Three Essays on Financial Economics

This dissertation is comprised of three main chapters that correspond to three papers I authored during my PhD studies. Each one is formatted like a separate article for publication in a scholarly journal. The dissertation discusses two realities of financial economic, high frequency trading and the shadow economy, which, despite their prevalence, are difficult to study due to their hidden nature. It attempts to shed more light on these two phenomena by using machine learning techniques and laboratory experiments. In addition to findings that contribute to the related literature and guide policymakers, it also contributes methodologically to the study of these difficult-to- measure phenomena. The first chapter develops a probabilistic model that can be applied to order book data for identification of high-frequency trading (HFT). The model was created using machine learning techniques and can make accurate intraday identifications. Because there is no widely accepted model for identifying HFT, scholars have used a variety of proxy to estimate HFT, resulting in inconsistent conclusions. Because our identification is based on data that academics have access to, it is expected to provide greater consistency and reproducibility to future HFT research. The fuzzy logic we used also provides policymakers with more flexible identification. The data for the first chapter came from the French capital market. We created a method and a classification model capable of accurately distinguishing HFT. Reverse engineering was used to turn the model into an understandable regression tree with the same level of predictability, thus resolving the issue of the model's indecipherability. The second chapter delves into the decision to engage in the shadow economy and evade taxes. This is done in the context of expected utility maximization and behavioral economics. We present a laboratory experiment in which participants on both the supply and demand sides of a hypothetical credit market invested and profited. Several simplified scenarios were examined to determine how those participants decided what type of economy to participate in and how they paid or avoided paying taxes. We investigated the role of semantic prosody in the framing effect in tax propensity and risk propensity. According to our findings, semantic prosody increases tax propensity and moderates the framing effect in risk propensity. Finally, using the data, several econometric models were developed to examine the impact of experimental variables, policy variables, and demographic variables on tax and risk propensity. In the third chapter, we carried out laboratory experiments to investigate the shadow economy and the development of the financial sector. We investigated the ways in which people's propensity to participate in the shadow economy is influenced by changes advancement in the financial sector. We examined the impact of a variety of shadow economy structures by employing a design that included both within-subject and between-subject components, with a primary emphasis on credit supply as a general proxy for financial development. According to the findings of our study, individuals are more likely to participate in the formal economy in situations where there is a higher level of financial development, even if this is associated with a higher tax rate. We also discovered that the structure of the shadow economy only has an effect on people's propensity toward it when financial development is sufficiently advanced to create certainty in the process of receiving credit.