Assessing Fund Performance:

The goal of benchmarking is to measure comparative performance. When it comes to cars and other manufactured products, measurements are precise and benchmarks are consistently applied. As a result, benchmarks create value by identifying performance gaps and enabling better decision making based on facts. Venture benchmarks, however, are a different story.

As a way to compare the performance of venture capital funds over time, available industry benchmarks can be inconsistent and confusing. Not surprisingly, LPs often seek independent verification of these claims. To assess the performance of our own funds, SVB Capital continuously examines the best methods of using these statistics. In this issue of Venture Capital Update, SVB Capital shares our findings about

venture capital industry benchmarks. We explain the metrics and methods and review the benefits and limitations of benchmarks commonly used in the industry. It is our hope that a better understanding of benchmarking across the investment community can lead to improved means of developing and gaining value from the benchmarks.

WHY WE NEED BENCHMARKS Making an investment in venture capital is a long-term commitment, with 10 years being the typical lifespan of a fund. During this period, LPs receive quarterly financial reports on capital calls and distributions related to their investment. However, they also need to understand how their investment is performing while the capital is put to work during the “J-curve” and before the fund’s portfolio

Venture Capital Update

written by:

View the Fourth Quarter 2007 U.S. Private Equity Snapshot

is completely realized. Financial statements alone will not provide this perspective, so LPs typically turn to benchmarks.

In addition to gauging the returns they might expect over time, LPs also require a way to compare the performance of investments across their portfolio. This is true even when investments are in different asset classes such as public equity and private equity, and regardless of whether the LP is a private individual, endowment, private or public pension fund.

But benchmarks are also an important indicator for venture capitalists (VCs). Sizing up the performance of a fund in the middle of its life cycle is key to assessing how portfolio companies are performing relative to the market.

PERFORMANCE METRICS: APPLES AND ORANGES

When assessing the return-on-investment performance of a venture fund, three different metrics are typically used:

• internal rate of return (IRR)

• distributions to paid-in capital (DPI)

• total value to paid-in capital (TVPI)

IRR provides an effective rate of return based on cash flows and current valuations of the fund portfolio, while DPI shows the realized portion of the portfolio that was distributed to the LP as a multiple of the contributed capital. By comparison, TVPI provides a multiple value on the entire portfolio—both distributed capital and the net asset value of the portfolio.1

Which of these metrics is the best assessment of fund performance? The short answer is, “it depends.” Many LPs rely on IRR measurements of

The performance of a venture capital fund can be calculated via at least one of the following metrics:

IRR: The annualized effective return rate which can be earned on the contributed (invested) capital, i.e. the yield on the investment.

DPI: The ratio of cumulative distributions to limited partners divided by the amount of capital contributed by the limited partners. TVPI: The sum of cumulative distributions to limited partners and the net asset value of their investment, divided by the capital contributed by the limited partners.

performance because they manage a portfolio that includes a mix of public and private investments.

Consequently, IRR reported

as a percentage provides an easy comparison to return percentages on public investments, even though IRR percentages are not completely comparable.2 _{LPs will often look for}

400 to 600 basis points over a public benchmark to justify the illiquid and long-term nature and risk profile of VC investing. DPI provides a clear metric of the actual multiple of cash invested which has been received by an investor, and TVPI provides a metric that accounts for potential returns that are the result of increased valuations of portfolio companies as they approach exit. Given this difference, many LPs rely on TVPI earlier in the life of a fund and DPI towards the end. In contrast to IRR, TVPI and DPI do not account for the time it takes to produce these gains.

Despite the shortfalls, the three metrics have become the standard for comparison. In our experience, we have found that LPs rely on a combination of all three metrics to assess the performance of their investments, with some favoring one over the other, in part, due to the preference of their board and their specific type of investment.

All three metrics can result in a biased assessment of fund performance due to the way each is calculated. For instance, the calculation of IRR is greatly influenced by the timing of returns in a fund, or more specifically, short holding periods.3 _{The example}

above shows two funds of identical size and capital call timing. Fund A provides a steady return of 2.0x to its investors during the last four years of the fund with an ending IRR of 14 percent. Fund B returns only 1.1x, but with an IRR of more than 2000 percent due to the large returns early in the fund’s life, soon after the investment was made.

