"Tax-Efficient Asset Management"
Prof. Clemens Sialm (University of Texas), and Prof. Hanjiang Zhang (Washington State University)
We investigate the relation between tax burdens and mutual fund performance from both a theoretical and empirical perspective. The theoretical model introduces heterogeneous tax clienteles in an environment with decreasing returns to scale and shows that the equilibrium performance of mutual funds depends on the size of the tax clienteles. Our empirical results show that the performance of U.S. equity mutual funds is related to their tax burdens. We find that tax-efficient funds do not only exhibit superior after-tax performance, they also exhibit superior before-tax performance due to lower trading costs, favorable style exposures, and better selectivity.
"Hedge Funds and Financial Intermediaries"
Prof. Magnus Dahlquist (Stockholm School of Economics), and Valeri Sokolovski (HEC Montréal)
In this project we will examine the link between hedge fund returns and financial intermediaries. There are at least two, non-mutually-exclusive, channels through which financial intermediaries, such as commercial and investment banks, impact hedge fund returns; direct and indirect. The first, direct, effect manifests itself in a hedge fund’s prime brokerage relationship. Prime brokers provide their clients with many services including clearing, securities lending and financing. The second, indirect, impact of financial intermediaries on hedge funds is through their effect on asset prices and risk premiums. Recent research finds that shocks to the intermediaries’ aggregate risk-bearing capacity can explain the cross-section of expected returns of multiple asset classes. Given that many hedge funds are essentially portfolios of such assets, one would expect them to have high exposure to the health of the intermediary sector. We will examine both effects. Methodologically, we will use standard approaches in the literature (such as panel regressions, portfolio sorts, and event studies) and recent techniques based on network theory. The structure of our enhanced hedge fund dataset allows us to identify the hedge-fund-broker relationships at each point in time, which makes it well-suited for network analyses.
"The Portfolios and Financial Decisions of High Net Worth U.S. Households"
Prof. Enrichetta Ravina (Kellogg School of Management)
The objective of our project is to describe the portfolio positions and transactions of a sample of high net worth US households over the period from 2005 to 2009. We will analyze the asset classes they invest in, their performance, the degree of diversification, the amount and timing of their rebalancing activity, and the characteristics, risk, liquidity, and tax status of their investments. We observe detailed transactions and portfolio positions data over time, and fees, interest, dividend income and capital gain for each security. While the strength of our data set is in the comprehensive view of the financial assets it offers, we also observe investments in real estate, and, in some cases, the value and characteristics of the business(es) the household owns directly. For each household, we also observe the different tax entities through which it invests, the different managers and advisors it employs, if any, as well as whether different households are part of the same family, which gives us a glimpse of the inter-family investment dynamics. Our time frame will also allow us to study in detail the behavior of these investors during the boom and the subsequent financial crisis.
"New Machine Learning Tools"
Prof. Fabio Trojani (University of Geneva), and Prof. Simon Scheidegger (HEC Lausanne)
This project develops new machine learning tools based on Gaussian process regression, which are customized for the nonparametric analysis of large panels of asset returns characterized by high-dimensional asset features. We develop various technologies for approximating and estimating high-dimensional asset pricing relations between returns and characteristics, and to quantify the resulting epistemic uncertainty under a nonparametric approach. To this end, we also develop efficient codes and parallelized algorithms for (i) estimating and testing high-dimensional risk-return tradeoffs and (ii) automatically selecting asset characteristics with incremental out-of-sample explanatory power for asset returns. We then test and train in various applications our methodologies using big data sets of international asset returns.