Risk and Return of Short Duration Equity Investments

Transcription

1 Risk and Return of Short Duration Equity Investments Georg Cejnek and Otto Randl, WU Vienna, Frontiers of Finance 2014 Conference Warwick, April 25, 2014

2 Outline Motivation Research Questions Preview of Results Short Duration Equity Investments Dividend Derivatives Data Investment Strategy Diversification Ex-ante Risk Premia Definition and Determination Implied Risk Premia Over Time Explaining Realized Excess Returns Conclusions Motivation 2 / 49

3 Motivation There is strong recent interest in the dynamics of the term structure of equity risk premia. Academia: Explicit modeling and calibration of asset pricing models to price dividend strips Binsbergen, Brandt, and Koijen (2012) Campbell and Cochrane (1999), Bansal and Yaron (2004) (upward sloping term structure) Gabaix (2012), Lettau and Wachter (2007) (downward sloping term structure) Practitioners: Increasing attention and trading activity in dividend derivatives Hedge Funds Institutional investors Lately, retail investors Motivation 3 / 49

4 Research Questions What is the magnitude of realized risk premia of short maturity equity investments compared to traditional equity benchmarks? What are the risk profiles of short duration equity investments? Risk exposures (market betas, downside market exposure, options-based risk factors, realized dividend variability) Ex-ante risk premia Motivation 4 / 49

5 Preview of Results Realized excess returns of short maturity equity investments are higher than for the benchmark in 3 out of 4 markets. There are higher international diversification benefits for short maturity equity investments as compared to the benchmarks. Market betas of short maturity strategies are low, but increase in times of market distress. Alphas of the short maturity strategies largely disappear when options-based risk factors are taken into account. Still, the attractive performance is puzzling in the light of historically stable cash dividends. Ex-ante risk premia are high on average and vary substantially over time. Motivation 5 / 49

6 Outline Motivation Research Questions Preview of Results Short Duration Equity Investments Dividend Derivatives Data Investment Strategy Diversification Ex-ante Risk Premia Definition and Determination Implied Risk Premia Over Time Explaining Realized Excess Returns Conclusions Short Duration Equity Investments 6 / 49

7 Dividend Derivatives The market for dividend derivatives emerged in the early 2000s (OTC). Since exchange listed dividend futures were introduced in 2008, the investor base has broadened substantially. Most dividend derivatives (swaps or futures) reference to an equity index; like for instance to the Euro Stoxx 50. The underlying of index dividend derivatives are cash dividends accumulated by all index members throughout a year. The aggregation works analogous to the traditional equity index (index shares, index divisor, etc.) Dividend derivatives are available for several maturities (several years ahead; each year one maturity). Futures prices / swap strikes of several subsequent maturity years establish a futures curve. Short Duration Equity Investments 7 / 49

8 Dividend Derivatives (2) The final payoff of a dividend derivative is simply the difference between the initial futures price / swap strike and the dividend amount realized throughout the maturity year. If held to maturity, investors in dividend derivatives are only exposed to cash flow (dividend) risk, not to valuation risk. Short Duration Equity Investments 8 / 49

9 Illustration: Current Termstructure Euro Stoxx 50 Dividend Futures (Bloomberg, April 17, 2014) Short Duration Equity Investments 9 / 49

10 Data Main sample We use a proprietary sample of dividend swaps for several maturities from Goldman Sachs. 1 Sample period is Jan 2006 May We use weekly frequency. We analyze 4 markets: Euro Stoxx 50, FTSE 100, S&P 500, and Nikkei 225 Benchmarks are constructed using index futures prices from Bloomberg. Additional data To estimate Lintner models we construct a long sample on annual dividends and earnings for France, Germany, the UK, the US and Japan using data from Global Financial Data. To proxy for macroeconomic conditions we use ISM data from Bloomberg. We employ data on consensus analysts estimates of current year and next year index dividends and earnings. 1 We thank Christian Mueller-Glissmann for providing us with data. Short Duration Equity Investments 10 / 49