This example shows that funds with lower IRRs can still provide higher

requiring large amounts of capital very early in a company’s life cycle, and requiring longer investment periods, may inherently generate lower IRRs. Larger capital calls would occur earlier and returns would be realized later in the life of this type of fund as compared to other funds. Companies in life sciences may fit this profile, while quicker exits might come from Web 2.0 companies.

Likewise, later-stage investments potentially would generate returns after a shorter time period than early-stage funds, with a potentially higher IRR due to the timing of the returns, assuming equal performance.

multiples on returned capital— that is, more money back into the investor’s pocket. This example also shows that big returns during the first few years of a fund’s life can lead to misleadingly high TVPI and DPI multiples early in the fund. In this example, Fund B showed TVPI and DPI figures above 5.0x in the first year in the life of the fund, but the fund ultimately returned only 1.1x at termination.

The built-in bias of the performance metrics may have greater implications for certain types of funds, particularly funds that focus on certain stages or sectors. For instance, funds with investments concentrated in sectors

fund a: returns spread evenly over 10 years

fund b: big return early in fund

Cost: ($100m) Returns: $200m Gain: $100m

IRR = 14%

TVPI = 2.0x Cost: ($100m)Returns: $110m Gain: $10m IRR = 2191% TVPI = 1.1x 7.0 6.0 5.0 4.0 3.0 2.0 1.0 - 2400% 2000% 1600% 1200% 800% 400% 0% 12/31/2001 12/31/20 00 12/31/200212/31/200312/3 0/200 4 12/30/20 05 12/3 0/20 06 12/3 0/20 07 12/29/200 8 12/29/200 9 12/29/201012/29/2011 D P I & TVPI IR R IR R 7.0 6.0 5.0 4.0 3.0 2.0 1.0 - 2400% 2000% 1600% 1200% 800% 400% 0% 12/31/2001 12/31/2000 12/31/200212/31/200312/3 0/20 04 12/3 0/2005 12/3 0/20 06 12/3 0/20 07 12/2 9/20 08 12/29/20 09 12/29/201012/29/2011 D P I & TVPI

OVERVIEW OF INDUSTRY BENCHMARK SOURCES

The most commonly used industry benchmarks are published by Cambridge Associates and Thomson

Reuters Venture Economics.

Cambridge Associates is a consulting firm that provides advisory services to institutional investors and in doing so, has access to financial information for a large number of funds. Thomson Reuters (formerly Thomson Financial) publishes a range of financial news and information, such as Private Equity Week and the Venture Capital Journal. A third benchmarking source, Private Equity Intelligence (known as “Preqin”), creates benchmarks using its Performance Analyst database of fund financials. Preqin also provides access to separate databases of funds in the market and limited partner information.

These organizations are more different than they are similar, not only in their business structure, but in how they gather, analyze and report benchmarks. Specifically:

The benchmarks use different 1.

methodologies for data collection

The benchmarks use different 2.

data samples

The benchmarks provide 3.

different performance results

Let’s explore each one of these differences in more detail.

1. The benchmarks use different

methodologies for data collection

Performance metrics vary widely from one benchmarking source to another. One factor is the different methodologies for collecting data from the funds. Cambridge Associates collects financial information from its clients’ investments as well as by soliciting information from managers, which it aggregates into its database for calculating performance benchmarks. Thomson Reuters Venture Economics uses surveys sent to private equity and venture funds relying on self-reporting. These surveys are not audited, but the information collected reveals cash flow information. Both organizations collect this information on a confidential basis.

By contrast, Preqin collects data on fund performance based on public data sources, typically reports from pension funds and other institutions that must provide their financial performance reports as mandated by the U.S. Freedom of Information Act (FOIA) or similar legislation in foreign countries. These organizations report performance, rather than cash flows, which form the basis of calculations by Thomson Reuters and Cambridge Associates.

