Forecasting Emerging Markets Equities the Role of Commodity Beta Huiyu(Evelyn) Huang Grantham, Mayo, Van Otterloo& Co., LLC June 23, 215 For presentation at ISF 215. The opinions expressed here are solely those of the presenter. Copyright HuiyuHuang 215.
Outline Emerging markets (EM) countries commodities exposure Country-sector rotation strategy based on commodity returns and equity beta to commodities Data and methodology How and why it delivers excess return over the EM benchmark Is it a proxy for Value (or other prevailing alpha factors)? 1
Emerging markets are a diverse group of countries In terms of exposure to commodities 6 Net exports of crude oil in 21 4 2-2 -4-6 Russia UAE Venezuela Mexico Qatar Colombia Brazil Argentina Malaysia Egypt Indonesia Jordan Peru Slovakia Hungary Morocco Pakistan Czech Republic Chile Philippines Portugal Turkey South Africa Greece Poland Thailand Taiwan South Korea India China Thousand Barrels per Day As of 2/28/215 Source: International Energy Statistics, U.S. Energy Information Administration 2
Why does the net oil exporting position matter? Change in the price of oil will impact earnings outlook for upstream oil and gas producers. It will also impact the outlook for an EM country s current account balance. Big net exporters will suffer, but big oil importers will benefit as oil price drops. Improving current account situation helps reduce currency vulnerability (Krugman, 1979). Currency risk is not negligible for US dollar investors in EM space. Currency hedging is expensive in EM and often ineffective in reducing portfolio volatility at investment horizons beyond one year (LeGraw, 215). P. Krugman, A model of balance-of-payments crises, Journal of Money, Credit, and Banking 11(3):311-325, 1979. C. LeGraw, The case for not currency hedging foreign equity investments: a U.S. investor s perspective, 215. GMO white paper at http://www.gmo.com 3
Commodity return cycles 2 15 1 5-5 -1 Jan-71 Oct-72 Jul-74 Apr-76 Jan-78 Oct-79 Jul-81 Apr-83 Jan-85 Oct-86 Jul-88 Apr-9 Jan-92 Oct-93 Jul-95 Apr-97 Jan-99 Oct- Jul-2 Apr-4 Jan-6 Oct-7 Jul-9 Apr-11 Jan-13 Oct-14 Returns of commodities over 4 years Return correlation between commodity indices (18-month rolling return) GSCI DJ-UBS WTI Crude GSCI 1 DJ-UBS 92% 1 WTI Crude 83% 73% 1 As of 1/31/215 Source: Bloomberg, GMO GSCI DJ-UBS WTI Crude 4
How do EM equities react to change in commodity returns? Russia Energy: simple OLS beta, contemporaneous, and lead-lag. Russia Energy's beta to trailing 18-mth oil return 1.2 1.8.6.4.2 -.2 -.4 -.6 Jan- Aug- Mar-1 Oct-1 May-2 Dec-2 Jul-3 Feb-4 Sep-4 Apr-5 Nov-5 Jun-6 Jan-7 Aug-7 Mar-8 Oct-8 May-9 Dec-9 Jul-1 Feb-11 Sep-11 Apr-12 Nov-12 Jun-13 Jan-14 Aug-14 ret18 fwd3 fwd12 fwd18 fwd24 fwd36 As of 1/31/215 Source: Bloomberg, Standard & Poor s, GMO 5
How do EM equities react to change in commodity returns? For all EM country-sectors, and on a combination of GSCI, DJ-UBS, and Oil A model for EM country-sector relative return prediction through lead-lag beta on commodities First, aggregate stock returns relative to benchmark to the country-sector level Then, run a rolling window PLS regression for each country-sector on lagged commodity returns (standardized):,, Finally, use, as return forecast to form portfolios out-of-sample..8.6.4.2 -.2 -.4 -.6 Lead-lag beta to oil in PLS regression Jan- Oct- Jul-1 Apr-2 Jan-3 Oct-3 Jul-4 Apr-5 Jan-6 Oct-6 Jul-7 Apr-8 Jan-9 Oct-9 Jul-1 Apr-11 Jan-12 Oct-12 Jul-13 Apr-14 Russia Energy China Energy India Energy As of 1/31/215 Source: Bloomberg, Standard & Poor s, GMO 6
