This dissertation investigates two significant areas in financial markets: the institutional dynamics of life insurance companies as major corporate bond investors, and the emerging challenges of large language models applications in finance. Together, these research streams contribute to our understanding of both traditional financial institutions and innovative technological disruptions in the field.In the first two chapters, I examine the critical role of life insurance companies as the largest domestic investors in the corporate bond market. Chapter 1 analyzes the implications of life insurance companies outsourcing their investment decisions to external advisers. Life insurers collectively outsource nearly $1 trillion in bond investments to a relatively small number of common external advisers, raising important questions about systemic risk through portfolio concentration. My research finds that portfolio similarity is three times higher between insurer pairs with the same adviser, and 1.5 times higher between insurer-mutual fund pairs sharing an adviser. This interconnectedness has dual implications for systemic risk: while portfolio similarity amplifies fire sale risks when these institutions face common shocks (such as during monetary tightening cycles), it can also enhance financial stability when institutions face divergent shocks. During mutual fund outflows, insurers purchasing bonds sold by associated mutual funds provide a stabilizing effect on market prices. Chapter 2 (joint work with David Ng and Xing Zhou) studies the sources and frictions of alpha in insurance corporate bond portfolios. We study the performance of corporate bond portfolios of life insurers based on detailed daily portfolios that we construct from the combination of holdings and transaction data from regulatory filings. Our analysis reveals that, on average, life insurers’ portfolios do not outperform the broader market, although performance varies significantly across insurers and over time. We find that investments in illiquid bonds can yield higher returns and alpha; however, regulatory restrictions on managing the duration gap between assets and liabilities offset these potential gains. Building on my interest in financial market dynamics, Chapter 3 (joint work with Will Cong, Xing Huang, and Lawrence J. Jin) extends my research into the frontier of financial technology by examining behavioral biases in Large Language Models (LLMs) when making economic and financial decisions. This work addresses critical questions about the reliability of AI systems in financial applications: Do generative AI models exhibit systematic behavioral biases similar to humans? If so, how can these biases be mitigated? We conduct a comprehensive set of experiments—originally designed to document human biases—on prominent LLM families with variations in model version and scale. Our findings reveal that for experiments concerning the psychology of preferences, LLM responses become increasingly irrational and human-like as models become more advanced or larger. Conversely, for experiments concerning the psychology of beliefs, the most advanced large-scale models frequently generate more rational responses. We further explore various methods for correcting these behavioral biases and find that prompting LLMs to make decisions according to the Expected Utility framework appears most effective. This dissertation contributes to both traditional finance literature on institutional investors and the emerging field of AI applications in finance, offering insights that enhance our understanding of financial market stability and the potential limitations of technological innovation in the sector.