Because the data are gathered from public sources, Preqin publishes the performance metrics for specific funds and firms. That is, it does not keep the fund or firm name confidential for performance on individual funds. Cambridge Associates and Thomson Reuters aggregate fund performance information and do not identify fund or firm names. Preqin advertises that its data have less selection bias than samples collected via surveys or client investments because Preqin’s information would not omit better funds or worse-performing funds or be skewed upwards by institutional clients’ investment picks. However, some in the industry assume that since Preqin gathers data from funds subject to disclosure, these investors cannot access the best performing funds and therefore, Preqin’s results will be skewed downward.

2. The benchmarks use different

data samples

One reason why the performance benchmarks from each of the providers are so different is because they use different samples of funds for their calculations. Simply put, different samples of funds yield different benchmarks. The graph below compares the number of funds per vintage year from 1995 to 2007 for each of the benchmark providers.

While Cambridge Associates typically uses the largest sample size to calculate benchmarks for almost all of the vintage years during 1995 to 2007, it is questionable whether even its sample size of funds per vintage year is large enough to provide efficient and unbiased estimators of performance.4

In other words, does a summary statistic of performance based on these sample sizes reflect actual performance of venture funds in the market? In statistics, the amount of variance in the population must be known or estimated in order to determine the appropriate sample size. The performance of venture funds is known to have a large amount of variability because the dispersion of returns is large. And the greater the variation, the larger the sample size required for the sample’s metrics to be

statistically significant. Because there is not a precise estimate of variation of funds’ performance, it’s difficult to estimate an accurate sample size for

venture. However, it is known that these samples are small percentages of funds in the U.S. market. The graph below shows the sample size for each of the benchmark providers as a percentage of funds invested.

The available sample sizes seem small given the perceived variance in fund performance, but this is beyond the control of the organizations providing the benchmarks. In one survey of private equity firm CFOs, almost 40 percent of respondents stated that they did not send financial information to Thomson Reuters because there was “no reason to do so.”5 _{In fact,}

there is little incentive for funds to complete and return surveys of their performance, particularly if the fund is one of the best or one of the worst performing.

Sources: Cambridge Associates, Thomson Reuters, and Private Equity Intelligence. Cambridge Associates and Thomson Reuters data are as of December 31, 2007; Preqin data are as of various dates but are the most recent obtained. The number of funds in each vintage year is the number of active funds based on Thomson Reuters Fund Statistics Report. Cambridge Associates data were provided at no charge.

sample size as a percentage of active u.s. venture funds

6 14 18 26 23 13 9 17 29 33 33 28 28 26 29 29 21 22 31 32 24 15 15 16 15 16 13 14 14 14 12 13 18 18 21 27 27 27 15 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 35% 30% 25% 20% 15% 10% 5% 0% P ercentage of A ctive F unds Vintage Year

Thomson Reuters Preqin Cambridge Associates

Sources: Cambridge Associates, Thomson Reuters, and Private Equity Intelligence. Cambridge Associates and Thomson Reuters data are as of December 31, 2007; Preqin data are as of various dates but are the most recent obtained. Cambridge Associates data were provided at no charge.

number of u.s. venture funds sampled by vintage year

11 27 34 25 17 23 16 17 37 30 19 35 56 57 44 72 125 162 118 109 77 82 73 62 35 35 25 25 43 49 36 54 38 44 33 28 50 55 58 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 180 160 140 120 100 80 60 40 20 0 Number of funds Vintage Year

Thomson Reuters Preqin Cambridge Associates

3. The benchmarks provide different

performance results

While each benchmarking source purports to report on the performance of the industry, there is a large variation in performance metrics among the three providers. The bar graph above compares pooled IRR performance metrics between Cambridge Associates and Thomson Reuters. Preqin does not provide benchmark metrics over these time horizons; rather the data are provided by fund, firm or vintage year. Note that Cambridge Associates’ 10-year pooled IRR figure is almost double the same metric published by Thomson Reuters. Given that the pooled 10-year IRR metric is more stable than, for instance, a short-term one-year metric, the large difference in the long-term benchmark is surprising.