Contrarian investor wins Strategy performance relative to S&P IFCI Composite index, 1/1995 to 1/215 For each month t, sort and group country-sectors into five buckets, based on the Commodity Alpha,, with each bucket consisting of 2% of the entire IFCI Composite index weight. Form a long-only fully invested portfolio of country-sectors based entirely on the first bucket using the square root of the country-sector s index weight. Strategy performance with varying forward return horizon h Raw commodity alpha h=3 h=6 h=12 h=18 h=24 h=36 Excess return 7.8 8.75 7.1 7.24 7.87 7.93 Stdev 11.55 12.1 11.36 1.9 9.65 8.42 Information Ratio.61.73.62.72.82.94 Turnover 16.82 16.28 15.72 15.95 16.42 18.9 Drawdown -18.99-27.2-25.4-21.2-2.9-11.46 8 7 6 5 4 3 Cumulative relative performance of strategies 6-month moving average Excess return 6.93 8.47 8.73 9.86 1.3 8.41 Stdev 1.86 11 1.7 9.96 9.5 8.45 Information Ratio.64.77.82.99 1.6.99 Turnover 11.78 11.57 11.13 11.86 12.4 12.6 Drawdown -17.99-17.51-16.98-12.24-1.9-11.45 2 1 Jan-95 Jan-96 Jan-97 Jan-98 Jan-99 Jan- Jan-1 Jan-2 Jan-3 Jan-4 Jan-5 Jan-6 Jan-7 Jan-8 Jan-9 Jan-1 Jan-11 Jan-12 Jan-13 Jan-14 Jan-15 h=3 h=6 h=12 h=18 h=24 h=36 As of 1/31/215 Source: Bloomberg, Standard & Poor s, GMO 7
A story of sector rotations between defensives and cyclicals Defensives: Consumer Staples, Health Care, Telecoms, Utilities Cyclicals: Consumer Discretionary, Energy, Financials, Industrials, Materials..8 Commodity alpha factor aggregated to sectors.6.4.2 -.2 -.4 -.6 Jan-95 Nov-95 Sep-96 Jul-97 May-98 Mar-99 Jan- Nov- Sep-1 Jul-2 May-3 Mar-4 Jan-5 Nov-5 Sep-6 Jul-7 May-8 Mar-9 Jan-1 Nov-1 Sep-11 Jul-12 May-13 Mar-14 Jan-15 Consumer Staples Materials Defensives Cyclicals As of 1/31/215 Source: Bloomberg, Standard & Poor s, GMO 8
Is it a proxy for Value (or other prevailing alpha factors)? Value: price-to-book ratio Size: log of investible market capitalization Momentum: log of 12x1-month return (one based) 1.8.6.4.2 -.2 -.4 -.6 -.8-1 Jan-95 Nov-95 Cross-sectional correlations between commodity alpha and other factors Sep-96 Jul-97 May-98 Mar-99 Jan- Nov- Sep-1 Jul-2 May-3 Mar-4 Jan-5 Nov-5 Sep-6 Jul-7 May-8 Value Size Momentum Mar-9 Jan-1 Nov-1 Sep-11 Jul-12 May-13 Mar-14 Jan-15 As of 1/31/215 Source: Bloomberg, Standard & Poor s, GMO 9
Is it a proxy for Value (or other prevailing alpha factors)? Regressions of forward h-month equity return on factors Single-factor regression (of relative return on) Multi-factor regression (of absolute return on) Commodity alpha R-squared Commodity alpha Value Size Momentum Adj. R- squared h=6 coefficient.39.6%.4.77.3 -.17.9% t-stat 9.2 6.98 7.15 1.17-1.1 h=12 coefficient.23.4%.21.47.3 -.82 1.7% t-stat 7.47 5.39 6.33 1.73-7.14 h=18 coefficient.9.1%.6.4.1 -.69 1.7% t-stat 3.73 1.96 6.68.86-7.92 As of 1/31/215 Source: Bloomberg, Standard & Poor s, GMO 1