My dissertation has two chapters. The first chapter is titled "Analysts' reinititations of coverage and market underreaction". I study a signal which has been completely ignored by the literature so far: reinitiations of coverage. Reinitiations are defined as the resumption of coverage of a stock by a broker after more than six months of interruption. They are associated with a significant short-term market response, in particular when the same analyst is assigned to the stock. However, this market response is incomplete. Interestingly, the price patterns that follow the issuance of regular upgrades of recommendation and reinitiations differ significantly, and this paper can help us better understand the phenomena of market underreaction and overreaction. Prices adjust quickly after a regular upgrade, while reinitiations are followed by a sustained price increase in the following six months. I assess the economic magnitude of this initial underreaction by setting up a trading strategy. Reinitiations of coverage are the only type of recommendation that delivers significant positive abnormal returns after transaction costs with a three- and six-month investment horizon. I investigate several explanations in relation to gradual information diffusion, limited attention and changes in firm profitability. Portfolio sorts on proxies for market attention indicate that firms subject to a lower level of initial attention experience the strongest cumulative abnormal returns. Reinitiations also coincide with improvements in firms' profitability.

The second chapter is co-authored with Ivo I. Welch and is titled "Are Economic Tracking Portfolios (ETP) useful? And What Fundamentals Are Driving Stock Prices?". Our paper shows that equity-based economic tracking portfolios, ([BGL89], [Lam01]) constructed from the ten Fama-French industry equity portfolios and three bond portfolios could not usefully track selected macro-series (inflation, industrial production growth, consumption, real estate, exchange rates, and oil) any better than simpler benchmarks (a constant, the T-bill, the S&P500, and/or a single stock). This suggests that the in-sample ability of ETPs to track macro-series was likely just overfitting. In most cases, the zero constant and/or Treasury Bill tracked best. This reinforces perhaps the most fundamental mystery in finance: what economic variables, if any, are really driving stock prices ([Rol88])?