The Multiple Mystery: At what P/E should the market trade?

Transcription

1 October 1, 2009 United States: Portfolio Strategy US Equity Views The Multiple Mystery: At what P/E should the market trade? Investor focus has shifted from earnings to valuation. We are now most often asked at what P/E should the market trade. We offer four approaches: (1) history of multiple expansions from bear market lows suggests 14.1x; (2) our top-down P/E model suggests 13.7x; (3) our DDM-derived 2009 price-target of 1060 implies 13.1x; and (4) the Fed Model implies 16.4x. History, Macro Model, DDM & Fed Model point to P/E of x Historically, P/Es expand by 34% from a bear market trough. Since March, the S&P 500 multiple has risen by 30%, on track with history. A further P/E expansion of 4% would lead to a year-end 2009 P/E of 14.1x. Our macro model incorporates inflation and the output gap and suggests a year-end P/E of 13.7x. Other valuation techniques, such as DDM and the Fed Model, imply year-end 2009 multiples of 13.1x and 16.4x, respectively. Conclusion: Target P/E elusive, but expect below-average multiple Historical pattern of bear market recoveries shows P/E multiples plateau after expanding for 10 months. Our macro model also suggests multiples will contract slightly given the anticipated slow economic recovery in Although this report offers various approaches and identifies a range of target multiples, our best estimate is roughly 13x-14x. We disagree with the increasingly consensus view that the market P/E will soon revert to the long-term average of 15x (SPX of 1200). We believe higher E not higher P/E is required for the market to sustain levels above our 1060 target. David J. Kostin (212) david.kostin@gs.com Goldman, Sachs & Co. Nicole Fox (212) nicole.fox@gs.com Goldman, Sachs & Co. Caesar Maasry (212) caesar.maasry@gs.com Goldman, Sachs & Co. Amanda Sneider (212) amanda.sneider@gs.com Goldman, Sachs & Co. Four approaches to year-end 2009 P/E suggests range of 13.1x 16.4x 2009 Year-End Upside from Implied 2009 Methodology Fair Value Current Price Year-End P/E US Portfolio Strategy DDM % 13.1x Top-Down P/E Regression % 13.7x Historical P/E Expansion from Bear Market Bottom % 14.1x Fed model % 16.4x US Treasury 10 Year Yield BBB Corporate 10 Year Yield Year TIPS Yield Note: Implied P/E of DDM and Fed Model based on GS top-down 2010 EPS estimate of $81. Source: Compustat, I/B/E/S, FactSet, Goldman Sachs Global ECS Research The Goldman Sachs Group, Inc. does and seeks to do business with companies covered in its research reports. As a result, investors should be aware that the firm may have a conflict of interest that could affect the objectivity of this report. Investors should consider this report as only a single factor in making their investment decision. For Reg AC certification, see the end of the text. Other important disclosures follow the Reg AC certification, or go to Analysts employed by non-us affiliates are not registered/qualified as research analysts with FINRA in the U.S. The Goldman Sachs Group, Inc. Goldman Sachs Global Economics, Commodities and Strategy Research Goldman Sachs Global Economics, Commodities and Strategy Research 1

2 The Multiple Mystery: Targeting the right P/E ratio P/E multiples will likely remain below average in 2010 Our top-down regression and DDM models cannot precisely pinpoint a correct P/E multiple. However, the current weak macroeconomic backdrop as well as the historical trading pattern of bear market recoveries suggest P/E multiples will stay below average in Our forecast that multiples will remain below average represents an out-of-consensus view based on recent conversations with a wide range of portfolio managers. Many investors expect S&P 500 will reach 1200 by year-end. However, we believe it will take a higher E not higher P/E for the