Franzoni, Francesco Barbon, Andrea 2020-02-19 My dissertation consists of three chapters, each of which focuses on a different area of research in asset pricing. The first chapter deals with the informational role of brokerage firms during fire sales in the equity market. The second chapter exploits the ETF program by the bank of Japan as a quasi-natural experiment to measure the slope of the equity demand curve. The last chapter presents the application of a novel machine learning model, based on an attention mechanism, to high-frequency data from the Nasdaq. In the first paper, co-authored with F. Franzoni, M. di Maggio, and A. Landier, we use trade-level data to study whether brokers play a role in spreading order flow information in the stock market. We focus on sizeable portfolio liquidations from large investment funds, which we name "fire sales", resulting in significant temporary price drops for the liquidated stocks. Next, leveraging on our unique dataset, we identify the brokers who intermediate these fire-sale trades. We show that these brokers’ clients are more likely to predate on the liquidating funds than to provide liquidity. Predation leads to profits of about 25 basis points over 10 days, on average, and increases the liquidation costs of the distressed fund by 40%. This evidence suggests a role of information leakage in exacerbating fire sales. The highlighted mechanism might be a concern to regulators as well, since it can exacerbate the costs associated with fire sales, especially during times of scarce liquidity. In the second paper, co-authored with V. Gianinazzi, we focus on the Japanese equity market to study a peculiar intervention by the Japanese central bank. Since the introduction of its quantitative and qualitative easing program in 2013, indeed, the Bank of Japan has been increasing its holdings of Japanese equity through large-scale purchases of index-linked ETFs to lower risk premiums. We exploit the cross-sectional heterogeneity of the supply shock to identify a positive and persistent impact on stock prices, consistent with a portfolio balance channel. The evidence suggests that long-run demand curves for stocks are downward sloping with unitary price elasticity. Moreover, we are able to quantify the effect of the intervention, showing that the purchases of ETFs tracking the price-weighted Nikkei 225 generate pricing distortions relative to a value-weighted benchmark. The third paper starts from the observation that, even though machine learning methods are able to deliver superior forecasting accuracy, they can hardly be used to make inference. To overcome this limitation, I propose an encoder-decoder neural network augmented with an attention-based mechanism that can autonomously learn to identify the most critical regions of the input data. I first train the model using high-frequency message data from the NASDAQ and show that it outperforms other state-of- the-art models in forecasting future transaction prices. Then, I develop a methodology that uses the attention mechanism to make inference on the relative share of information content of market orders versus limit orders, concluding that the most informative events are executions of market orders while submission and cancellations of limit orders are less relevant. Finally, I test the model’s behavior during the execution of real block orders from institutional investors, showing that it favors liquidity provision rather than front-running strategies. application/pdf https://susi.usi.ch/global/documents/318980 https://localhost:5000/ark:/12658/srd1318980 https://susi.usi.ch/documents/318980/files/2020ECO002.pdf eng info:eu-repo/semantics/altIdentifier/urn/urn:nbn:ch:rero-006-120842 info:eu-repo/semantics/altIdentifier/ark/12658/srd1318980 info:eu-repo/semantics/openAccess License undefined Asset pricing Demand shocks Institutional investors Monetary policy Market microstructure Machine learning info:eu-repo/classification/udc/33 Asset prices and demand shocks http://purl.org/coar/resource_type/c_db06