GLOBAL TACTICAL ASSET ALLOCATION: BOTTOM-UP
Rajeev Seth BeatIndex
http://www.beatindex.biz/html
September 22-23, 2008
Outline
Background Global tactical asset allocation: bottom-up process Pros / cons of the bottom-up process Example of bottom-up approach Universe of ETFs Alternative Ways to Do Bottom-Up ETF Portfolio Selection Bottom-up GTAA ETFs model results Aug 30, 2008
(c) 2008 BeatIndex 2
Background
What is Global Tactical Asset Allocation (GTAA)
Decision making process for choosing the appropriate short-term asset weights from the targeted universe of ETFs. Comes after the long-term or normal asset mix has been set using strategic asset allocation Shifts the asset mix in the short term based on predictions of the relative excess returns offered by the various ETFs. Attempts to answer the question of the relative expected future returns/risk of the constituent ETFs. Manages risk while producing a commensurate excess-return
GTAA – overall process
(c) 2008 BeatIndex
3
GTAA – Bottom Up Process
Such forecasts of relative future returns to the ETFs could have been generated in two alternative ways: Qualitative judgments of experienced market strategists and economists, or Quantitative modeling that uses statistical techniques on crosssectional or historical relationships in the data. BeatIndex uses the latter approach of using quantitative techniques to generate bottom-up forecasts for the expected return for each ETF. In this bottom-up approach, forecasts for individual stocks in an ETF are aggregated to generate a forecast for the entire ETF. Contrast this with the former top-down approach, where gross forecasts would simply have been made for each ETF reflecting a judgment on the expected performance of its industry group or sector.
(c) 2008 BeatIndex
4
GTAA – Bottom Up Approach
Alpha forecasts for fundamental data on individual stocks in an ETF are combined using their weightings to generate a market-cap weighted average forecast for the entire ETF. The process is repeated for each ETF in the universe of targeted ETFs, whether they are for country markets or industry specific to an ETF. ETFs are ranked based on their under-valuation percentage and then the top and bottom quintiles picked to go long and short. Fundamental insight: cheap beats expensive more than it should.
A tacit assumption is made that mean-reversion will happen over time.
A separate set of forecasts is done for the risk or volatility for each ETF in our universe. Using a portfolio optimizer, the alpha forecasts are combined with the risk forecasts to generate the proportion or weights of recommended ETFs to be held. This portfolio could be optimized to maximize the expected information ratio relative to either holding cash or relative to holding a benchmark, such as the Russell 3000 US Stock Index (IVW).
(c) 2008 BeatIndex 5
Pros/Cons of the Bottom Up Process
Takes into account all available information for all stocks belonging to that ETF, and therefore its industry group. Reflects a synthesis of expectations and experience of all industry analysts that follow each constituent stock of an ETF in detail. Fails to work during infrequent times of hundred-year floods even if there is no equity bear-market
such as during the Internet bubble, when outrageously expensive tech stocks kept getting more expensive or in the October, 1987 stock market crash, when all sectors tanked together.
(c) 2008 BeatIndex
6
Example of Bottom Up Approach
Let us assume we believe that expected EPS growth rate is a statisticallysignificant priced fundamental factor for a US stock in our investment universe of the Russell 3000 stock index.
Other popular factors include Value/Growth or Momentum
Investors pay up (accord a high Price/Sales multiple) to any stock in this universe sporting a high expected EPS growth rate. They penalize and pay a lower P/S to a stock with a low expected EPS growth rate. Having tested and verified this belief, we gather forecasts on expected EPS growth rate for each individual stock in any target ETF, e.g. IGN, and weight them by their proportional market capitalizations in the ETF, to generate a weighted-average forecast for the entire ETF. A tacit assumption is made that mean-reversion will happen over time, and so the under-valuation or over-valuation shown by an ETF will correct itself.
A very pricey ETF will become cheaper and a very undervalued ETF will become dearer as the information filters through to all investors in that ETF.
The time-interval over which the mean-reversion is expected to happen determines the time-horizon of the portfolio re-allocation.