market to sustain levels above our current 1060 target. The debate has shifted from EPS to valuation. During the past 18 months, we have focused our research primarily on the earnings power of the market and the trajectory of S&P 500 EPS. This emphasis was appropriate given forward bottom-up consensus EPS estimates for the S&P 500 plunged by 37% from $98 in September 2008 to $62 in May During the past five months, however, EPS estimates have edged up slightly, currently hovering around $68. Given the recent stability of earnings estimates, investor focus has shifted from earnings to valuation. (1) History suggests 14.1x, or 34% expansion from the trough Historical analysis reveals a consistent pattern of bear market recoveries: first multiple expansion, then earnings growth. The market multiple tends to expand by 34% during the first 10 months following a bear market trough but then remains flat during the subsequent 12 month period. Since the March low, multiples have risen by 30%, on track with historical precedent, to 13.6x currently. A further 4% rise would lead to a 14.1x P/E. (2) Top-Down P/E Regression suggests 13.7x Macro model of inflation and the output gap suggest multiples will remain near current levels into Both variables appear negatively correlated with P/E (output gap is positive when GDP is below potential). We acknowledge that these two variables may not have a completely linear relationship with P/E multiples. For example, a deflationary environment probably would not lead to higher P/E (despite the negative correlation between P/E and inflation). However, our 2010 headline CPI forecast equals 1.2%, a rate within our historical regression sample set. The standard error of our model is 3.4 points, indicating the regression results should not be interpreted with a false sense of accuracy. (3) Comparison with DDM (13.1x) and Fed Model (16.4x) Independent valuation models provide varying implications for year-end P/E multiples. Our 2009 year-end price target of 1060 stems from our Dividend Discount Model (DDM) and implies a year-end 2009 P/E multiple of 13.1x, based on our 2010 pre-provision and write-down EPS estimate of $81. This result represents the lowest implied P/E of our various approaches. The Fed Model suggests P/E multiples may expand significantly in the near term, rising to 16.4x by year-end This model suggests equities are undervalued relative to bonds. The usefulness of this methodology in the current environment is limited given the unprecedented government actions that have kept interest rates unusually low. P/E is hard to target; using 15x long-term average is too simplistic Most investors agree with a $75-$80 EPS estimate for Bullish investors argue for 15x P/E multiple based on long-term history, and a year-end 2009 target of Our year-end 2009 price target remains We recognize the possibility the S&P 500 could rally sharply in 4Q boosted by better than expected earnings results and increased risk appetite. However, we reject the argument of many investors that re-valuation alone will propel the market higher through multiple expansion back to the long-term average. Goldman Sachs Global Economics, Commodities and Strategy Research 2

3 (1) History suggests market P/E may expand to 14.1x by year-end Historically, the market multiple has expanded by an average of 34% during the first 10 months following the bear market troughs in 1982, 1987, 1990, and 2002 (10 months posttrough in the current episode would be year-end 2009). After the initial expansion period, P/E multiples typically remained flat or contracted slightly over the subsequent 12 months. Multiple compression does not necessarily mean the S&P 500 must trade lower. In fact, history suggests that the market index level continues to rise even as the multiple stays flat or even declines. Multiple