Understanding the Private Equity Premium Eric Johnson Managing Principal, TVPI Advisors Learning Outcomes

• Distinguish between required return premiums for illiquidity and other types of associated with PE investments • Identify LP-specific factors affecting the level of required return premium • Describe key performance methodologies available via ILPA toolkit • Historical return premiums versus various public stock indices • Limitations of commonly-used approaches such as “horizon returns” • Apply these methodologies to generate return premium expectations: • Consistent with actual historical results • Customizable based on current market conditions & LP’s own access/skills

2 Is THIS the Key Question for LPs?

“Does Private Equity Beat Public Markets?”

3 It’s Close, But a Better Question for LPs Is:

“Does Private Equity Beat Public Markets by Enough, for Us?”

4 Standard Assumptions for Asset Allocation Modeling Typically Presume a “PE Premium” Above Return for Stocks

. Assumes “riskier” asset classes have an inherently higher expected return.

. Often shown as a roughly linear relationship, moving upwards and to the right.

5 However, It’s Better to Think of Risk/Return in Terms of Distributions

. Not a straight Another View of Capital Markets Assumptions progression (H/T Howard Marks) upwards and to the right. Public Equities (Stocks)

Private Equity/VC . Potential for higher returns, but also Bonds greater risk of Annualized Return Cash lower returns as LP moves to right.

Volatility (aka "Risk")

Source: “Risk Revisited”, Howard Marks, OCM Client Letter, September 3, 2014. Retrieved April 4, 2018 from https://www.oaktreecapital.com/docs. 6 KEY ISSUE: Do We Really Know Where the Distribution of Future Returns for PE/VC is Centered?

. It is possible the entire distribution is in a different place.

. Not necessarily where it “SHOULD” be.

. If so, we would still expect some PE/VC investments to beat equities by a decent margin (“top quartile”?!) 7 BIG PICTURE  $2.1 Trillion of US PE/VC: Mediocre Results vs. Average S&P 500 Stock, Mid Caps, or Small Caps

. Academic & industry studies often show outperformance versus S&P 500 (or Russell 3000).

. Outperformance drops or disappears entirely when compared to S&P 600 Small Cap, S&P 400 Mid Cap, or Equal-Weighted S&P 500 indices.

8 How Could This Be Possible? Inefficiencies, of Course!

. Short history of institutionalized PE/VC (i.e., decades)? . Poor benchmarking choices and limited history of specialized stock indices? . Changing nature of PE/VC industry (How relevant are earlier results?) . Limitations on transparency of true “comprehensive” PE/VC industry results? . Is available data “representative”, or perhaps skewed? If so, how? . Long time horizon for results & feedback loops to affect LP behavior? . “Math issues” that complicate assessment of results even once they’re known? . Capture of too large a portion of any true premium by GPs and Sellers? . Human nature (e.g., greed, self , anchoring, overconfidence)? . New LP entrants eager to replace those LPs that exit, cut back, or push for better economic terms?

9 What Should an LP Be Trying to Do?

STEP 1 STEP 2

Determine Determine Required Level of Premium Expected Level of Premium to Meet LP’s Portfolio Should Be the LP Can Reasonably Plan Objectives Lower to Achieve Than or Equal to A. Illiquidity A. LP-specific issues Premium?? B. Illiquidity Risk B. Historical Returns C. Other PE Risks C. LP-specific issues D. Market Environment

10 STEP 1: Determining the Required Level of Return Premium, Including LP-Specific Factors

. What are our objectives for allocating to PE/VC? . What is our “Opportunity Cost” versus other uses of capital? . Could other more liquid investments provide similar exposure to “factors” such as broad equity market, small-cap premium, leverage, technology? . How much value-added do we expect from our active stock managers? . What is the size/nature of our future spending requirements? . How valuable is “liquidity” to us, and how does that change at different levels of overall portfolio illiquidity? . How large is our allocation to PE/VC and other illiquid assets? . Do we have other non-return-based objectives for PE/VC exposure? . How confident are we in our assumptions for potential PE/VC value-added?

11 Distinguishing between “Illiquidity Risk” and Other PE/VC Risks for Which LPs Should Be Compensated

. How Much Additional Premium is Needed for Other PE/VC Risks: . How Large a Premium is Needed . Smaller size & lower quality of for Illiquidity Risk: companies (vs. mega/large-caps) . Blind-pool risks . Average illiquidity of PE/VC assets . Manager-selection risks . Extra compensation for poor results in liquidity shock?* . Reduced legal rights . Agency risks . Relatively high direct costs & increased oversight costs . Distraction & reputation risks for board and institution?