11 Investment Strategy To obtain realized risk premia for an equity strategy with short maturity, we create investment strategies with constant maturity, using dividend swaps. Our strategy is as follows: Invest in dividend swaps with several maturities. Each year in December (maturity for the current year contract), roll the exposure into the subsequent contract. At the roll date, adjust the exposure according to the shape of the futures curve. Note that these rolling strategies have varying maturities. To establish constant maturity, combine two subsequent contracts. Start out with 52 contracts of the shorter contract. Each week, roll one contract over into the subsequent maturity contract. Short Duration Equity Investments 11 / 49

12 Investment Strategy (2) The swap price of the portfolio therefore reads as w t F t,fy (t) + (1 w t )F t,fy (t+1) Assuming fully collateralized positions, we calculate excess returns of the strategies as follows R t+1 = F t+1 F t F t We focus on the one year constant maturity strategy; we provide some results for longer (constant) maturities though. Short Duration Equity Investments 12 / 49

13 Weighting the Contracts 100% weight % weight % weight % weight year constant maturity dividend strategy (for 2011) maturity/roll date dividend swaps 2010 January 2011 December st maturity/roll date index futures 2 nd maturity/roll date index futures Short Duration Equity Investments 13 / 49

14 Investment Performance - Euro Stoxx Short Duration Equity Investments 14 / 49

15 Investment Performance - FTSE Short Duration Equity Investments 15 / 49

16 Investment Performance - S&P 500 Short Duration Equity Investments 16 / 49

17 Investment Performance - Nikkei Short Duration Equity Investments 17 / 49

18 Investment Performance - Summary Stats CMDS 1 Benchmark Mean St.dev. Sharpe Mean St.dev. Sharpe Euro Stoxx FTSE S&P Nikkei Global Annualized mean returns of the CMDS strategy exceed benchmark returns for all markets but the S&P 500. In all markets, the standard deviation is lower. Short Duration Equity Investments 18 / 49

19 Transaction Costs R Div,n,TC t = R Div,n t TC D 52 where R Div,n,TC t are returns after transaction costs TC D. We set { Rt bm,tc R bm = t if t rolldate Rt bm TC bm if t = rolldate TC D to 20 bps (per round trip) TC bm to 5 bps (per round trip) Short Duration Equity Investments 19 / 49

20 International Diversification Comparison of CMDS correlations across markets (upper right) and stock market index futures correlations (bottom low) Euro Stoxx FTSE S&P Nikkei Euro Stoxx FTSE S&P Nikkei Difference is significant economically and statistically (Jennrich test) Short Duration Equity Investments 20 / 49

21 Investment Performance - Global Short Duration Equity Investments 21 / 49

22 Global Strategies Performance per Year CMDS 1 Benchmark Mean St.dev. Sharpe Mean St.dev. Sharpe Overall Short Duration Equity Investments 22 / 49

23 Conditional Correlations Conditional on market state, correlation structure appears more attractive for 1-year CMDS than for the benchmark. Up Down Out-of-Phase CMDS BM CMDS BM CMDS BM Euro Stoxx / FTSE Euro Stoxx / S&P Euro Stoxx / Nikkei FTSE / S&P FTSE / Nikkei S&P / Nikkei Results are qualitatively similar if we condition on ISM data as a proxy for macroeconomic conditions. Short Duration Equity Investments 23 / 49

24 Risk Profile To evaluate risk exposures of the strategies we compute market betas. Following recent literature on downside risk (e.g., Lettau et al., 2014), we account for increasing market exposures during times of distress. Thus we estimate R Div,n t = α + βr bm t + β down R t D t + ɛ t, where D t is 1 if the benchmark return is less than µ σ and 0 otherwise. Short Duration Equity Investments 24 / 49

25 Accounting for Potential Illiquidity We use proprietary data for markets that might be illiquid. Therefore we regress dividend returns on contemporaneous as well as lagged market returns. We report aggregated coefficients and use a Wald test to determine the joint statistical significance of both lags. This corresponds to the Dimson (1979) method for stale prices. Short Duration Equity Investments 25 / 49