expansion typically drives market returns during the initial 10-month period after a bear market trough. Subsequent equity market returns are primarily earnings-driven as rising forward EPS estimates drive share prices higher. Since the March low (7 months ago), the S&P 500 multiple has expanded by 30%, rising to 13.6x. History suggests the multiple may expand another 4% by year-end, reaching 14.1x. Based on our 2010 pre-provision & write-down EPS estimate of $81, a 14.1x multiple implies a year-end 2009 level of 1140, or 8% above our current price target of History suggests upside potential exists to our forecast. However, the typical historical multiple expansion is arguably not warranted in the current environment given expectations of a slower-than-average recovery into next year. Exhibit 1: P/Es tend to expand for 10 months and then remain relatively flat as of September 28, 2009 P/E Multiple Current Cycle (2009, LHS) P/E 10.5x SPX Trough Current top-down NTM P/E 10 months of multiple expansion (Median expansion = 34%) (Current expansion = 30%) (12) (10) (8) (6) (4) (2) x Months from S&P 500 Trough Typical P/E expansion 10-mo from SPX trough 14.1x Average Recovery ( , RHS indexed) Flat multiple after month 10 (median change = -2%) , S&P 500 Price 1,400 1,300 1,200 1,100 1, Current Cycle (2009, LHS) SPX Trough Average Recovery ( , RHS indexed) SPX continues to rally despite flat/down multiple Price (12) (10) (8) (6) (4) (2) S&P 500 EPS Current Cycle (2009, LHS) SPX Trough Average Recovery ( , RHS indexed) After 10 months of P/E expansion, rising EPS drives market higher, EPS (12) (10) (8) (6) (4) (2) Months from S&P 500 Trough Months from S&P 500 Trough Source: Compustat, I/B/E/S, and Goldman Sachs Global ECS Research. Goldman Sachs Global Economics, Commodities and Strategy Research 3

4 Fitting P/E expansion into our Pop, Stall, and Sustained Rally framework as a guide for multiple expansion into year-end. Historically, the market multiple tends to expand during the Pop (+16%) and Stall (+9%) phases, but EPS explains the market return during the Sustained Rally phase. Given multiples have already expanded by 3% during the Sustained Rally phase, we believe P/E expansion is limited between now and year-end. This view is also consistent with our top-down regression that suggests a 13.7x year-end 2009 multiple, up from 13.6x currently. Exhibit 2: Multiples expand during the Pop and Stall but remain relatively flat during the Sustained Rally Bear Market Decline Bear Market Recovery "Pop" "Stall" "Sustained Rally" 1-month after trough months 2-4 months 5-10 Price EPS Multiple Price EPS Multiple Price EPS Multiple Price EPS Multiple Peak Trough Decline Change Change Rebound Change Change Rebound Change Change Rebound Change Change Nov-80 Aug-82 (27)% 5 % (31)% 18 % (2)% 21 % 15 % (4)% 21 % 17 % 15 % 1% Aug-87 Dec-87 (34) 8 (39) 14 (2) (5) 6 12 (6) Jul-90 Oct-90 (20) (3) (18) 6 (4) (6) Mar-00 Oct-02 (49) (10) (44) 15 (1) 17 (7) (1) (6) Historical Average (32)% 0 % (33)% 13 % (3)% 16 % 6% (2)% 9% 11 % 8 % 3% Oct-07 Mar-09 * (57)% (38)% (31)% 27 % (4)% 31 % 3% 4% (0)% 20 % 16 % 3% Multiples tend to rise during the "Pop" and "Stall" Phases of an equity market recovery. But during the "Sustained Rally" phase, multiples tend to stay flat, while changes in forward EPS views explain a larger portion of the return. Source: Compustat, IB/E/S, and Goldman Sachs Global ECS Research. Consistent with our top-down P/E model, history suggests P/E multiples may contract by 2% in P/E multiples tend to expand by 34% during the first 10 months after a bear market bottom, and decline by 2% during the subsequent year. The notable historical outlier is the 1987 bear market, during which there was no economic recession. In that episode, the market recovery was fueled in the reverse order of other recoveries: