This dissertation comprises three articles which are covering various aspects in the area of dynamic asset allocation, i.e., how to control an investor's portfolio composition over time in light of changing market conditions and various asset classes with differing return and risk profiles. In the first article, we propose the advancement of the traditional value-based rebalancing framework by setting the bandwidth weights in a non-static manner depending on current market characteristics, namely the relative price and volatility levels. The empirical analysis reveals significant excess returns in comparison to a buy and hold strategy, an idealized SAA strategy, as well as a rebalancing strategy with static bandwidths but otherwise comparable characteristics. The proposed approach also proves to be robust in various subsamples. In the second and third article, we break new ground by explicitly incorporating the risk dimension in the dynamic asset allocation process. In the second article, we present a rebalancing approach which applies the various asset classes' risk contributions to control the rebalancing process during the investment period. In strong contrast to traditional value-based rebalancing strategies, the resulting risk contribution strategies are capable of triggering rebalancing procedures based on deviations in an asset class's stand-alone volatility or correlation to the portfolio's remaining asset classes. Thus, this method lets investors closely maintain the asset classes' initial risk contributions. In the third article, we develop another approach which applies the loadings of the investor's portfolio on various key risk factors as indicators for triggering the rebalancing process. We implement the approach by monitoring the loadings on the risk factors interest rate, term spread, credit risk, equity premium, and volatility. We further define bandwidths for every risk factor loading. Once the effective loading.

This dissertation studies three different topics in empirical finance, specifically, portfolio management, short selling constraints and stock price informational efficiency, and one of the puzzling calendar anomalies: turn-of-the-month effect. The first chapter studies whether educational endowments earn superior returns. This is an interesting question, given the strong returns earned by some legendary endowments (e.g., Yale under the management of David Swensen), which has led to the widespread adoption of the so-called endowment model of investing. Using NACUBO/Commonfund data from 1991 to 2011, Brad M. Barber (UC Davis) and I analyze the returns earned by US educational endowments using simple style attribution models pioneered by Sharpe (1992). We document that for the average endowment, models with only public stock and bond benchmarks explain virtually all the time-series variation in returns, yield no alpha, and generate sensible factor loadings. Elite institutions perform better than public stock and bond benchmarks because of large allocations to alternative investments. We found no evidence that manager selection, market timing, and tactical asset allocation generate alpha. The second chapter uses the event of short selling ban removal in China in March, 2010 to study the relation between short selling and stock returns. First, I document that an increase in short interest predicts negative future returns, indicating that short sellers are informed about future stock returns. The long-short portfolio that buys stocks with no increase in short interest and shorts stocks with an increase in short interest earns a daily return of 0.085% (t=3.97). Second, consistent with the prediction of the Diamond and Verrecchia (1987) model, I find that the reduced short sale constraint leads to smaller price adjustments in response to earnings surprises. Specifically, I document that the price reaction to earnings announcements during the period that allows short selling is 67% lower than the price reaction during the period in which short selling is banned. In combination, these results indicate that short sellers play an important role in setting prices in financial markets. In the last chapter, Nathan George (UC Berkeley), Ethan Namvar (UC Berkeley), and I study the turn-of-the-month effect (TOM)--stocks have significantly higher returns during the period spanning from the last trading day of the previous month to the third trading day of the current month than during other trading days. Specifically, using the 13F institutional ownership data over the last three decades, we study the cross-sectional difference of the TOM effect across stocks held by different investors. First, we confirm the existence of the TOM effect in the stock market across stocks with different institutional ownership. Second, we document two patterns: (1) For stocks mainly held by individuals, the stock return out-performance during the TOM period mainly comes from the last trading day of the previous month; and (2) For stocks mainly held by institutions, the TOM effect in raw returns is evenly distributed across each day in that period, and that effect is completely explained by their exposures to the market. Furthermore, for stocks with high institutional ownership, the three days leading up to the last trading day of a month exhibit a significantly positive abnormal return compared to those of the other days. We propose that the difference in the trading behaviors of individuals and institutions may explain this dispersion.

The author reports on situations experienced by investors which clearly break the rules and norms of accepted rationality, for example when a fall in the price of an asset is accompanied, contrary to the classical hypotheses, by a fall in demand. He outlines a general theory for the fields of choice and the marginal rates of substitution within the framework of a specific referential of the economic subject, of a psychological nature and structured by time-values, and the framework of a limited rationality. In this important and timely book, Blanqué made a powerful claim for positive investment in order to avoid fanciful illusions. He assigned it the task of understanding what is happening in today's world. Blanqué argued that the investment world contains many elements of a fairy tale and showed that a mere confrontation with reality leaves its mark on the impressive procession of theoretical, sacred cows, of established beliefs and truths. As does the confrontation with crises.

Artificial intelligence (AI) has grown in presence in asset management and has revolutionized the sector in many ways. It has improved portfolio management, trading, and risk management practices by increasing efficiency, accuracy, and compliance. In particular, AI techniques help construct portfolios based on more accurate risk and return forecasts and more complex constraints. Trading algorithms use AI to devise novel trading signals and execute trades with lower transaction costs. AI also improves risk modeling and forecasting by generating insights from new data sources. Finally, robo-advisors owe a large part of their success to AI techniques. Yet the use of AI can also create new risks and challenges, such as those resulting from model opacity, complexity, and reliance on data integrity.