PortfolioStrategyA valuation anchor for EquitiesWe introduce GS DDM, a new approach to valuing equity marketsglobally, to sit alongside our other regional valuation models. We takethree approaches to fair value: 1) Fair Value under our current centralassumptions, applying current bond yields and estimated ERP. 2) FairValue adjusted for current ‘fair value’ bond yields using our Sudokubond model. 3) Equilibrium Value, assuming the ERP and bond yieldsrevert to long run averages.Estimating 'fair value' using a DDMIn Part I we develop a model that: 1) is estimated in the same way ineach major region to establish a benchmark for ‘fair value’; 2) lendsitself to reverse engineering – assessing what assumptions the marketis ‘implying’. We look at the sensitivity of fair values to changes in theunderlying assumptions. In our forthcoming Part II, we will extend thisanalysis to forecast the ERP and market levels.Substantial undervaluation from 'equilibrium'Our analysis shows that, using current central assumptions, mostmarkets are undervalued by between 3% and 15% with Europe themost undervalued. Using our ‘fair value’ Sudoku bond yieldassumptions instead of current yields increases the undervaluation inall regions. In all cases, an assessment of ‘equilibrium’ valuation, basedon assuming the ERP reverts to a long run average and using a longrun ‘trend’ real yield of 2% shows substantial undervaluation of 71% forEurope, 36% in the US, 31% in Asia ex Japan and 13% in Japan.Fair value and upside/(downside)Ce n tra l S ce n a rio Usin g S u do ku Fa irV a lu e Bo nd m od e lUsin g e qu ilib riu m ERP* a n d2% re a l in te re st ra teCu rre n t Le ve lFa irV a lu eL e ve lF a ir V a lu eUpsid e /(Do w nsid e)F a irV a lueL e ve lFa ir V a lu eUp sid e /(Dow n sid e)F a ir V a lueLe ve lF a ir V a lu eUpsid e /(Do w nsid e)US (S&P 500)84694111%94312%115236% Eu ro p e (S to x x 600)19522415%27441%33371% Ja pa n (T O P IX)7868093%91616%88613% Asia (M S CI AP x J)2662774%NA NA34931%* We estimate equilibrium ERP at 3% for all regions (4% for Asia).Source: Goldman Sachs Global ECS Research. Peter Oppenheimer+44(20)7552-5782 | peter.oppenheimer@ Goldman Sachs InternationalJessica Binder, CFA+44(20)7051-0460 | jessica.binder@ Goldman Sachs InternationalAnders Nielsen+44(20)7552-3000 | anders.e.nielsen@ Goldman Sachs InternationalGerald Moser+44(20)7774-5725 | gerald.moser@ Goldman Sachs InternationalSharon Bell, CFA+44(20)7552-1341 | sharon.bell@ Goldman Sachs InternationalThe 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 /research/hedge.html. Analysts employed by non-US affiliates are not registered/qualified as research analysts with FINRA in the U.S.Executive summary• We introduce a 4-stage dividend discount model that is estimated in the same way ineach major region. This is used as a valuation anchor for international comparison. •In generating results from GS DDM, we have taken three approaches. 1) ‘Fair value’ today using our central scenario (how should we think of equities relative to current bond yields?). 2) ‘Fair value’ today using ‘fair value’ bond yields, based on our Sudoku model (how should we think of equities assuming bonds were fairly valued in the current environment?). 3) ‘Equilibrium value’ using long run average real bond yields and ERP (how much of the current valuation gap is due to the fact that the discount rate is higher/lower than normal?).•We estimate that the upside to ‘fair value’ given the current economicenvironment and bond yields is, 15% in Europe, 11% in the US, 4% in Asia ex Japan and 3% in Japan .• Adjusting the ‘fair value’ for current ‘fair value’ bond yields increases the undervaluation in all equity markets.•On a long run ‘Equilibrium Value’ basis we find an upside to fair value of 71% in Europe, 36% in the US, 31% in Asia ex Japan and 13% in Japan. We estimate the upside to long run equilibrium fair value by using a real interest rate of 2% and an equity risk premium of 3% (except in Asia ex Japan where we use 4%) in our dividend discount model.