(c) 2008 BeatIndex
7
Universe of ETFs
iShares Cohen & Steers Realty Majors Index Fund iShares Dow Jones Transportation Average Index Fund iShares Dow Jones U.S. Aerospace & Defense Index Fund iShares Dow Jones U.S. Basic Materials Sector Index Fund iShares Dow Jones U.S. Broker-Dealers Index Fund iShares Dow Jones U.S. Consumer Goods Sector Index Fund iShares Dow Jones U.S. Consumer Services Sector Index Fund iShares Dow Jones U.S. Energy Sector Index Fund iShares Dow Jones U.S. Financial Sector Index Fund iShares Dow Jones U.S. Financial Services Index Fund ICF IYT ITA IYM IAI IYK IYC IYE IYF IYG iShares Dow Jones U.S. Pharmaceuticals Index Fund iShares Dow Jones U.S. Real Estate Index Fund iShares Dow Jones U.S. Regional Banks Index Fund iShares Dow Jones U.S. Technology Sector Index Fund iShares Dow Jones U.S. Telecommunications Sector Index Fund iShares Dow Jones U.S. Utilities Sector Index Fund iShares FTSE EPRA/NAREIT North America Index Fund iShares FTSE NAREIT Industrial/Office Index Fund iShares FTSE NAREIT Mortgage REITs Index Fund iShares FTSE NAREIT Real Estate 50 Index Fund iShares FTSE NAREIT Residential Index Fund iShares FTSE NAREIT Retail Index Fund iShares Nasdaq Biotechnology Index Fund iShares S&P North American Natural Resources Sector Index Fund iShares S&P North American Technology Sector Index Fund iShares S&P North American Technology-Multimedia Networking Index Fund iShares S&P North American Technology-Semiconductors Index Fund IEO IEZ iShares S&P North American Technology-Software Index Fund IGV IHE IYR IAT IYW IYZ IDU IFNA FIO REM FTY REZ RTL IBB IGE IGM IGN IGW
iShares Dow Jones U.S. Healthcare Providers Index Fund
iShares Dow Jones U.S. Healthcare Sector Index Fund iShares Dow Jones U.S. Home Construction Index Fund iShares Dow Jones U.S. Industrial Sector Index Fund
IHF
IYH ITB IYJ
iShares Dow Jones U.S. Insurance Index Fund
iShares Dow Jones U.S. Medical Devices Index Fund iShares Dow Jones U.S. Oil & Gas Exploration & Production Index Fund iShares Dow Jones U.S. Oil Equipment & Services Index Fund
IAK
IHI
Benchmarks:
iShares Russell 3000 Index Fund iShares S&P 500 Index Fund
IWV IVV
(c) 2008 BeatIndex
8
Alternative Ways to Do Bottom-Up
Another bottom-up approach uses the earnings and dividend forecasts for an individual stock in the ETF as inputs to a proprietary dividend discount model (DDM) for generating the expected return on that one stock. Then, expected returns for all stocks in the ETF are aggregated to generate a bottom-up estimate for the expected return for the entire ETF. Yet another approach may be to estimate a probability of achieving an equity risk premium and a bond risk premium using logistic regression of time series of past returns.
It is assumed that a high equity risk premium would mean-revert to a low value Investors are assumed to fluctuate between over pessimism and optimism Leads to contrarian investment positions.
(c) 2008 BeatIndex
9
ETF Portfolio Selection
Whatever be your proprietary sauce to arrive at the fair market value (FMV) of an ETF, once you have all of the FMVs for your universe of ETFs, you should rank order them, with the most undervalued down to the most overvalued. A simple ETF Portfolio Selection then boils down to simply going long the most under-valued quintile (bottom 20%) of ETFs, and going short the most over-valued quintile (top 20%) of ETFs, in equal dollar proportions. A somewhat more sophisticated approach would be to do a meanvariance optimization to arrive at the different unequal dollar holdings vector for the long-short ETFs portfolio. This represents the lowest-risk operating point for the most possible expected return for a possible portfolio on the efficient frontier.
Bottom-up GTAA ETFs model results on August 30, 2008
ITB
IHF
iShares Dow Jones U.S. Home Construction Index Fund
iShares Dow Jones U.S. Healthcare Providers Index Fund iShares Dow Jones U.S. Insurance Index Fund iShares Dow Jones U.S. Industrial Sector Index Fund iShares Dow Jones U.S. Aerospace & Defense Index Fund iShares Dow Jones U.S. Consumer Services Sector Index Fund iShares Dow Jones U.S. Telecommunications Sector Index Fund iShares Dow Jones U.S. Oil & Gas Exploration & Production Index Fund iShares Dow Jones U.S. Technology Sector Index Fund iShares Dow Jones U.S. Healthcare Sector Index Fund iShares Cohen & Steers Realty Majors Index Fund iShares Dow Jones U.S. Pharmaceuticals Index Fund iShares FTSE NAREIT Mortgage REITs Index Fund iShares Nasdaq Biotechnology Index Fund
Go long these ETFs:
IAK IYJ ITA IYC IYZ
IEO
Go short these ETFs:
IYW IYH ICF IHE REM IBB
For more up-to-date signals on ETFs and models, check out http://www.beatindex.biz/html
(c) 2008 BeatIndex 11
August 30, 2008
(c) 2008 BeatIndex
12
What should ETF investors do now?
Pick a set of ETFs to go long and short from the model results presented
Use further due diligence or insight into industry group expectations that the model does not capture e.g. market expectations of Insurance (IYK) growing at160%+ average expected EPS growth rate, are unlikely to be realized.
If managing risk quantitatively, then optimize the portfolio holdings using past risk characteristics of these ETFs to be on the efficient frontier. Visit http://www.beatindex.biz/html regularly for updated bottom-up GTAA model signals on ETFs
(c) 2008 BeatIndex
13