26 Beta Regressions 1-yr CMDS CMDS 1yr Euro Stoxx FTSE S&P 500 Nikkei Global Alpha *** *** *** *** *** *** 8e-04*** *** *** *** Beta *** *** *** *** *** *** *** *** *** *** Down-Beta *** *** *** *** *** *** *** *** *** *** No. obs 384*** 384*** 384*** 384*** 384*** Adj. R *** *** *** *** *** Short Duration Equity Investments 26 / 49

27 Options-based Risk Factors In the Hedge Fund literature, options-based risk factors capture non-linear exposures (see Naik, 2004 and Mitchell, 2001). We download time series of implied volatilities (1 month maturity) of various levels of moneyness that we interpolate linearly to get a continuum of implied volatilities. We compute Black-Scholes prices for European put options for all four benchmark markets. We consider the return of a systematic put writing strategy, where each week a 5% OTM put option is sold, and bought back one week later. Short Duration Equity Investments 27 / 49

28 Options-based Risk Factors (2) Since this strategy is non-linear in nature, we use it as an alternative proxy for downside risk. We estimate R Div,n t = α + βr bm t + β put R put t + ɛ t, where we use again contemporaneous as well as 1 week lagged returns. Short Duration Equity Investments 28 / 49

29 Option Sensitivity Regressions 1-yr CMDS CMDS 1yr Euro Stoxx FTSE S&P 500 Nikkei Global Alpha *** *** *** *** *** *** *** *** *** *** Beta *** *** *** *** *** *** *** *** *** *** Option-Beta *** *** *** *** *** *** *** *** *** *** No. obs 384*** 294*** 384*** 384*** 384*** Adj. R *** *** *** *** *** Short Duration Equity Investments 29 / 49

30 Variability of Dividends vs Index Changes At maturity, dividend swaps are settled according to realized dividends of the year. As time to maturity shortens, discounting the expected settlement price (the risk premium) plays less a role for price determination. Therefore, the risk inherent in short term dividend swaps should be closely related to the risk of cash dividends. We report annual dividend and price index changes for the US, the UK, Japan, Germany and France. Short Duration Equity Investments 30 / 49

31 Variability of Dividends vs Index Changes (2) Short Duration Equity Investments 31 / 49

32 Outline Motivation Research Questions Preview of Results Short Duration Equity Investments Dividend Derivatives Data Investment Strategy Diversification Ex-ante Risk Premia Definition and Determination Implied Risk Premia Over Time Explaining Realized Excess Returns Conclusions Ex-ante Risk Premia 32 / 49

33 Components of the Dividend Swap Price swap price F t,t dividend expectations E t (D T ) risk premium µ risk free rate r f does not enter! Ex-ante Risk Premia 33 / 49

34 Components of the Dividend Swap Price Binsbergen et al. (2013) show that the present value of future dividends relates to current dividends through expected dividend growth, the risk-free rate of interest, and the risk premium. Note that the dividend swap price F t,t = CEQ(D T ) Its present value P t,t = F t,t e r F can also be obtained by discounting expected future diviends at the risk-adjusted rate: P t,t = E t (D T ) e (r F +µ) Hence, µ = ln ( Et(D T ) F t,t ) = ĝ + ln ( ) Dt F t,t Or, using discrete returns µ = Et(D T ) F t,t 1 = Dt (1+ĝ) F t,t 1. Ex-ante Risk Premia 34 / 49

35 Ex-ante Risk Premia and Dividend Expectations We want to obtain an estimate for the ex-ante risk premium contained in the dividend swap price. The ratio of the current dividend level to the dividend swap price combines information on expected dividend growth and the dividend risk premium. If we have an estimate for dividend growth, we can back out the risk premium. We use a Lintner (1956) model to obtain estimates for the future dividend level. Ex-ante Risk Premia 35 / 49