•Exhibit 2 shows that the deviation of market values from long runequilibrium levels tend to close over time. One percentage point of additional upside to fair value at the beginning of the year on average increases the total return during the year by 0.3 percentage points (Exhibit 3). Exhibit 3 also shows that the upside to fair value explains 15% of the variation in yearly returns.Exhibit 1: GS DDM “fair-value” and “equilibrium” levelsCe ntra l S ce na rioUsing S udoku Fa ir V a lue Bond m ode l Using e quilibrium ERP * a nd2% re a l inte re st ra te Curre nt Le ve l Fa ir V a lue Le ve l Fa ir V a lue Upside / (Dow nside )Fa ir V a lue Le ve l Fa ir V a lue Upside / (Dow nside )Fa ir V a lue Le ve lFa ir V a lue Upside / (Dow nside )US (S &P 500)84694111%94312%115236%Europe (S tox x 600)19522415%27441%33371%Ja pa n (TO P IX )7868093%91616%88613%Asia (M S CI AP x J)2662774%NANA34931%Source: Goldman Sachs Global ECS Research.Exhibit 2: S&P 500 reverts to fair value over timeExhibit 3: Deviations from fair value predicts returns-60%-40%-20%0%20%40%60%Jan-89Jan-92Jan-95Jan-98Jan-01Jan-04Jan-07UndervaluedOvervalued-50-40-30-20-1001020304050-60-40-202040Upside to lo ng run fair value (%)R e t u r n o v e r t h e n e x t y e a r (%)Source: Datastream, Haver analytics, Goldman Sachs Global ECS ResearchSource: Factset, Haver analytics, Goldman Sachs Global ECS ResearchFinding a ‘fair value’ anchor for equities; Introducing GS DDMValuing equity markets is always controversial because any chosen approach relies onassumptions. Equally, valuation is not the only driver of markets. Markets can stay aboveor below theoretical ‘fair’ value for long periods of time. Nevertheless, using a standard,easily comparable, valuation model that provides an ‘anchor’ to fair value and can be usedto compare across markets can help to identify opportunities and risks; it can also be usedas a forecasting tool. At the very least, it might also shed some light on the kinds ofassumptions that implicitly need to be made in order to justify deviations from equilibriumvaluation over time.Across the regions, GS Strategists use a variety of different tools and measures to assessvalue. The importance given to any particular approach varies by region and over time,according to the structure of the market and the stage of the cycle. Our interest here wasnot to replace these models but to supplement them with a single approach that can beapplied globally, providing a guide to ‘fair value’ using an assessment of current driversand also as a guide to ‘equilibrium’ – the fair value level of the market assuming that theEquity Risk Premium converges to the long run average and real bond yields are at theirlong run trend.Two approaches to valuationWe introduce a common basis for the global valuation of equities with GS DDM (dividenddiscount model). In each region, we monitor and forecast with several different types ofvaluation tools. These include earnings based multiples (cyclically adjusted), dividend andfree cash flow yields, book valuations, relative valuations against competing assets and anumber of other approaches. For each market we also have a dividend or earningsdiscount model.However, many of these models are used as a guide to valuing the market in isolation atany time, rather than to compare across markets or focus on an intrinsic ‘fair value’.Across the Global Economics, Strategy and Commodities Research Group (ECS) we have,over time, developed a number of valuation tools intended to generate a measure of‘equilibrium’ value for each asset class. These can be thought of not so much as a specificforecast for a market but more as a valuation ‘anchor’. The advantage of these is that theycan provide useful warning signals when markets become very stretched relative to theunderlying fundamentals. Furthermore, they provide a framework for understanding whatkinds of assumptions investors would need to make at any time to justify current marketlevels if they are significantly away from fair value.These models include our Sudoku model for the bond markets and GSDEER for foreignexchange markets. In this paper we highlight a new dividend discount model approachthat we have developed with the aim of providing an internationally comparable valuationanchor for equities.In effect there are two ‘approaches’ to equity valuation models, both of which we introducein a new approach to valuing equity markets globally.• A ‘fair value’ approach: by fixing specific inputs, such as the equity risk premia, onecan back out a ‘fair value’; this approach is helpful in providing a valuation anchor.• A ‘forecast’ approach: using a fair value to generate an expected future price level.Valuation as a ‘fair value’ guideOur intention in this paper, Part I of a two part series, is the following.