investment managers risk-adjusted performance measures. net. Each decision, action, outcome is and plan sponsors segregate their 9 The efficient frontier is usually prospective, based lending program from their port- on expected asset class return, volatilities, and cor- placed in a separate folder in the file cab- relations. Figure 5 is entirely retrospective using folio management. They accept inet. The folder contains the costs and monthly returns from 1/1/98 to 7/31/2005. significant portfolio volatility as benefits associated with a particular 10 The analysis used the following utilization “part of investing,” but they levels: S&P 500, 4%; Russell 2000, 25%; MSCI decision....[This] affects investors’per- structure their lending program to EAFE, 25%; Lehman Government, 75%; and Lehman Credit, 15%; Lehman Global Aggregate, ceptions of portfolio risks. The tendency avoid even small losses. The 75%. to overlook the interaction between superior risk-adjusted perform- 11 This assertion may be conceptualized using the investments causes investors to misper- ance of securities lending chal- Markowitz risk-return frontier. The Sharpe ratio equals the slope of the Capital Market Line. ceive the risk of adding...to an existing lenges this model, highlighting Extensive use of “Pure Leverage” requires borrow- the value of focus and an integrat- ing at rates above the risk-free rate (negative portfolio....The process does not lead to demand spread), degrading risk-adjusted returns. ed perspective. A rigorous the benefits of diversification. approach to optimize the risk- References Portfolio optimization demands adjusted performance of securities that investors integrate their deci- Nofsinger, John R, 2002, The Psychology of lending within a portfolio frame- Investing, Prentice Hall, Upper Saddle River, NJ. sion-making about portfolio allo- work represents the next frontier, cations and collateral reinvest- Reilly, Frank K., and Keith C. Brown, Investment offering low-risk, incremental Analysis and Portfolio Management, South- ment. Today, most clients per- Western Mason, OH, 2003. returns for investors. ❐ ceive lending as simply a source Wright, Christopher, 2005, “Ripe for the Picking,” of incremental earnings; they CFA Magazine (Sept-Oct 2005), pp. 26-35. Contact Ben Atkins at forgo additional returns by falling [email protected]; into the trap of “mental account- contact Glenn Horner at ing.” Personal finance offers an [email protected]. example of this behavior: Many Share your comments or questions people allocate funds for retire- about articles appearing in The Notes ment savings with an expected RMA Journal with Beverly Foster at return of 4-5%, while they contin- 1 Agent lenders typically maintain the cash collat- [email protected]. If contacting eral at 102-105% of the value of borrowed securi- ue to pay higher interest on mort- ties, further reducing the default risk inherent in the author directly, please "cc" gage and credit card debt. The the collateralized loan. [email protected]. Thanks. current securities lending para- 2 Typical guidelines limit reinvestment assets to digm exemplifies this phenome- A1/P1 commercial paper and the portfolio duration to 90 or 120 days. non in institutional investing. 3 To foster a consistent comparison with the lend- Several organizational factors ing performance, index returns are “excess may impede plan sponsors’ broad returns”—i.e., the analysis assumes that invest- ment capital is raised at the risk-free rate of bor- acceptance of the enhanced rowing, so returns are net of the overnight Treasury money market strategy. For exam- reverse repo rate. ple, some plans use commingled 4 The VaR is calculated as a one-month measure pools for their collateral reinvest- with 95% confidence—i.e., the VaR represents the minimum shortfall from the mean return that is ment. The perception of reduced expected once every 20 months.

48 The RMA Journal May 2006