Large differences also remain in the three-year pooled IRR performance benchmarks.

Do the differences in aggregated performance indicate that these sources contain completely different collections of funds? An examination of the sampling distributions and the sample means, or averages, would provide a definitive answer to this question. However, the data for individual funds in Cambridge Associates and Thomson Reuters’s samples are not available to conduct these statistical tests. As a proxy, box and whisker plots help show the quartile ranges of funds and the best and worst performing funds. The chart below show the maximum, top quartile, median, lower quartile and minimum fund performance for vintage year 2000.6 _{The top quartile}

and lower quartile provide the top and bottom edges of the box; the median is the line in the middle; the minimum and maximum are dots connected by extended lines.

Sources: Cambridge Associates and Thomson Reuters. Note that Preqin does not provide cumulative benchmarks over specific time horizons. Pooled IRR is calculated based on cash flows of all funds regardless of vintage year during the specified time horizons. All data are as of December 31, 2007. Cambridge Associates data were provided at no charge.

comparison of investment horizon benchmarks

40% 35% 30% 25% 20% 15% 10% 5% 0% P ooled IR R (%)

Thomson Reuters Cambridge Associates

10-Year Venture IRR Performance

5-Year Venture IRR Performance

3-Year Venture IRR Performance

1-Year Venture IRR Performance

Sources: Cambridge Associates, Thomson Reuters, and Private Equity Intelligence. Cambridge Associates and Thomson Reuters data are as of December 31, 2007; Preqin data are as of various dates but are the most recent obtained. Cambridge Associates data were provided at no charge.

comparison of irr ranges for vintage 2000

IR R (%) 40% 20% 0% -20% -40% -60% -80% Thomson Reuters Cambridge Associates

Lower Quartile Minimum Median Maximum Upper Quartile

This simple analysis confirms the wide range of performance across venture capital funds—not only within samples, but also across different benchmarking sources. This disparity in performance between the best and worst funds is exceptionally large—particularly for Cambridge Associates’ funds in 2000—but the performance of these funds may be outliers compared to other funds in each sample. Nonetheless, the benchmarking sources show ranges of more than 10 percentage points between the upper quartile and lower quartile IRRs and almost 20 percentage points’ difference in Preqin’s sample. For vintage year 2000, the median fund performance is similar across the three benchmarking sources, although the IRR is positive according to Preqin and negative according to Cambridge Associates and Thomson Reuters.7

Many in the venture industry would argue that a fund’s performance must place it in the top quartile in order to achieve attractive, risk-adjusted returns over time. The box and whisker plots above show how high the performance can be for some of the funds in the top quartile (represented by the extended lines on top of the boxes). The graph below shows the variation in the IRR performance of the top quartile

fund in samples from Cambridge Associates, Thomson Reuters and Preqin. The difference in IRR metrics of the sources’ top quartile funds was narrow (2 percent) in 2001 and wide (12 percent) in 2005. The graph also shows that no one benchmark has an upper quartile fund performance that is consistently higher or lower than the other benchmarks; moreover, the benchmarks do not trend together.

OVERCOMING BENCHMARK LIMITATIONS

While benchmarks can provide a quick comparison of one investment to the performance of another in the same asset class, many LPs invest in venture capital and private equity to add diversification to their portfolio and to provide returns that are not correlated to public

Sources: Cambridge Associates, Thomson Reuters, and Private Equity Intelligence. Cambridge Associates and Thomson Reuters data are as of December 31, 2007; Preqin data are as of various dates but are the most recent obtained. Preqin does not provide IRR benchmarks for vintage 2007 funds. Cambridge Associates data were provided at no charge. Note: IRR performance during the first three years of a fund is typically considered not meaningful.

upper quartile comparison

2000 2001 2002 2003 2004 2005 2006 30% 25% 20% 15% 10% 5% 0% -5% P ooled IR R (%)

Thomson Reuters Preqin Cambridge Associates 10 10 11 11 11 14 16 12 12 21 24 5 -3 12 4 7 13 5 3 9

In the process of benchmarking performance, LPs must decide the objective of an investment. Which is it?