•To have a model that is estimated in the same way in each major region so that aconsistent benchmark for ‘fair value’ can be established. In this way we can comparevaluations across regions and also relative to history.•Develop an approach that lends itself to reverse engineering – assessing whatassumptions the market is ‘implying’. This can be particularly useful at turning pointsor periods when the market appears to have deviated materially from ‘fair value’. If amarket appears expensive, then what kinds of assumptions would be required in orderto justify such valuations?•To generate an assessment of ‘Fair Value’ for the market by applying assumptionsabout it (bond yields, ERP, earnings etc).•To establish an ‘Equilibrium Level’ level for the markets by assessing the ‘fair value’assuming that the ERP converges to a long run average level and real bond yields areat their long term trend.Valuation as a forecast toolThe forecast returns approach attempts to assess how quickly the convergence of prices tofair value is likely to happen. In a separate report, Part II of our Valuation series, we intendto invert the fair value model presented in this paper to back out the ERP required toequate market levels to the theoretical fair value. By doing this at different points in timewe generate a time series of the implied equity risk premium. We then analyze therelationship of the premium with macroeconomic variables, allowing us to forecast theERP and, therefore, market levels.Setting out the frameworkWe considered several valuation approaches to find the one that might be useful for thisexercise. One distinction when comparing valuation tools is whether to value the equity orthe enterprise value, another is to consider models based on cash flows, returns ormultiples.Cash flow models focus on cash flow to equity holders – dividends or flows to equity anddebt holders – free cash flow. Returns models focus on the capital stock and the spreadbetween the return and the cost of capital.For the purposes of this exercise we focus on equity holders and are interested in adiscount model rather than a multiple model so that we can more easily test assumptionsand scenarios when markets deviate from ‘equilibrium’. It is for this reason that we settledon a DDM approach. The model that we have chosen covers the US (S&P 500), Europe (DJStoxx 600), Japan (TOPIX) and Asia (MSCI Asia-Pacific ex Japan).The model is based on four ‘phases’ and assumes that a proportion of earnings are paidout each year as dividends. The length of the phases and the proportion of value for theS&P 500 index that falls in each phase are given in Exhibit 4.Exhibit 4:Model time linePhase I Phase II Phase III Phase IVFinancial years% of total value5%5%23%67%Source: Goldman Sachs Global ECS Research.Phase I: Forecast growth – years 1 and 2We use our top-down estimates for operating earnings growth for year 1 and year 2. For each year, we adjust the previous year’s payout ratio by a factor that relates to the forecast growth rate (for example if earnings are expected to rise rapidly, we assume that last year’s payout ratio falls moderately as dividends are more stable than earnings).Phase II: Fade to trend ROE – years 3 and 4We assume that the market gets back to trend ROE by the end of year 4 and that the earnings change to achieve this occurs equally over the two years. In a normal part of the cycle this may not result in very different growth in earnings from an average year but at major turning points it could result in quite a large jump in growth in either direction. Having an assumption that gets back to a long-term trend is important. Without it, for example, a model could imply ongoing growth from an unsustainably high (or low) level following a boom (or collapse) in profits. Growing from an unsustainable high or low level of profits would alter the implied fair value materially.Phase III: Long-term growth rate – years 5-20We assume that profits grow at their trend rate of growth (equal to the long-term real economic growth rate plus inflation) but assume that the proportion paid out by companies over this period equals the average of the past 5 years.Phase IV: The terminal valueWe assume that any profits growth in perpetuity is offset by a commensurate change in the payout ratio; this way we ensure that the return on equity is equal to the cost of equity and profits grow in line with trend real GDP.Other key assumptions for current ‘fair value’In order to calculate a fair value, we also need estimates for the following.