EPS growth explained the first 10 months of recovery and P/E expansion drove the market during the following year. Given that the bear market occurred during an economic recession and that P/E expansion has fueled the initial recovery, we believe the historical precedent set by 1982, 1990, and 2002 is more relevant. Those experiences suggest P/E multiples are likely to decline next year. Exhibit 3: Multiples tend to remain flat during the year following a bear market bottom P/E Multiple Expansion from Bear Market Bottom 10 Months post Trough.One Year Later (22 Months Post Trough) Price EPS Multiple Price EPS Multiple Peak Trough Rebound Change Change Rebound Change Change Nov-80 Aug % 7 % 48 % (6)% 15 % (18)% Aug-87 Dec Jul-90 Oct (8) Mar-00 Oct (0) (6) Historical Median 28 % 3 % 34 % 9% 11% (2)% Oct-07 Mar-09 * 57 % 21 % 30 % Multiples tend to remain flat or contract in the 2 nd year following a bear-market bottom. * Current episode data from Mar 9 Sep 21 (not full 10 months), using GS top-down EPS estimates Source: Compustat, I/B/E/S, and Goldman Sachs Global ECS Research. Goldman Sachs Global Economics, Commodities and Strategy Research 4

5 (2) Top-down regression implies P/E of 13.7x We find that the Output Gap and CPI headline inflation are the two macro variables that best explain the market s P/E ratio. Both variables appear negatively correlated with P/E (output gap is positive when GDP is below potential). Our US Economics team forecasts relatively modest GDP growth next year (+2%), implying that the output gap will remain high, but inflation will remain low in their view (headline CPI +1.2%). The persistent high output gap suggests P/Es will remain below average in the near term, but the low inflation environment is generally supportive of a higher market P/E. Using our current macroeconomic forecasts, our top-down P/E model suggests the S&P 500 will trade around 13.7x at year-end 2009 (please see Appendix for regression details). Using our 2010 operating EPS estimate of $75, a 13.7x multiple implies a year-end 2009 target price of 1030, and using our pre-provision & write-down estimate of $81, the model implies a target price of Our current price target of 1060, which is derived from a separate process using our Dividend Discount Model, sits in the middle of the two goal posts (1030 and 1110) implied by our top-down P/E model. Exhibit 4: Our top-down P/E model suggests the market will trade at 13.7x forward earnings at year-end 35x Top-Down P/E Regression S&P x Top-Down P/E 25x (Out of Sample) 20x 15x 10x 5x Avg P/E = 15.1 S&P 500 Estimated Top-Down P/E 28-Sep Top-Down P/E = 13.6x Year-End 2009 Estimated Top-Down P/E = 13.7x 0x R 2 = 0.65 Dec-55 Dec-60 Dec-65 Dec-70 Dec-75 Dec-80 Dec-85 Dec-90 Dec-95 Dec-00 Dec-05 Dec-10 Dec-15 Dec-20 Source: Compustat and Goldman Sachs Global ECS Research. Exhibit 5: Higher GDP growth would lead to a higher P/E multiple, but higher inflation generally lowers P/Es 2010 Headline CPI Sensitivity of Average 2010 P/E Multiple 2010 Real GDP Growth 0.0% 0.5% 1.0% 1.5% 2.0% 2.5% 3.0% 3.5% 4.0% 4.5% 5.0% 3.0 % 10.1x 10.4x 10.6x 10.8x 11.1x 11.3x 11.6x 11.8x 12.0x 12.3x 12.5x 2.5 % 10.7x 10.9x 11.2x 11.4x 11.6x 11.9x 12.1x 12.4x 12.6x 12.8x 13.1x 2.0 % 11.2x 11.5x 11.7x 12.0x 12.2x 12.4x 12.7x 12.9x 13.2x 13.4x 13.6x 1.5 % 11.8x 12.1x 12.3x 12.5x 12.8x 13.0x 13.3x 13.5x 13.7x 14.0x 14.2x 1.0 % 12.4x 12.6x 12.9x 13.1x 13.3x 13.6x 13.8x 14.1x 14.3x 14.5x 14.8x 0.5 % 13.0x 13.2x 13.4x 13.7x 13.9x 14.1x 14.4x 14.6x 14.9x 15.1x 15.3x 0.0 % 13.5x 13.8x 14.0x 14.2x 14.5x 14.7x 15.0x 15.2x 15.4x 15.7x 15.9x (0.5)% 14.1x 14.3x 14.6x 14.8x 15.0x 15.3x 15.5x 15.8x 16.0x 16.2x 16.5x (1.0)% 14.7x 14.9x 15.1x 15.4x 15.6x 15.9x 16.1x 16.3x 16.6x 16.8x 17.0x (1.5)% 15.2x 15.5x 15.7x 15.9x 16.2x 16.4x 16.7x 16.9x 17.1x 17.4x 17.6x (2.0)% 15.8x 16.0x 16.3x 16.5x 16.7x 17.0x 17.2x 17.5x 17.7x 17.9x 18.2x Source: Compustat and Goldman Sachs Global ECS Research. Goldman Sachs Global Economics, Commodities and Strategy Research 5