* See “The Ins and Outs of Investing in Illiquid Assets”, Robeco, 2015, which cited research on marketable securities published in 2011 by A. Khandani and A.W. Lo. 12 Marginal Required Premium Curve versus “Average” Premium

. As the percentage of the LP’s total assets invested in illiquid assets gets larger, the marginal required premium for additional levels of illiquidity likely increases.

. Steepness & actual levels depend on LP-specific factors.

From “Policy Benchmark Selection”, ILPA White Paper Series, TVPI Advisors, August 2017. For further information on required premium curves, see Mark Hayes, James Primbs, Ben Chiquoine, A Penalty Cost Approach to Strategic Asset Allocation with Illiquid Asset Classes, Journal of Portfolio Management, Winter 2015. 13 Some Academics Believe Required Premium Just for Illiquidity is Quite Large

. Andrew Ang of Columbia and colleagues developed model suggesting an LP owning an asset with a 5-year required holding period would require an additional 4.3% per year just for illiquidity.

. For shorter, 2-year period, Ang’s required premium would be 2.0% per year.

. For 10-year period, Ang’s illiquidity premium would be around 6.0% per year. (Or higher, since model didn’t include other important issues.)

. How much more premium is needed for other PE risks? Are the most common 300- to 500-bps overall targets in ILPA & CA survey data sufficient?

Source: Ang, Andrew, 2014, Asset Management: A Systematic Approach to Factor Investing (New York: Oxford University Press), p. 438. 14 Remember Our Question?

“Does Private Equity Beat Public Markets by Enough, for Us?”

15 STEP 2: Determining Expected Level of Premium an LP Can Reasonably Plan to Achieve

. Not simply a question of “asset class” returns, as for stocks & bonds . Heavily dependent upon active management skills of GPs . Impossible to deploy capital to obtain a passive “index” return . LP’s size, access, and active portfolio management skills affect results

. Develop hypotheses about future expected returns & premiums (“By Enough”?) . Be consistent with returns data now available on USD trillions in PE/VC . Consider which public indices are most relevant for comparisons

. Customize for market conditions and LP’s own access/abilities (“For Us”?)

16 STEP 2-A: Can LPs Expect Compensation for Illiquidity Risk (i.e., an “Illiquidity Premium” Just for Participating)?

. “…the magnitude of the illiquidity premium is primarily related to supply and demand. If there is a relatively small supply of illiquid asset[s] with an alpha and large demand for illiquid asset[s], then the premium will be squeezed….If my experience is representative, there is plenty of demand compared to supply and it is difficult to see how a significant illiquidity premium could exist….

. …Similarly, if you believe in the illiquidity premium, then you will get the same one if you just buy anything directly yourself, no need to go through a specialized intermediary.” -- Ludovic Phalippou, Private Equity Laid Bare, 2017- 2018, pp. 178-185 (emphasis added)

17 Can LPs Expect Compensation Simply for Illiquidity Risk? (continued)

. The published industry statistics of most relevance to asset owners are fund returns net to LPs (after GP fees, expenses, and carried interest).

. These PE/VC peer group returns do not include LPs’ implementation and oversight costs (except, to a large degree, for funds of funds).

. Underlying assets in which PE/VC funds invest are priced in a broader competition among GPs & other actors, including strategic investors who also have long-term horizons and may bid up prices because of synergies the PE fund lacks (e.g., leading to “the winner’s curse”).

18 Can LPs Expect Compensation Simply for Illiquidity Risk? (continued)

. Total Gross multiple of capital for US PE/VC funds of 1.87x is: . Well above Net TVPI multiple to LPs (1.63x) . Well above mPME multiple for S&P 600 small caps (1.61x) . In-line with S&P 600 plus 300 bps (1.88x). . Below S&P 600 plus 400 bps (1.99x). . If Gross is like a ceiling, it’s hard to reach “Stocks plus 300-500”

19 Can LPs Expect Compensation Simply for Illiquidity Risk? (continued) . Even in the case that a substantial “illiquidity premium” for private transactions does exist, it could be based on the economics of the entire deal, rather than the implied pricing for the capital provided by LPs via one or two layers of intermediary funds (i.e., direct and FoFs) & including LP’s own costs.