36 Lintner Model (1) Aggregate dividends depend on the past dividend level and current earnings For robustness, estimation is usually done to explain changes in dividends: D Y (t) = λ D Y (t 1) + β E Y (t) + ɛ t, where D Y (t) are dividends of year t, E Y (t) are earnings of year t, and D Y (t) = D Y (t) D Y (t 1). We estimate the regressions on data ending 2005, as our dividend swaps data start in Ex-ante Risk Premia 36 / 49

37 Lintner Model Estimation Results France*** Germany*** UK*** US*** Japan*** ˆλ *** *** *** *** *** ˆβ *** *** *** *** *** No. obs 34*** 34*** 60*** 60*** 49*** Adj. R *** *** *** *** *** Ex-ante Risk Premia 37 / 49

38 Obtaining Expected Dividend Growth Rates Using the estimated coefficients from the Lintner model and estimated earnings from analyst forecasts, we obtain estimated growth rates for year-end dividends: ĝ D t,y (t 1),Y (t) = ˆλ + ˆβ E(E Y (t)) D Y (t 1) For next year s growth rate we use analyst forecasts for both earnings and current year dividend levels ĝ D t,y (t),y (t+1) = ˆλ + ˆβ E(E Y (t+1)) E(D Y (t)) Ex-ante Risk Premia 38 / 49

39 CMDS ex-ante Risk Premia We back out a time series µ t for the risk premium of the one-year CMDS which is actually a portfolio containing two swaps. Solving for µ t w t ˆD Y (t) (1+µ t) (Y (t) t) + (1 wt) ˆD Y (t+1) (1+µ t) (Y (t+1) t) = w t F t,fy (t) + (1 w t )F t,fy (t+1). Ex-ante Risk Premia 39 / 49

40 Ex-ante Risk Premia Summary Stats Euro Stoxx FTSE S&P 500 Nikkei Mean Median St.Dev Min Max Ex-ante Risk Premia 40 / 49

41 1 Year Implied Risk Premium EuroStoxx 50 Ex-ante Risk Premia 41 / 49

42 1 Year Implied Risk Premium FTSE 100 Ex-ante Risk Premia 42 / 49

43 1 Year Implied Risk Premium S&P 500 Ex-ante Risk Premia 43 / 49

44 1 Year Implied Risk Premium Nikkei 225 Ex-ante Risk Premia 44 / 49

45 Relation of ex-ante Risk Premia to Realized Excess Returns How do ex-ante risk premia relate to realized excess returns? We estimate the following regression R Div,1 t,t+1y = α + βµ t,t+1y + ɛ t We find a significant relation Ex ante risk premia explain about half the variation of ex post realized risk premia! Dynamic strategies might be even more attractive than long only. Ex-ante Risk Premia 45 / 49

46 Results - Explaining Excess Returns Euro Stoxx FTSE S&P 500 Nikkei Intercept *** *** *** *** *** *** *** *** ex-ante RP *** *** *** *** *** *** *** *** Adj. R *** *** *** *** No. Obs. 332 *** 332*** 332*** 332 *** Ex-ante Risk Premia 46 / 49

47 Outline Motivation Research Questions Preview of Results Short Duration Equity Investments Dividend Derivatives Data Investment Strategy Diversification Ex-ante Risk Premia Definition and Determination Implied Risk Premia Over Time Explaining Realized Excess Returns Conclusions Conclusions 47 / 49

48 Conclusion Summary Historically attractive performance of short term equity claims; better international diversification benefits than the benchmarks. This can partly be explained by downside risk exposure. Yet still puzzling in the light of sticky cash dividends. Strong indication for time-varying expected returns in dividend strategies. Our findings give empirical support to recent asset pricing models consistent with a downward sloping term structure of equity risk premia. Conclusions 48 / 49

49 Conclusions Takeaways For academics: Asset pricing models should be able to match not only magnitude and volatility of risk premia, but also the term structure. For practitioners: In their portfolio selection problem, equity investors should not only consider country indices, but also allocate maturity buckets similar to fixed income investors. Conclusions 49 / 49