•Risk-free rate: We use the current 10-year bond yield for each of the regions (for Asia we use a cap-weighted average).•Inflation: We have settled on a 5-year moving average of core inflation as the most stable and consistent measure of “expected future” inflation that is easily measurable in each region.•ERP: While we intend to model the ERP separately in our valuation ‘forecast’ paper, we cannot use the results of that model as an input into the DDM because it becomes circular (the DDM ‘fair values’ are used to extract the historical series of ERP).Consequently, we are constrained to making an assumption for the ERP in this model.This is an important limitation since we cannot observe the required ERP directly at any time. For the purposes of our central assumptions used in our assessment ofcurrent ‘fair value’, we make an assessment about the current appetite for risk, and chose an ERP by reference to the long run ERP series from our ERP model (to bediscussed in detail in a follow up report). Currently, for example, we use an ERP of 5%for the developed markets and 6% for Asia, which is in the top quartile of the historical distribution. While changing the assumption for the ERP can make big differences to the output of the model, we examine the sensitivities to these assumptions later in this report. While we do make an assumption about the ERP for our central caseassessment of fair value, we also assess the ‘equilibrium’ level of the market byassuming the ERP and real bond yields are at their long run average.All the assumptions are summarized in Exhibit 5:Exhibit 5:Summary of assumptions for our central scenarioEarningsFY1-5%-16%-37%-15%FY231%8%-3%17%FY33%3%9%5%FY43%3%8%5%Payout RatioFY141%54%67%40%FY238%54%48%38%FY339%56%43%38%FY440%58%39%38%Year 5-2032%41%34%35%ROE14%13%6%15%Long termInflation2%2%1%4%Real Growth3%3%2%4%10 year bond3%3%1%5%ERP5%5%5%6%Source: Goldman Sachs Global ECS Research.What the model says nowIn generating results from GS DDM, we have taken three approaches.1. ‘Fair Value’ today using our central scenarioSuppose the bond markets are right and the ERP is where we think it belongs in this kindof environment, where would you expect fair value for stocks to be? (How should we thinkof equities relative to current bond yields)2. ‘Fair Value’ today using ‘fair value’ bond yields (based on our Sudoku model)Suppose the bond markets are themselves at fair value and the ERP is where we think itshould be, where would you expect fair value for stocks to be? (How should we think ofequities assuming bonds were fairly valued in the current environment)3. ‘Equilibrium Value’Suppose the discount rate (both the real bond yield and ERP) was at its average historicallevel, where would you expect fair value for stocks to be? (How much of the currentvaluation gap is due to the fact that the discount rate is higher than or lower than normal).Fair Value under central scenarioTaking the final results form our model, explained in more detail below, the ‘fair value’central scenario levels and percentage upside currently are shown in Exhibit 6.Exhibit 6:DDM equity ‘fair value’ based on central scenarioCurre nt Le ve lFa irV a lueLe ve lFa ir V a lueUpside /(Dow nside)US (S&P 500)84694111%E urope (S tox x 600)19522415%Japan (TO P IX)7868093%A s ia (M S CI A P x J)2662774%Source: Goldman Sachs Global ECS Research.All four markets are trading below fair value but the upside is limited. This is because the valuation is taking into account the current weak economic environment by using an equity risk premium of 5% for all markets except Asia ex Japan, where we use 6%. The upside to fair value is substantially higher if economic assumptions consistent with long run equilibrium are used. Under such assumptions the upside to the European market is 71% for example (see further discussion of this below).This analysis is in line with other work from Goldman Sachs strategists, where we have argued that markets are currently attractively valued but that other criteria are needed to be satisfied in order for that value to be unlocked. Specifically, risk premia need to be reduced.In recent work, many of our strategists have written about conditions that may need to change in order for the long term fair value to be unlocked in a sustained recovery. We believe three key criteria are the following.