. Much or all of such an “illiquidity premium”, if it exists in theory, could be captured by GPs, fund expenses, and other intermediaries.

. With abundant PE/VC capital available, Sellers may also capture too large a share of the potential economics of many deals. While Sellers & their advisors may have been more naïve in industry’s early days, now they are well-prepared (and familiar with stories of “riches” made by Yale, GPs, other founders, etc.)

20 STEP 2-B: Historical Market Returns

. The most commonly-used “evidence” in favor of the long-run “market- beating returns” of PE/VC have been “Horizon Returns” such as those below:

Horizon Returns for Periods Ending December 31, 2016 (%)

Asset Class/Strategy 5 Years 10 Years 15 Years 20 Years 25 Years

US Venture Capital: Early Stage 15.2 9.8 6.1 57.7 35.2 US Venture Capital: Total 14.0 9.4 6.8 26.1 25.4

. Although the 15-Year results appear mediocre, those long-term results sure look good, way above stocks. Perhaps we’re done with STEP 2-B?!

. Well, not quite… 21 Why Not Just Use Long-Term “Horizon Returns”?

. Fixing the inception date shows how strong early results can “lock in” long- term horizon returns that are frequently reported as strong evidence of a VC premium.

. Effect is rarely visible, since common reports are based on trailing 10-year, 20-year periods that roll inception point forward each quarter.

22 As a Result, We Can Predict Future Horizon Returns for US VC Industry --- Just Not In a Way that Is Useful

. Much of what appears to be strong “long-term performance” for US VC is caused by the IRR calculation’s reinvestment assumption.

. Future extreme results do little to change the long-term “returns” that will be in WSJ and PPMs in 2021.

23 Even Horizon Returns over Shorter Periods Can Be Unreliable, Though for Different Reasons . Horizon Returns rely on valuation, size, and maturity of existing illiquid assets at start of measurement period, in ways that are rarely transparent to LPs. . Example: US VC “returns” for 5 and 10 years through 2016 are 14.0% & 9.4%. . But…just what is included? . Two-thirds of initial NAV is in funds of vintages already five or more years old, which would not be available to LPs making new commitments in 2007 or 2012 (& which are wrongly valued). . Without older VYs, 5-year return jumps to 22.5%, & 10-year return to 13.3%. . Beware: Also affects mPME horizon return comparisons versus stocks.

24 TOOLKIT: Some Key Methodologies Available via ILPA

. Direct Alpha

. CA mPME (especially comparisons to DPI and TVPI)

. KS PME

. Screening by Quartile and Other Percentiles

. Fund Sequencing: New (I & II), Developing (III & IV), Established (5+)

25 CA mPME Shows LPs Could Have Achieved 1.60x TVPI Multiples Just by Investing in Average S&P 500 Stock

. Despite of annual stock returns and periodic market crashes, total multiples from holding US stocks per the cash flow patterns of PE/VC funds were high & stable. . S&P 500 Equal-Weighted (EW) Index around 1.5x or above for all mature VYs. (Average is 1.6x)

. With premium of 300 bps, average TVPI multiple for these stocks would be 1.85x.

26 US Funds Exceeded 2.0x TVPI by Wide Margins for 1991 to 1995 Vintages (Helped by Internet Bubble Exits)

. Returns for earlier, smaller vintages 1991 to 1995 exceeded 2.5x, with strong VC results in lead (though S&P Info Tech mPME also averaged 2.65x for 5 VYs).

. BUT, only 3 vintages since 1997 have DPI above 1.60x. None since then have a DPI above the 1.85x represented by S&P 500 EW + 300 bps. 27 STEP 2-B: US PE/VC: Mediocre Results vs. Average S&P 500 Stock, Mid Caps, or Small Caps for 1995-2016 VYs

. Recall the earlier overall results versus S&P 600 Small Cap, S&P 400 Mid Cap, or Equal-Weighted S&P 500 indices: modest/no premiums.

. Even inclusion of strong VC results for 1995 to 1997 vintage-years only gets Direct Alpha up to 100-150 bps.

28 Could Data Be WRONG? It Is Unlikely that “Hidden” Funds Change Industry’s “True” Outperformance Dramatically

. Could poor coverage by benchmark providers mask strong “true” industry results?