6 Sensitivity analysis of regression model to GDP and CPI A sensitivity analysis of our model shows that each incremental percentage point of real GDP growth next year is associated with a market multiple ½ of a point higher. A higher GDP growth estimate would likely translate into higher earnings growth, suggesting the stock market return would be amplified from both a higher P/E ratio and a higher EPS level. Although our model uses the Output Gap, we prefer showing sensitivity to real GDP growth, as most investors forecast this metric vs. Output Gap (see Appendix for details). The 2010 Goldman Sachs Economics forecasts of 2.0% real GDP growth and 1.2% CPI growth suggest a 13.3x multiple. Our model implies that even much stronger GDP growth of, for example, 4% next year would lift the expected P/E by just 1 point. This result is due to the current, wide output gap that will take several years of above-trend growth to close. P/E ratio may contract during 2010 Our top-down model suggests that P/E ratios will contract to 13.3x during 2010 from 13.7x at 2009 year-end, given our forecast of a muted recovery. We expect the output gap to remain relatively flat next year, which does not affect the P/E estimate of our model, but headline inflation may rise modestly, causing the estimated P/E to contract. Our model implies a P/E contraction of 3% next year (13.7x to 13.3x), which is consistent with past bear market recoveries (see discussion on page 4). Historically, P/E multiples tend to expand for 10 months following a bear market bottom and then remain relatively flat, which is the general trajectory predicted by our top-down P/E model. Exhibit 6: Macro regression suggests P/E of 13.7x by year-end, and 13.3x during x 22x S&P 500 Estimated Top-Down P/E 20x 18x 16x Avg P/E = x 12x 10x Year-End 2009 P/E = 13.7x Avg P/E P/E = 13.3x Dec-85 Dec-87 Dec-89 Dec-91 Dec-93 Dec-95 Dec-97 Dec-99 Dec-01 Dec-03 Dec-05 Dec-07 Dec-09 Dec-11 Dec-13 Source: Compustat, I/B/E/S, Goldman Sachs Global ECS Research. (3) Dividend Discount Model (DDM) implies P/E of 13.1x Our 2009 year-end price target of 1060 stems from our Dividend Discount Model (DDM) and implies a year-end 2009 P/E multiple of 13.1x, based on our 2010 preprovision and write-down EPS estimate of $81 (see Exhibit 7). This methodology does not use a P/E forecast as an input, and is based on our dividend estimates, cost of equity assumption (8.5%), and long-term earnings growth rate (6.7%). Interestingly, the implied P/E of our DDM valuation is similar to our top-down P/E model estimate, which is based on independent macro variables. Goldman Sachs Global Economics, Commodities and Strategy Research 6

7 Exhibit 7: Goldman Sachs Top-Down Dividend Discount Model (DDM), as of September 25, S&P 500 level Path of S&P 500 fair value based on DDM estimates and assuming S&P 500 appreciates at the rate of dividend growth Current SPX (1044) Current Top-down Fair Value 2009 Year-end Top-down Fair Value implies 13.1x P/E Goldman Sachs Top-down DDM 250 Current 2009E 2010E 2011E 2012E 2013E 2014E 2015E 2016E 2017E 2018E S&P 500 Top-Down Dividend Discount Model 2008A 2009E 2010E 2011E 2012E 2013E 2014E 2015E 2016E 2017E 2018E 2019E Top-down Forecast Dividends $28.39 $22.10 $22.03 $22.80 $24.81 $26.74 $28.39 $29.84 $31.36 $32.96 $34.64 $36.40 Annual Dividend Growth (%) 2.4 (22.1) (0.3) Payout Ratio (%) EPS ($) Annual EPS Growth (%) (40.0) Cost of Equity 8.5% Years from cash flow Discount factor PV of future dividend $7 $20 $19 $19 $19 $18 $18 $17 $17 $16 $16 Assumptions Assumed LT EPS growth rate 6.7% Risk-free rate (a) 4.5% Equity risk premium 4.00% Cost of Equity (risk free rate + ERP) 8.5% Calculation