•Evidence that economic activity is stabilizing. In this regard we focus on our proprietary indicators such as the GLI and the Financial Conditions Index (as ameasure of policy efficacy) as a guide to a shift in the first and second derivative of growth.•We also expect to see evidence of the market absorbing disappointing data, both macro economic and company specific, without a negative price reaction.•We have argued that an improvement in the credit markets is also required as a prerequisite for a sustained recovery in equities since, on a risk adjusted basis, parts of the credit market have overshot fair value even more than equities (see for example our report of November 14, 2008, Strategy Matters: Credit versus equity: Credit offers better value). Again, in this regard, recent improvements in our Financial DistressIndex are also encouraging.Fair Value, adjusting for Sudoku ‘fair value’ bond yieldsOne of the issues that may affect the current fair value of equities is the bond yield. In the central case assumptions that we have used in the fair value levels shown in Exhibit 6, we applied the current 10-year bond yields as our risk free rate. In practice this gives some advantage to the US where real rates are much lower. Some investors argue that it is the bond market that is ‘mis-priced’ (particularly given growing supply risk) and that this distorts equity valuation measures that use the bond yield as an input. Our bond strategists do not believe this to be the case (see Bond Snapshot January 26, 2009) but as a cross check, we also show the current equity ‘fair values’, based on our fundamental assessment of ‘fair value’ in the bond market using the results of our Sudoku bond fair value model. While we do not have estimates for Asia, the conclusions are broadly the same for equities in general: equity markets are currently attractively valued. Europe’s fair value upside, in particular, rises sharply to 30% when we apply our fair value bond estimates.Exhibit 7:DDM equity ‘fair value’ based on Sudoku bond modelCurre nt Bond Yie ld Fa ir V a lueUpside /(Dow nside)S udokuBondYie ldFa ir V a lueUpside /(Dow nside)US(S&P500) 2.9%11% 2.9%13% E urope (S tox x 600) 3.5%15% 2.6%30% Japan (TOP IX) 1.3%3%0.9%16% A s ia (M S CI A P x J) 4.9%4%NA NA Source: Goldman Sachs Global ECS Research.Establishing a long run ‘Equilibrium Value’In the above ‘fair value’ calculations, we use current bond yields (or those adjusted by our ‘fair value model) and an assessment about current risk appetite. As discussed above, probably the most controversial input into this model is the ERP. In the central assumptions we plug in an ERP that relates to a band in relation to long run averages. If we think that uncertainty is high (as we do currently) we input an ERP in the highest quartile of historical patterns and vice versa when the market is trending and growth is robust. The advantage of this approach is that it gives an idea about what fair value is at the moment by using an ERP that is reasonable given the current economic environment.Arguably, however, this adds a further layer of subjectivity. We can think of two sensible alternatives that help pin down this issue, which each illuminate different issues. The first is to try to model more formally what an appropriate ERP "should" be given a particular economic environment. This is one of the things our follow up report: Part II will address. The second is to think about what the fair value of the market would be if economic conditions were to reach long run equilibrium. We consider an ERP of 3% and a real bond yield of 2% as reasonable benchmark case. Using those assumptions, Exhibit 8 shows how fair value has fluctuated over time.Exhibit 8: GS DDM, US equilibrium value% deviation form ‘fair value’ assuming fixed ERP at 3% and real bond yield of 2%-60%-40%-20%0%20%40%60%Jan-89Jan-92Jan-95Jan-98Jan-01Jan-04Jan-07Undervalued OvervaluedSource: Datastream, Haver analytics, Goldman Sachs Global ECS Research.While we do not explicitly use the deviation from equilibrium fair value as a forecasting tool, a good measure of equilibrium fair value should have the property that market values tend to move towards the equilibrium value over time. Exhibit 