. Coverage is never complete.

. It would require heroic assumptions about size & results of “hidden” funds to move industry’s “true” mPME Delta to 3%-5%. 29 A Handful of US Private Equity (without VC) Vintages Have Had > 3% to 5% Premium to Small Caps

. Direct Alpha by Vintage Year for US PE (ex VC) is below 300 bps for most years when compared to S&P 600 Small Cap Index or S&P 500 Equal-Weighted Index.

. Exceptions are 2000 to 2004 and 1995 vintages, most of which had relatively small amounts of capital raised. 30 These New Results Are Broadly Consistent with 2016 Study on US Buyouts by L’Her et al. (Using Earlier Burgiss Data)

. Over 200 basis points of “outperformance” versus S&P 500 disappeared when using S&P 600 Small Cap Index.

. Nearly 200 more bps disappeared when adjusting for sector and leverage (not covered today, but which would further reduce any premiums). 31 STEP 2-C: LP-Specific Issues (Skills, Access, Scale)

. Manager selection skills and access that are “above average” may be necessary simply for an LP to gain its full proportionate share of the “winners” whose performance drives the industry’s pooled index returns. . Hiring a talented internal team and/or an experienced external advisor is not necessarily sufficient to obtain these pooled returns. . Future top funds are also being sought by other talented, experienced LPs fervently trying to get more than their proportionate share of the winners. . Without an index-proportional weighting to the top performers, an LP with only “average” skills and access may get results well below the pooled averages. . Scale is also an important factor: affects returns via internal and external costs, and can cause access constraints (for both large & small LPs).

32 Without Top 5% or 10% of Funds, Much of an LP’s Potential Direct Alpha Vanishes from Pooled Averages

. The effect of missing out on the very top performers can be seen by removing the top 5% or 10% of funds over a full time period.* . Without the top 5% of US VC funds and US Buyout/Growth Equity funds, outperformance dropped by 238 basis points and 95 basis points. . Without the top 10%, the remaining 90% of the funds had Direct Alpha versus the S&P 600 Small Cap Index of negative 414 basis points for US VC and negative 13 basis points for US Buyouts/Growth Equity. . The absence of a significant premium versus small cap stocks for 90% of the industry during these periods provides further support for the argument that an LP should expect little, if any, general illiquidity premium just for being in PE/VC.

* The 2001 to 2014 vintages were used for US VC to avoid effects of the internet bubble, and 1995 to 2008 vintages were used for US Buyouts/Growth Equity to cover a period with mature funds. Data from Cambridge Associates, Thomson Reuters Datastream, TVPI Advisors’ analysis. Returns in USD through September 30, 2017. 33 Example LPs: US Funds of Funds Have Trailed Objectives of “Stocks Plus 300 to 500 Basis Points” By Wide Margins . US-focused PE/VC FoFs dramatically trailed the most common targets by 500 to 700 basis points.

. More than just a fee effect; suggests manager selection is quite challenging.

. 419 FoFs, $137 billion. Sample may have positive skew, however (i.e., truth

may be even worse). 34 Screening by Quartiles Provides Another Useful Tool for LPs to Analyze Composition of Pooled Results

