Equity premium puzzle | EPFL Graph Search

The equity premium puzzle refers to the inability of an important class of economic models to explain the average equity risk premium (ERP) provided by a diversified portfolio of U.S. equities over that of U.S. Treasury Bills, which has been observed for more than 100 years. There is a significant disparity between returns produced by stocks compared to returns produced by government treasury bills. The equity premium puzzle addresses the difficulty in understanding and explaining this disparity. This disparity is calculated using the equity risk premium: The equity risk premium is equal to the difference between equity returns and returns from government bonds. It is equal to around 5% to 8% in the United States. The risk premium represents the compensation awarded to the equity holder for taking on a higher risk by investing in equities rather than government bonds. However, the 5% to 8% premium is considered to be an implausibly high difference and the equity premium puzzle ref

Essays in Information-Based Asset Pricing

Michael Hasler

I started my Ph.D. studies in the Fall 2008, a period ex-post perceived as being at the core of the Financial Crisis. At that time my ideas were vague and I struggled to find a good research topic. As surprising as it might appear, in one single week the discussions I had with my relatives provided me with enough ideas needed to write my thesis. Indeed, during that single week my relatives contacted me and complained about the fact that they had lost money invested in the financial market. They wanted me to advise them. I simply told them to invest their money in relatively safe bonds. My relatives did not realize how fruitful their behavior was for me. Indeed, after having advised them I wondered whether their willingness to gather information in bad times was optimal or pure irrationality. This issue is addressed in the first chapter of my thesis. The last two chapters investigate the impact of the willingness to gather more information in bad times than in good times on the dynamics of asset prices. My thesis answers two very important questions: When should investors acquire information given that information-processing is costly and what is the impact of the optimal information acquisition strategy on asset price dynamics? Consequently, the main topic of my thesis is the transmission of acquired information in financial markets. In the first chapter I consider a dynamic portfolio choice and costly information acquisition problem. Consistent with my relatives behavior, I show that it is optimal to gather more information in bad times than in good times. As the state of the economy worsens, investors optimally re-balance their portfolio towards safer investment and spend more and more money to gather more and more accurate information regarding future market moves. This result derived in a purely rational setup helps to understand the empirical observation that people tend to pay fluctuating attention to news, focusing more in downturns than in upturns. To the best of my knowledge, this paper is the first to consider costly information acquisition in a dynamic setting. The main contribution is to provide insights on how investors should optimally allocate their wealth to the financial market on the one hand and to information market on the other hand given current market conditions. In the second chapter (with Daniel Andrei) we model the optimal information acquisition strategy obtained in the first chapter in a general equilibrium framework and we show that this strategy (fluctuating attention paid to news) has very interesting implications on the dynamics of asset prices. Our main contribution is to highlight a J-shaped relationship between attention and volatility. By constructing an empirical measure of attention to news, we show that the theoretical prediction of our model holds in the data. Moreover, we are the first to provide a theoretical foundation for the fact that the slope of the term structure of equity premia is pro-cyclical. In the third chapter I investigate the impact of fluctuating attention to news in a multi-asset general equilibrium framework. My main contribution is to show that fluctuating attention to news generates return and volatility spillover effects among financial markets. As a bad shock hits one market, investors raise their attention to that particular market. This implies a simultaneous increase in each market volatility and in each cross-market correlation, a phenomenon called financial contagion. Moreover, I document that the interconnection between attention and uncertainty generates persistence in cross-market correlations. This very interesting result proves that persistence can be generated through Bayesian learning. Finally, I am the first to provide a theoretical foundation for the observed shape of the term structure of correlation. Consistent with empirical findings, I show that fluctuating attention implies an upward sloping term structure of correlation where the shorthand is wider than the longhand.

EPFL2013

The total benefit of alternative assets to pension fund portfolios

Anna Slavutskaya

Pension funds only quite recently have explored alternative assets, prodded by financial crises that devastated equity returns and led to low bond returns. We assess the addition of alternative assets to pension fund portfolios in terms of the total benefit derived from diversification, addition of positive skewness, and the elimination of left tails in returns. During 1994-2012, adding portfolios of hedge funds produced significantly higher total benefits than adding real estate, commodities, foreign equities, mutual funds, funds of funds, as well as some counter cyclical and non-cyclical assets. Conditioning on past total benefits improves the out-of sample performance even further. (C) 2016 Elsevier B.V. All rights reserved.

Elsevier Science Bv2016

Essays in Empirical Asset Pricing

Alexis Arilès Marchal

This thesis consists of three applications of machine learning techniques to empirical asset pricing.In the first part, which is co-authored work with Oksana Bashchenko, we develop a new method that detects jumps nonparametrically in financial time series and significantly outperforms the current benchmark on simulated data. We use a long short-term memory (LSTM) neural network that is trained on labelled data generated by a process that experiences both jumps and volatility bursts. As a result, the network learns how to disentangle the two. Then it is applied to out-of-sample simulated data and delivers results that considerably differ from the benchmark: we obtain fewer spurious detection and identify a larger number of true jumps. When applied to real data, our approach for jump screening allows to extract a more precise signal about future volatility.In the second part, which is co-authored work with Oksana Bashchenko, we develop a methodology for detecting asset bubbles using a neural network. We rely on the theory of local martingales in continuous-time and use a deep network to estimate the diffusion coefficient of the price process more accurately than the current estimator, obtaining an improved detection of bubbles. We show the outperformance of our algorithm over the existing statistical method in a laboratory created with simulated data. We then apply the network classification to real data and build a zero net exposure trading strategy that exploits the risky arbitrage emanating from the presence of bubbles in the US equity market from 2006 to 2008. The profitability of the strategy provides an estimation of the economical magnitude of bubbles as well as support for the theoretical assumptions relied on.In the third part, I propose a new tool to characterize the resolution of uncertainty around FOMC press conferences. It relies on the construction of a measure capturing the level of discussion complexity between the Fed Chair and reporters during the Q&A sessions. I show that complex discussions are associated with higher equity returns and a drop in realized volatility. The method creates an attention score by quantifying how much the Chair needs to rely on reading internal documents to be able to answer a question. This is accomplished by building a novel dataset of video images of the press conferences and leveraging recent deep learning algorithms from computer vision. This alternative data provides new information on nonverbal communication that cannot be extracted from the widely analyzed FOMC transcripts. This chapter can be seen as a proof of concept that certain videos contain valuable information for the study of financial markets.