9 shows the yearly return of the S&P 500 index plotted against the deviation from long run fair value at the beginning of the year. The R 2 shows that 15% of the variation in yearly returns over time can beexplained by the market’s deviation from fair value at the beginning of the year. The slope coefficients shows that for each 1 percentage point increase in upside to equilibrium fair value at the beginning of the year, returns during the year increase by 0.3 percentage points.Exhibit 9:Returns are higher when the upside to fair value is large at the start of the year-50-40-30-20-1001020304050-50-40-30-20-1010203040Upside to long run fair value (%)R e t u r n o v e r t h e n e x t y e a r (%)Source: Factset, Haver analytics, Goldman Sachs Global ECS ResearchComparison of approachesThe three approaches are compared in Exhibit 10.•On current assumptions, all markets are undervalued, with the Europe being the most attractive and Japan and Asia the closest to fair value.•By applying an assessment of current ‘fair value’ bond yields as defined by our Sudoku model, the broad conclusions remain the same though the undervaluation of European equities becomes significantly larger.•Assuming the ERP (which we believe is currently unusually high) reverts to a long run average of 3% in the developed markets and 4% in Asia and using a trend average real bond yield, all markets are substantially undervalued. The range of undervaluation is from 13% in Japan to 71% in Europe.Exhibit 10:Fair value and upside/(downside)Ce ntra l S ce na rio Using S udoku Fa irV a lue Bond m ode lUsing e quilibrium ERP* a nd2% re a l inte re st ra teCurre nt Le ve lFa irV a lueLe ve lFa ir V a lueUpside /(Dow nside)Fa irV a lueLe ve lFa ir V a lueUpside /(Dow nside)Fa ir V a lueLe ve lFa ir V a lueUpside /(Dow nside)US (S&P 500)84694111%94312%115236%Europe (S tox x 600)19522415%27441%33371%Ja pa n (TO P IX)7868093%91616%88613%Asia (M S CI AP x J)2662774%NA NA34931%* We estimate equilibrium ERP at 3% for all regions (4% for Asia)Source: Goldman Sachs Global ECS Research.Sensitivities to assumptionsWhile the model helps to identify the potential upside or downside to ‘fair value’, it is verysensitive to changes in inputs and structure. But how much do changing assumptionsmatter and which variables are the most sensitive? In the section we examine theseassumptions and sensitivities in more detail.Near-term earnings expectations Phase IHow do you measure earnings?The first phase of the model is an explicit forecast period based on our own top downoperating profit assumptions generated from our regional profit models (details can beobtained on request). We assume that earnings revert to the historical trend rate of growthby the end of 2012 (Phase II).Exhibit 11: Short-term earnings growth expectationsFY indicates fiscal year. For Europe, US and Asia, fiscal years end December 31. For Japan,fiscal years end March 31.Earnings growth assumptionsFY1FY2FY3 and FY4CAGR Trend ROE US-5%31%3%14%Europe -16%8%3%13%Japan -37%-3%8%6%Asia-15%17%5%15%Source: Goldman Sachs Global ECS Research.Given how the model is set up, changes in near-term earnings expectations do not affectfair value to a very large extent, since we assume that earnings recover to trend by the end of the fourth year in the model. Exhibit 12 shows the impact on fair value in Europe, as an example, of changes in the assumptions for earnings growth in fiscal years 1 and 2 (out to 2010).Our current assumptions for the DJ STOXX 600 are for operating profits to fall 16% this year and to rise by 8% in 2010. As with many other markets, these numbers may be heavily distorted by financials for this year. For example, our top down model assumes that financial earnings rise by 29% this year after a fall of 47% in 2008 (mainly as a result of lower writedowns). However, we do not have much confidence in these numbers given the uncertainty over banks writedowns currently. Excluding financials, we assume earnings fall 30% this year. If we assume that this is true for the market as a whole, and the recovery in 2009 is still 8% from that lower level, it implies that the market is close to fair value (10% upside).Exhibit 12: Sensitivity of fair value for European market to short-term earnings forecasts shading represents current estimatesN o m i n a l F Y 1 e a r n i n g s g r o w t hSource: Goldman Sachs Global ECS Research。