. Striking difference in performance between pooled results and middle of the PE/VC distribution. . Simple proxy: use funds within the second and third quartiles only. Direct Alpha drops by ~160 bps . LP with only 2nd quartile funds (100% “winners”): Just 70 to 100 bps above S&P 500 Equal- Weighted & S&P 600. . Even a Q1-Q2-Q3 portfolio only beat S&P 600 by 200 bps. Source: Cambridge Associates, Standard & Poor’s, Frank Russell Company, Thomson Reuters Datastream, MSCI Inc., Dow Jones Inde xes, TVPI Advisors’ analysis. Direct Alpha in USD (%) through September 30, 2017. 35 How Good Would Your Manager Selection Have Needed to Be to Achieve 3% to 5% Outperformance vs. S&P 600? Pro-forma mPME Outperformance by Manager Selection Skill, . Actual: 16% of LPs’ capital went to US PE/VC/Credit Distressed Funds, 1998-2014 VYs Q4 funds, and 26% to Q1 funds, Percentage in Bottom Quartile Funds 25% 20% 15% 10% 5% 0% leading to mPME IRR delta -0.35%. 5% -4.2% -3.7% -3.2% -2.7% -2.3% -1.8% 10% -3.5% -3.0% -2.5% -2.1% -1.6% -1.1% . Pro-forma: LP would have needed 15% -2.9% -2.4% -1.9% -1.4% -0.9% -0.5% 50% of capital in Q1 funds to get 20% -2.2% -1.7% -1.2% -0.7% -0.2% 0.2% 25% -1.5% -1.0% -0.5% 0.0% 0.5% 0.9% 300 bps mPME outperformance. 30% -0.8% -0.3% 0.2% 0.7% 1.2% 1.6% 35% 0.0% 0.5% 0.9% 1.4% 1.9% 2.4% . Or 65% in Q1 to reach 500 bps (at 40% 0.7% 1.2% 1.7% 2.2% 2.7% 3.1% same level of Q4 exposure). 45% 1.5% 2.0% 2.4% 2.9% 3.4% 3.9% 50% 2.2% 2.7% 3.2% 3.7% 4.2% 4.7% . If LP avoided Q4 entirely, would still 55% 3.0% 3.5% 4.0% 4.5% 5.0% 5.4% 60% 3.8% 4.3% 4.8% 5.3% 5.8% 6.2% have needed 40% in Q1 to reach 65% 4.6% 5.1% 5.6% 6.1% 6.6% 7.1% 300 bps or 55% to get 500 bps. 70% 5.4% 5.9% 6.4% 6.9% 7.4% 7.9% Percentage in Top QuartileFunds in Top Percentage 75% 6.3% 6.8% 7.2% 7.7% 8.2% 8.7%

Source: Cambridge Associates, Standard & Poor's, Thomson Reuters Datastream, TVPI Advisors' analysis. Outperformance in mPME IRR terms is based on custom-weighted cash flows by quartiles for US PE, VC, Credit, Distressed funds of 1998-2014 vintage years, in USD net to LPs as of September 30, 2107. Remaining capital split evenly across second & third quartiles. 36 “Persistence”, Access, and LP’s Manager Selection Skills

. Persistence is central to the case for PE & its most common fund & fee structures, but itself is very hard to prove because of long time lags. . By the time there may be reasonable data on performance of a firm over multiple, mature funds, the firm and/or the environment may have changed so dramatically the data is irrelevant or misleading.

. There is evidence that persistence of top-quartile performance is not as strong as many LPs believe it to be, or as it may have been in past. . If persistence is “high”, then Access to the top firms becomes more crucial. . On the other hand, if persistence is lower or “weak”, then investing with future top GPs may be harder and require greater LP Manager Selection Skills and resources (or may depend to a greater degree upon luck).

37 STEP 2-D: Market Environment

. A final step for an LP is to adjust its long-term expectations for private investment premiums to reflect current market conditions. . Preqin’s 2018 report cites a record $453 billion of fundraising in 2017 & a record $2.83 trillion of AUM, reflecting an industry that is now 2.43 times larger than it was at the end of 2016, with ~5400 GP firms (up 48% in 10 years). . Bain & Company’s 2018 Global Private Equity report states that “PE consulting at Bain has grown sevenfold over the past 15 years and now represents about one-quarter of the firm’s global business.” . Are these “Goldilocks” conditions, as Preqin notes? Or cause for caution and concern? . If overall PE/VC historical results have actually been modest/poor despite more favorable conditions in the past, should we adjust expectations now?

38 Results of Pre-GFC Funds: Asymmetric “Value-Added” by Top & Bottom Quartiles, Especially versus Distributions . LP investing $100 mm would have received $119 mm in distributions to date . $56 mm less than in stocks; almost 10 years in . Top Quartile total value of $233 mm . But only $12 mm more in distributions so far . Opportunity Cost in Bottom Quartile is Huge . Stocks: $143 mm more in “spendable” results by now 39 STEP 2: Bringing It All Together - What Expected Level of Premium Can You Reasonably Plan to Achieve as an LP?

. LPs can use these four building blocks, the data in these slides, and own insights/analyses to develop an overall expected level of premium vs. stocks.

STEP 2-C STEP 2-A STEP 2-B STEP 2-D LP-specific issues Illiquidity Historical Returns Adjust for Market (Size, Access, Premium? Environment Skills/Resources)

40 With Our Required Return Premium from STEP 1 in Hand, We Can Now Begin to Answer the Question:

“Does Private Equity Beat Public Markets by Enough, for Us?”

