Not logged in.
Quick Search - Contribution
|Title||Timing Models for Factor Investing|
|Institution||University of Zurich|
|Faculty||Faculty of Business, Economics and Informatics|
|Number of Pages||55|
|Zusammenfassung||recent years factor investing has been picked up by the financial industry and is often marketed as so called Smart Beta investing. Smart Beta products try to capture risk premia and are offered to the investor such that by investing in them he can gain exposure to apparently very specific risks in exchange for a risk premia he will eventually receive. However, there is a problem with these products: it can be observed that these risk premia are not constant over time (see for instance Gulen et al. (2011)). So first, the timing of these risk premia is still left to the investor. Second, as the index providers are more concerned about the outperformance of their products than on how well they capture an individual risk premia, the investor may not get exactly the kind of risk exposure he wants. In this thesis we thus ask: Do these premia really exist? How does their dependence structure look like? Are the magnitudes of the premia predictable and can these predictions be used to construct a dynamic factor portfolio which outperforms the market? Finally, are they a better suited to be used in an asset allocation than traditional asset classes like the sector based classification? In this thesis we will thus look at some of the most well-known factor premia. We will give a short summary of the existing literature, construct factor portfolios, backtest them over the time period from 1995 to 2015 and look at their performance and dependence structure. We will also compare these results to the performance and dependence structure of more traditional asset class indices, the sector based indices. Doing so we find that especially the long period momentum and low volatility factors offer sustained risk adjusted outperformance when looking at the Sharpeand Sortino Ratio, while the value premium seems to be time-varying and the size premium does not even exist anymore. We also find that that the Health Care and Consumer Staples sector indices have highly outperformed over our sample period and of course there have been sub-periods where other sectors have outperformed. Furthermore find that the factor excess returns offer diversification but that their dependence structure also changes over time. Again, this is true for the sector indices as well. With respect to return predictability, using a linear regression we find that some leading fundamental indicators seem indeed to be related to future factor excess returns more than they are related to future sector returns. However using the fundamental indicators to make some actual predictions about future factor returns does not provide any benefit for constructing a dynamic factor portfolio. In fact we find that using simple historical risk/return based allocation techniques or even the simple equal weighted approach work best. IV|