• Provide a return commensurate with the added risk and illiquidity of the investment • Provide more dollars back to the fund

• Outperform public investments by a certain margin

Determining the primary goal of the investment will help to guide LPs to find the appropriate benchmarking tool.

markets. These LPs often compare the IRR of the venture portion of their portfolio to the performance of public investments, with an expectation that the venture portion will return a certain level higher than public market investments. The logic of this assessment is based on risk and reward. Venture investing presents greater risks to an investor than investing in public markets, in part, because it is a long-term and relatively illiquid investment; likewise, investors expect greater returns from their venture investments.

The business of benchmarking venture capital funds has many complications simply because it is

hard to collect accurate financial data on private investments. It’s also difficult to report consistently on performance due to the metrics, the sample sizes and the collection methodologies.

With clear shortcomings and

inconsistencies in industry

benchmarks, how can investors assess the performance of their funds? Given the long-term nature of the investment and the lack of access to information on returns in the private market, accurately benchmarking venture capital remains elusive. While individual funds may have little incentive to contribute their financial

information to benchmarking

organizations, SVB Capital believes that the venture industry as a whole should have an incentive to create more credible and statistically reliable performance metrics. With more accurate benchmarks, the venture industry could assess more fully how funds are performing, especially as compared to other asset classes, and communicate these results with current and potential investors. Today it’s commonplace to study automotive industry benchmarks that yield meaningful insights as a basis for decisions. Tomorrow it’s possible we will be able to say the same about venture capital.

Recognizing the limitations of venture capital benchmarking for assessing performance, SVB Capital recommends supplementing benchmark analysis with other information. Consider the following:

• Gain a better understanding of portfolio companies and the return potential of the active portfolio. Annual meetings can typically be the place to obtain this information. It is widely known that one “home run” in a venture capital portfolio can move the fund into top-tier territory. Returns to the top-tier venture capital funds are typically driven by a few deals.

• Become part of the conversation. Those closest to the business of the fund have good information and instincts about current and future performance. Discuss the performance of the fund with the fund managers and learn the details of the companies in the portfolios that drive—or drag—performance.

• Look at the track record of individual venture investors, many of whom have made previous investments at other funds. The performance of past investments—including which sectors provided the returns—might help to inform expected performance.

TELL US WHAT YOU THINK

Send your comments and suggestions for topics to Bronwyn Bailey at [email protected].

Net Asset Value is the market value of the portfolio plus any cash held by the fund.

See Austin M. Long and Craig J. Nickels, “A Method for Comparing Private Market Internal Rates of Return to Public Market Index Returns.” Manuscript. The University of Texas System, August 28, 1995.

The IRR calculation assumes that distributed capital is reinvested at the same IRR over the life of the fund, when in fact, investors may not find similar investment opportunities for each distribution. See Oliver Gottschalg and Ludovic Phalippou, “The Truth about Private Equity Performance,” Harvard Business Review, December 2007 for this analysis. For a detailed discussion on the benefits and drawbacks of using IRR as a performance metric, see Paul Gompers and Josh Lerner, “Assessing the Performance of Private Equity Funds.” Manuscript. Harvard Business School, January 2003.

The game of darts can be used as an analogy for the quality of a sample statistic. The darts of an efficient and unbiased player would land clustered closely together on the bulls’ eye of the target. The darts of a less efficient player would land scattered around the dartboard, and the darts of a biased player would be tightly clustered outside the bulls’ eye.

Results from an informal survey conducted by Thomson Reuters (formerly Thomson Financial) presented at the Private Equity CFO Conference, July 2007. Respondents were attendees at the conference.

Vintage year 2000 was chosen due to the large number of funds in each sample, which would provide a more conservative estimate of variation in fund performance. This analysis is limited to IRR performance because Cambridge Associates does not publish quartile ranges for DPI and TVPI calculations.