of DCF value Terminal year multiple (b) 55.6 x Terminal year value 2022 PV of terminal year value 875 PV of dividends years PV of terminal year value + PV of dividends 1045 S&P 500 DDM Fair Value 1040 Current S&P 500 Price: 1044 Premium / (Discount) to Fair Value 0% 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% Decomposition of S&P 500 fair value Sensitivity Analysis Present Value of Dividends ( ) Present Value of Dividends (2019+) % 84% % Long Term EPS Growth Terminal Multiple Equity Risk Premium Risk Free Rate Premium / Premium / Premium / Premium / Input (Discount) Input (Discount) Input (Discount) Input (Discount) Value to Fair Value Value to Fair Value Value to Fair Value Value to Fair Value 0% 0% 0% 0% 6.25% 17% 45 x 16% 4.50% 21% 4.00% (38)% (16) (2) 60 (6) 3.75 (16) (17) 65 (14) 3.50 (38) (a) Current 10-year Treasury yield equals 3.2%. (b) Terminal multiple calculated as 1 / (cost of equity long term EPS growth rate). Alternatively, terminal multiple equals the inverse of the normalized dividend yield. Source: Compustat, FactSet, and Goldman Sachs Global ECS Research. Goldman Sachs Global Economics, Commodities and Strategy Research 7

8 (4) Fed Model implies P/E of 16.4x The Fed Model suggests P/E multiples may expand significantly in the near term, rising to 16.4x by year-end At its core, the Fed Model compares the S&P 500 earnings yield (inverse of P/E using NTM earnings) with the 10-year Treasury yield. The rationale for the model is that investors ultimately choose whether to invest in equities or bonds, and that the yield differential of the two assets should be roughly comparable over time. We calculate the S&P 500 upside or downside from fair value by assuming the spread between the 10-year Treasury yield and the earnings yield reverts to the 10-year trailing average spread with a price change by the S&P 500 that closes half the gap. We base our Fed Model valuation on an average of three versions of the model: the traditional model comparing the S&P 500 earnings yield to 10-year Treasury yield, one which substitutes 10-year BBB corporate bond yields for US Treasuries, and a third that incorporates 10-year Treasury Inflation Protected Securities (TIPS). In the BBB version, investors choose whether to invest in the equity or credit of a corporation. The yield differential between the S&P 500 and the BBB corporate bonds is typically narrower than the spread between S&P 500 equity yield and US Treasuries. Skeptics of the traditional Fed Model would also contend that Treasury yields are nominal while earnings yields are real. Our TIPS version seeks to settle this argument by using the real interest rate. Currently, all three versions of the Fed Model point to similar equity valuation, suggesting the 2009 year-end fair value ranges from 1320 to The average of the Fed Models using our top-down pre-provision and write-down earnings estimates suggests a year-end 2009 fair value of 1330, or 25% above the current S&P 500 price. We assume the yield spread closes by half through the price appreciation of the S&P 500, and the other half closes through changes in bond yields. The Fed Model suggests P/E multiples may expand significantly in the near term, rising to 16.4x by year-end 2009 (see Exhibit 8). This model suggests that equities are undervalued relative to bonds. In our view, this methodology is flawed given the current environment of unprecedented government actions to keep interest rates low. Although we do not expect rates to rise in the short term, the yield gap between bonds and equities may close by a larger move in bond prices rather than equity prices in coming years. Exhibit 8: Fed Model implies P/E of 16.4x, as of September 25, 2009 Top-Down 10-Yr / BBB Yield Spread Implied 2009 Year-End Methodology EPS Yield Yield Current 10yr Avg Price P/E Fed model x US Treasury 10 Year Yield 7.4 % 3.3 % (412)bp (154)bp BBB Corporate 10 Year Yield (207) Year TIPs Yield (589) (361) Note: Top-Down EPS Yield reflects NTM pre-provision & writedown estimate of $78. Implied 2009 year-end P/E reflects 2010 pre-provision and writedown EPS of $81. Yield spread reflects bond yield less earnings yield. Source: Compustat, I/B/E/S, Goldman Sachs Global ECS Research. Goldman Sachs Global Economics, Commodities and Strategy Research 8