41 STEP 2: Example Expected Premium for a Small, Skeptical LP

0% + 1% + -2% + -2%

STEP 2-C STEP 2-A STEP 2-B STEP 2-D LP-specific issues Illiquidity Historical Returns Adjust for Market (Size, Access, Premium? Environment Skills/Resources)

42 STEP 2: Example Expected Premium for a Large Optimistic LP

0.5-1.0% + 4% + -1% + -1%

STEP 2-C STEP 2-A STEP 2-B STEP 2-D LP-specific issues Illiquidity Historical Returns Adjust for Market (Size, Access, Premium? Environment Skills/Resources)

43 In Conclusion, as Historians and Mathematicians: “We Now Know”*

. Much like academics writing the history of the Cold War while US and USSR were still engaged in the ongoing conflict itself, investors face many difficulties assessing the PE premium while PE/VC industry has been rapidly evolving.

. Our initial views from 20+ years ago were based on limited data and an incomplete understanding of the world that only became clearer over time, yet we all remain within the historical period that we are attempting to analyze.

. If we were starting from scratch, with what we now know, would we expect 300 to 500 basis points of premium? Or might we be “anchored” on dated ideas?

*With a nod to John Lewis Gaddis, We Now Know: Rethinking Cold War History, 1997. 44 In Conclusion (continued)

. How can ILPA’s members increase their chances of success? What might a truly “Fair Deal” between GPs and LPs look LPs Need to Target Here like, given the history of modest or negative premiums? . Carry of 20% still make sense? Lower it and/or add hard hurdles? . Fee reductions? . Answer  LPs need to capture larger portion of Top Quartile’s gross return. Fee reductions across other quartiles would not have been sufficient without lower carry on “winners”.

45 Questions?

46 THANK YOU

Eric Johnson Managing Principal, TVPI Advisors t +1(650) 515-2950 e [email protected]

These materials are copyright 2018 TVPI Advisors LLC. All rights reserved. This material is for informational and educational purposes only, and is not intended to be investment advice recommending the purchase, hold, or sale of any security, partnership interest, or other investment. Further reproduction or distribution outside of the ILPA, its member organizations, and their relevant board/committee members without permission of TVPI Advisors is prohibited.

47 Appendix: Supplemental Materials and Data Summaries

48 Illiquidity Risk Premium -- A Simple Experiment

. Fund A: Hypothetical index fund, but with 5- or 7-year lockup via a structure with a deep-pocketed counterparty that will not allow early exit

. Fund B: Index fund with identical equity holdings, and quarterly liquidity

. What illiquidity premium would make an LP ambivalent between A & B? . B would allow LP to rebalance or exit should the index rise to overvalued levels (such as NASDAQ in 2000), should other more attractive investments be identified (EM in 2009), or should the LP wish to reduce its . Is 100 basis points enough for you? 150 or 200 basis points? More?

49 Illiquidity Risk Premium -- A Simple Experiment (continued)

. An LP in Fund A would still be in a better risk position than an LP in a typical PE/VC portfolio . Near certainty that it had ownership in a transparent pool of securities in a diversified group of actual functioning companies . No manager selection or security selection risks . Essentially guaranteed the ability to exit the position fully on agreed date and at same market prices as Fund B . Investments would not perform more poorly than the liquid index Fund B if exit date happened to coincide with a systematic liquidity shock that hurt valuations of less-liquid assets

50 CA mPME -- Compare TVPI Multiples With Better Matching of Cash Flows . Can’t use Time-Weighted public market returns because of cash flow timing . Ideally would “perfectly match” each cash flow with underlying asset and sell corresponding stock index position as each is sold, summing for TVPI. But, not practical for most LPs or benchmarking firms because of data issues.

51 CA mPME Matches Inflows; Makes Assumptions about Outflows; Doesn’t Capture All Timing . CA mPME is designed to make reasonable estimate of remaining cost basis of fund without “breaking”, using fund-level cash flow data available to LPs. . Not the only solution, though. One example: a “Fixed Step” PME could provide useful information on effects of GP’s capital deployment timing other PMEs miss.

52 How Much Has Manager’s Experience Helped in US Private Equity?

. LPs can now compare results using “fund sequencing” to assess whether new or more experienced GPs have performed better.