This finding for vintage year 2000 does not support the notion that public institutions, Preqin’s data source, have problems accessing better performing funds.

1 2 3 4 5 6 7

most active venture investors

Source: Dow Jones VentureSource

u.s. venture investing activity

Source: Dow Jones VentureSource

fundraising by u.s.-based venture and lbo/

mezzanine firms

Source: Thomson Financial Venture Economics/National Venture Capital Association

venture investment by region, all

industries

Source: Dow Jones VentureOne

$ (MILLIONS)

Firm Name

Assets

Under Mgmt Num of Deals

Polaris Venture Partners $ 3,049 18 New Enterprise Associates 8,500 16 Draper Fisher Jurvetson 4,214 15

Atlas Venture 2,100 14

Mohr Davidow Ventures 1,400 13

Sequoia Capital 2,153 13

Accel Partners 4,000 12

Austin Ventures 3,000 11

Bessemer Venture Partners 2,000 11 Duff Ackerman & Goodrich 1,125 11

Ignition Partners 1,475 11

InterWest Partners 2,002 11 Kleiner Perkins Caufield & Byers 2,760 11 Highland Capital Partners 2,964 10 Q4 2007 US Region Num of Deals Num of Investing Firms Average

Per Deal Sum Inv.

Bay Area 187 267 $ 14.8 $ 2,499.5 Boston Area 77 127 13.4 967.2 New York Metro 48 81 9.2 447.6 San Diego Metro 24 66 19.0 426.7 Washington State 32 59 13.2 424.1 Research Triangle 18 40 20.1 361.7

Texas 35 31 10.6 356.4

Midwest 38 62 8.0 290.2

Potomac 32 56 8.8 278.8

Los Angeles Metro 30 36 8.2 230.9 Philadelphia 15 8 11.5 154.4 Southeast 18 22 9.1 151.0 Colorado 15 33 5.5 82.3 Oregon 4 7 10.1 43.3 $ (MILLIONS) $ (MILLIONS) 700 600 500 400 300 200 100 0 $10,000 $8,000 $6,000 $4,000 $2,000 $0 3Q 4Q 1Q 2Q 3Q 4Q 2006 2007 Deals Amount Invested ($M) $ (BILLIONS) Venture Capital Buyout and Mezzanine

$80 $70 $60 $50 $40 $30 $20 $10 $0 3Q 4Q 1Q 2Q 3Q 4Q 2006 2007 5 31 6 68 60 6 9 56 9 61 68 12

cumulative irr performance (%) by stage

(u.s.)

Source: Thomson Financial Venture Economics/National Venture Capital Association, data as of December 31, 2007

Fund Type Num of Funds Cap Wtd Avg Pooled Avg Upper Quartile Median Lower Quartile Early/Seed VC 591 9.2 19.3 16.0 3.3 (4.9) Seed Stage VC 66 5.5 9.4 13.3 4.4 (1.2) Early Stage VC 525 9.4 20.2 16.3 3.3 (5.2) Balanced VC 452 10.0 14.1 15.7 5.6 (0.3) Later Stage VC 199 6.4 13.7 17.0 7.4 (0.6) All Venture 1,242 8.9 15.8 16.0 4.9 (2.1) Small Buyouts 178 8.7 15.4 17.3 7.4 (0.3) Med Buyouts 112 12.3 17.6 22.2 9.3 (0.1) Large Buyouts 92 11.3 12.9 19.5 7.3 0.1 Mega Buyouts 124 8.1 12.4 18.0 8.7 0.4 All Buyouts 506 8.9 13.3 18.4 8.0 (0.1) Mezzanine 72 6.9 8.7 12.6 7.5 1.5 Buyouts and Other

PE 668 9.3 12.7 18.0 8.0 0.1 All Priv Equity 1,915 9.2 14.1 16.7 6.3 (1.4)

us. venture liquidity events by industry

Source: Dow Jones VentureOne

2005 2006 2007

Industry IPO M&A IPO M&A IPO M&A

Biopharmaceuticals 14 40 20 29 17 25 Healthcare Services 0 11 0 6 2 8 Medical Devices 7 14 6 23 9 18

Medical IS 1 18 2 16 3 9

Comm. And Networking 3 40 5 49 10 34 Elect. & Computer Hdw. 1 12 1 14 4 9 Information Services 0 43 5 48 6 65 Semiconductors 4 17 2 18 6 18

Software 3 148 7 157 8 144

Other 10 70 8 60 10 68

TOTAL 43 4413 556 420 75 398

u.s. ipo vs m&a transactions for

venture-backed companies

Source: Dow Jones VentureSource

irr performance (%) by vintage year (u.s.)