9 Exhibit 9: Fed Model implies P/E of 16.4x, as of September 25, 2009 Yield Spread (10 yr yield - S&P 500 E/P) 600 bp 400 bp 200 bp 0 bp (200) bp (400) bp (600) bp 10-year Treasury yield less S&P 500 earnings yield Bonds Rolling 10-yr average Goldman Sachs Top-down (-412 bp) Yield Spread (BBB yield - S&P 500 E/P) 500 bp 400 bp 300 bp 200 bp 100 bp 0 bp (100) bp (200) bp BBB Corp yield less S&P 500 earnings yield BBB Goldman Sachs Top-down (-207 bp) Rolling 10-yr average (800) bp (300) bp bp Yield Spread (TIPS yield - S&P 500 E/P) 0 bp (200) bp (400) bp (600) bp (800) bp year TIPS yield less S&P 500 earnings yield Rolling 10-yr average TIPS Goldman Sachs Top-down (-589 bp) P/E Implied by the Fed Model Implied 2009 Year-End Methodology Price P/E Fed model x US Treasury 10 Year Yield BBB Corporate 10 Year Yield Year TIPs Yield Source: Compustat, I/B/E/S, Goldman Sachs Global ECS Research. Goldman Sachs Global Economics, Commodities and Strategy Research 9

10 Appendix Top-Down P/E regression details Our top-down P/E model incorporates the output gap and headline CPI to forecast P/E multiples. We find these indicators fit from both a statistical perspective and a theoretical standpoint. P/E multiples represent a simplistic discounting model. The two main inputs of such models are a growth rate and a discount rate. We believe the output gap captures the growth view and CPI the discounting metric. The output gap provides a better fit than GDP growth. We find that the time series of GDP growth is much more volatile than the output gap and results in a less accurate forecasting model for P/E. We show GDP growth in our P/E sensitivity analysis in Exhibit 5 but investors may easily convert a view on GDP growth into a view on the output gap, given the CBO s estimate of potential GDP. Currently, we assume 1.2% headline CPI and a 6.8% output gap in our regression, resulting in an average P/E for 2010 of 13.3x. Exhibit 10: Top-down macro model incorporates the Output Gap and CPI, as of September 25, x 30x Top-Down P/E Model S&P 500 Top-Down P/E 25x (Out of Sample) 20x 15x 10x 5x 0x Dec-55 R 2 = 0.65 Dec-60 Dec-65 Dec-70 Dec-75 Dec-80 Dec-85 S&P 500 Estimated Top-Down P/E Dec-90 Dec-95 Dec-00 Dec-05 Average 2010 Estimated Top-Down P/E = 13.3x P/E = ( x Output Gap ) + (-1.13 x CPI ) Note: Output Gap calculated using CBO estimates of potential GDP and Goldman Sachs US Economics GDP forecasts. CPI calculated as year-over-year change, based on Goldman Sachs US Economics forecasts. Source: Compustat, I/B/E/S, CBO, Goldman Sachs Global ECS Research. Dec-10 Dec-15 Dec-20 Exhibit 11: Output gap is negatively correlated with P/E Exhibit 12: Output gap to remain wide in 2010 NTM P/E 30x 25x 20x 15x 10x 5x 0x Dec-50 = Dec-55 Dec-60 Dec-65 Output Gap Dec-70 Dec-75 Dec-80 NTM P/E Dec-85 Dec-90 Dec estimate Dec-00 Dec-05 Dec-10 Dec % 8% 6% 4% 2% 0% (2)% (4)% (6)% (8)% Output Gap (%) GDP ($ trillions) Dec-02 Dec-03 Dec-04 Dec-05 Dec-06 Potential GDP Dec-07 Nominal GDP Dec-08 Dec-09 Dec-10 Output Gap 2010 GDP growth (2.3% Nominal, 2.0% Real) leaves wide Output Gap Dec-11 Dec-12 Source: Compustat, I/B/E/S, CBO, Goldman Sachs Global ECS Research. Source: CBO, Goldman Sachs Global ECS Research. Goldman Sachs Global Economics, Commodities and Strategy Research 10