. For US PE, Funds III and IV have slight edge of ~150 bps over New & Established funds; all failed to meet 300-500 bps for 1998-2014 VYs. 53 LPs May Seek to Outperform by Emphasizing Small Cap Buyouts, But They’re a Very Limited Part of Entire Market

. US Small Buyouts exceeded 400 bps premium for 1998-2014 vintages, and are popular with LPs. Limited capacity, though: Funds’ average size: $192 mm; just 4.6% of all capital in US Buyouts. CA: ~12 funds with $2.2 Bn total size/year.

54 European PE/VC Has Delivered Higher Direct Alpha; But, Similar Gap between Broad Market and Small Caps

. Aggregated European PE/VC results for 1995-2016 and 1998-2014 look relatively good vs. stocks. . Most outperformance versus small/midcap stocks, though, was for vintages prior to 2005. . No CA mPME data on EW European indices, but EW beat market-cap for S&P Europe 350 by 2.2% annually for 15 years to 12/2017.*

* Source: “Outperformance in Equal-Weight Indices”, Tim Edwards, et al., S&P Dow Jones Indices, January 2018. Comparable outperformance for S&P 500 EW was 2.1%. 55 Maybe Secondary Funds Will Have Had Better Success vs Stocks?

. Despite the ability to purchase assets at a discount, US-focused secondaries funds have also struggled overall to beat “Stocks plus 300 to 500 basis points”.

. Little evidence here, either, for an “illiquidity premium” (based on 158 funds, with $113 billion in commitments). 56 Commitment Pacing and Contrarianism Can Help LPs

. LPs should assess whether they are likely to continue investing in PE/VC when it’s unpopular and fundraising dries up, as well as whether they are likely to slow down commitments when the market is overheated.

. The only US PE vintages years after 1995 that substantially beat a 300-basis-points premium above S&P 600 Small Cap Index are 2000 to 2004. 57 Expected Premiums: Some Key Questions

. Do you believe there is a general “illiquidity premium” that you can expect to harvest simply by allocating to PE/VC? . How do you think about historical “pooled” returns versus the popular market-cap-weighted indices such as S&P 500 or Russell 3000, as opposed to: . Equal-weighted S&P 500 Index (i.e., average results for monkeys & darts) . Small-cap (e.g., S&P 600 Small Cap Index) or Mid-cap indices? . Sector and leverage effects? . Should you adjust expectations downward toward median or middle of the Distribution (2d & 3rd Quartile) if your access/skills aren’t above average? . What internal & external costs do you have at your scale (FoF, internal staff)? . Are current market conditions increasing the risk of a lower future premium versus stocks? Or vice versa? 58 US PE, VC, Credit, Distressed IRRs and Relative Performance by Vintage Year versus Various Broad-Market, Mid-Cap, & Small-Cap US Indices (to Q3 2017) DATA TABLE Direct Alpha versus:

S&P 500 Dow Jones MSCI US Total Pooled Median Russell Equal- S&P 500 S&P 600 Russell US Small Small Cap Vintage Fund Cap. IRR IRR 3000® S&P 500 Weighted S&P 400 Info Tech Small Cap 2000® Cap 1750 Year Count ($Bn) (%) (%) (%) (%) (%) Mid Cap (%) (%) (%) (%) (%) (%, Gross) 1991 26 3.6 23.0 21.0 4.7 4.2 5.1 5.5 -3.3 7.4 1992 40 4.7 29.6 18.5 8.9 8.3 10.2 10.2 -0.4 12.6 1993 55 9.2 32.5 18.8 10.6 9.6 12.8 11.7 1.2 16.6 15.1 14.1 1994 74 15.5 35.4 19.9 13.3 12.5 16.0 13.1 4.8 20.2 18.2 17.0 1995 71 17.6 33.3 21.5 13.8 13.1 16.3 12.5 5.6 19.0 20.4 18.3 17.0 1996 76 19.1 30.8 9.5 14.8 14.4 16.8 11.6 7.7 18.8 19.7 16.8 16.0 1997 116 34.5 21.5 11.1 14.1 14.6 11.8 7.6 14.1 11.0 12.9 9.9 9.5 1998 135 59.6 8.9 6.9 6.6 7.6 1.0 -1.1 14.8 -1.3 1.5 -0.8 -1.0 1999 161 66.0 4.2 0.1 1.8 2.7 -3.4 -4.3 10.6 -5.1 -2.6 -4.1 -4.4 2000 228 115.4 9.4 3.0 4.1 4.8 0.3 0.0 7.3 -0.6 1.0 -0.2 -0.4 Source: Cambridge Associates, 2001 113 51.6 14.3 8.0 6.7 7.4 3.8 3.4 7.5 3.1 4.2 3.0 2.9 Standard & Poor’s, Frank Russell 2002 72 35.4 15.7 8.8 7.1 7.7 4.6 4.5 6.7 4.6 5.4 4.2 4.2 Company, Thomson Reuters 2003 77 37.9 17.3 8.6 10.1 10.6 8.2 7.6 9.5 8.2 9.2 8.0 8.0 Datastream, MSCI Inc., Dow Jones 2004 119 59.2 10.6 6.7 4.6 4.9 3.0 2.5 2.7 3.4 4.3 3.1 3.2 Indexes, TVPI Advisors’ analysis. MSCI 2005 159 97.0 9.1 7.4 3.2 3.4 1.5 1.2 0.7 1.8 2.9 1.7 1.7 2006 187 212.6 7.9 7.9 0.3 0.5 -1.4 -1.4 -2.2 -1.3 0.1 -0.9 -1.1 data provided "as is" without any 2007 177 194.5 11.0 11.1 0.8 1.0 -1.0 -0.5 -1.4 -0.8 0.7 -0.1 -0.5 express or implied warranties. Pooled 2008 160 180.3 12.5 10.0 -0.8 -0.7 -2.2 -1.5 -2.9 -2.0 -0.5 -1.0 -1.5 returns are in US Dollars net to 2009 78 53.3 16.3 15.1 1.4 1.4 0.6 1.2 -0.1 0.3 2.0 1.8 1.2 Limited Partners, through September 2010 105 57.8 13.8 12.6 -0.3 -0.4 -0.7 0.1 -2.8 -1.0 0.7 1.0 0.2 30, 2017 for US private equity, venture 2011 123 107.1 16.6 12.7 2.5 2.3 2.2 2.8 -1.3 1.7 3.3 4.0 3.0 capital, credit, and distressed funds. 2012 126 106.6 15.6 12.9 2.5 2.3 2.5 2.9 -2.6 1.7 3.4 4.5 3.3 2013 137 125.7 13.9 11.0 1.9 1.6 2.4 2.5 -4.9 0.9 2.4 4.1 2.8 Direct Alpha in dark shading 2014 150 145.2 12.6 10.8 0.3 0.1 1.2 0.9 -6.7 -1.3 0.0 2.3 0.7 represents outperformance by 500 2015 147 146.6 15.8 9.0 0.5 0.5 1.6 1.3 -7.7 -1.5 -0.8 2.1 0.1 basis points or higher, whereas figures 2016 120 143.2 10.2 2.3 -6.7 -6.6 -4.8 -4.6 -16.4 -7.8 -8.5 -5.4 -7.2 in light shading represent 3,032 2,099.2 14.0 9.5 4.5 4.7 2.7 2.1 2.7 4.0 outperformance by 300 to 499 basis points. Vintage years determined by 1995-2016 Only 2,837 2,066.1 12.0 9.0 3.8 4.1 1.6 1.1 2.8 1.2 2.7 1.7 1.4 timing of first cash flow. 1998-2014 Only 2,307 1,705.1 10.4 9.0 2.7 3.0 0.3 0.1 2.1 -0.2 1.4 0.5 0.2

1995-2004 Only 1,168 496.3 13.2 6.7 7.1 7.7 3.8 2.4 8.9 3.0 4.8 3.1 2.8 2005-2011 Only 989 902.6 10.6 10.4 0.8 0.9 -0.7 -0.4 -1.6 -0.6 0.9 0.2 -0.2 59 Ten-Year Time Horizons: Even the Post-1997 VYs with a Positive TVPI Premium Trail in DPI Terms after a Decade . $100 mm committed per VY . 1991-1997 VYs have distributed over $100 mm . 2000-2004 VYs are ahead on TVPI terms, but LPs would have $20 mm to $30 mm more liquidity from stocks for most vintages . Capital raised per vintage is in gray; big shortfalls in DPI terms over 10 years versus stocks when size of industry rising sharply or large 60