Source: Thomson Financial Venture Economics/National Venture Capital Association; data as of December 31, 2007

Vintage Year Num of Funds Cap Wtd Avg Pooled Avg Upper Quartile Median Lower Quartile 1996 35 59.6 83.6 113.9 31.0 1.5 1997 62 46.3 49.6 59.7 20.2 (0.8) 1998 77 24.3 19.2 11.9 2.0 (4.0) 1999 109 (7.1) (5.8) 0.9 (6.9) (15.0) 2000 125 0.8 2.1 3.0 (2.5) (7.9) 2001 56 4.3 5.8 11.3 2.6 (3.5) 2002 19 1.0 2.7 3.6 (0.6) (2.4) 2003 16 8.0 8.0 15.7 0.9 (1.6) 2004 23 4.0 7.3 11.8 (1.0) (7.8) 2005 17 1.8 5.2 21.0 3.2 (2.3) 2006 25 (2.7) (0.5) 4.7 (9.6) (20.2) 2007 11 (31.0) (0.1) (21.7) (26.1) (60.2) Number of IPOs Number of M&As 160 140 120 100 80 60 40 20 0 3Q 4Q 1Q 2Q 3Q 4Q 2006 2007 9 120 18 83 13 105 94 24 100 11 25 110

price change

venture capital barometer

Source: Fenwick & West L.L.P.

The direction of price changes for 103 San Francisco Bay Area companies receiving financing, compared to their previous rounds.

Average per share % price change from previous round of Silicon Valley companies receiving VC investment in the applicable quarter. Complete report available at http://www.fenwick.com/vctrends.htm

Source: Fenwick & West L.L.P.

u.s. venture-backed m&a activity

Source: Dow Jones VentureOne

$ (BILLIONS) 8 Number Deals Amount Paid ($B) 140 120 100 80 60 40 20 0 8 10 9 12 16 122 83 105 94 100 110 3Q 4Q 1Q 2Q 3Q 4Q 2006 2007 67 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% Down Flat Up 11 24 67 22 8 81 11 12 79 9 7 79 14 ₉ 69 22

Net Results of Rounds

90% 80% 70% 60% 50% 40% 30% 20% 10% 0% 3Q 4Q 1Q 2Q 3Q 4Q 2006 2007 69 49 75 74 79 55 $18 $16 $14 $12 $10 $8 $6 $4 $2 0 9 3Q 4Q 1Q 2Q 3Q 4Q 2006 2007

*This update is for informational purposes only and is not a solicitation or recommendation that any particular investor should invest in any particular industry, security, or fund.

This material, including without limitation to the statistical information herein, is provided for informational purposes only. The material is based in part on information from third-party sources that we believe to be reliable, but which have not been independently verified by us and for this reason we do not represent that the information is accurate or complete. The information should not be viewed as tax, investment, legal or other advice nor is it to be relied on in making an investment or other decision. You should obtain relevant and specific professional advice before making any investment decision. Nothing relating to the material should be construed as a solicitation, offer or recommendation to acquire or dispose of any investment or to engage in any other transaction.

©2008 SVB Financial Group. ® All rights reserved. Member Federal Reserve. SVB, SVB> and SVB>Find a way are all trademarks of SVB Financial Group. SVB Capital is a non-bank member of SVB Financial Group. Products and services offered by SVB Capital are not insured by the FDIC or any other Federal Government Agency and are not guaranteed by Silicon Valley Bank or its affiliates. Rev. 06-02-08.