11 Comparing macroeconomic indicators with P/E multiples on a one-to-one basis, most indicators point to higher P/E multiples. We compare the distribution of P/E multiples with the time series of major macro variables and find that four of the six macro indicators we study suggest that P/E multiples may continue to expand. For example, real GDP growth of 2.0% (the Goldman Sachs Economics 2010 forecast) is generally associated with a P/E of 16.3x (see Exhibit 13). Similarly, headline CPI growth of 1.0%-1.5% generally coincides with a P/E multiple of 16.6x. However, this simplistic analysis does not take into account the fact that GDP is growing from a depressed level. Our top-down model uses the output gap as an input, and in our view is better able to model the business cycle, which is relevant to P/E ratios. Furthermore, earnings growth of 18% (our 2010 forecast) has historically been accompanied by a P/E of 13.7x, in-line with our top-down P/E model. Exhibit 13: Distribution of P/E multiples across various macroeconomic indicators Earnings Growth Real GDP Growth Core CPI % of Avg % of Avg % of Avg Min Max Time P/E Min Max Time P/E Min Max Time P/E (5) or lower (5) or lower (2.0) (5) (3) (5) (3) (3) (1) (3) (1) or higher or higher or higher When EPS growth is between 17% and 19%, the average P/E multiple is 13.7x Headline CPI 10-Year Bond Yield Output Gap % of Avg % of Avg % of Avg Min Max Time P/E Min Max Time P/E Min Max Time P/E (5.0) (4) or lower (4.0) (3.0) (3.0) (2.0) (2.0) (1.0) (1.0) or higher or higher or higher Note: Boxed values represent current 2010 Goldman Sachs forecasts. P/E data uses realized forward earnings between 1928 and 1976 and estimated forward earnings through GDP data from 1950, Core CPI from 1958, Headline CPI from 1950, 10-year Bonds from 1962, and the Output Gap from Source: Compustat, I/B/E/S, and Goldman Sachs Global ECS Research. Goldman Sachs Global Economics, Commodities and Strategy Research 11

12 Exhibit 14: Bottom Line: Our key forecasts for the S&P 500 index, recommended sector positioning and top trade ideas as of September 30, 2009 S&P 500 Index Forecasts 2009 year-end price target 1,060 0% above current price level of 1057 S&P 500 EPS estimates 2007A 2008A 2009E 2010E GS Portfolio Strategy top-down $83 $50 $52 $75 Growth (% year/year) (6) (40) 5 45 Recommended Sector Positioning Recommended GS Overweight / S&P 500 Total Return GS Alpha S&P 500 Sector Positioning Underweight (a) Weight YTD YTD Energy 300 bp 12% 8% (18)bp Materials Financials Overweight Information Technology (1) Industrials Utilities Neutral Consumer Discretionary (14) Telecom Services (200) 3% 1 9 Health Care Underweight (300) (22) Consumer Staples (400) 12 9 (7) S&P 500 0bp 100% 19% (16)bp Thematic Trade Recommendations Initiation Our best trade ideas: Portfolio Strategy thematic baskets (b) Date Return BUY High Operating Leverage (GSTHOPHI); SELL Low Operating Leverage (GSTHOPLO) See Macro to Micro Shift Part II: 2H 2009 Update (20-Jul-09). 20-Jul % BUY BRICs Sales Basket (GSTHBRIC); SELL S&P 500 See Portfolio Passport: Coming to America (5-Nov-08). 4-May % BUY Hedge Fund Very-Important-Position (VIP) list Basket (GSTHHVIP); SELL S&P 500 See Hedge Fund Trend Monitor: Fund (24-Aug-09). 24-Aug % BUY Invest for Growth Basket (GSTHINVG); SELL S&P 500 See Investing for Growth: Capex and R&D (31-Aug-09). 31-Aug % (a) Sector weightings last rebalanced on 20-Jul-09. (b) US Portfolio Strategy baskets may be found on Bloomberg by typing <GSSU5>. The Bloomberg page provides real-time basket performance and current basket constituents. To obtain access to our baskets on Bloomberg, please contact your Goldman Sachs salesperson. The ability to trade these baskets will depend upon market conditions, including liquidity and borrow constraints at the time of trade. Source: FactSet and Goldman Sachs Global ECS Research Goldman Sachs Global Economics, Commodities and Strategy Research 12