In this dissertation, I study the financial systemic risk from firm-level perspectives. Chapter 1 investigates a breakdown of the total financial system risk into individual contributors and sources. Chapter 2 studies a theoretical model about the active balance sheet management of individual bank in securitization. Chapter 3 and 4 present empirical evidence about securitization asset choices of banks when they face different constraints. Chapter 5 provides a brief summary of findings in this dissertation. In chapter 1, I propose a novel systemic importance (SI) index that tracks the contribution of a financial institution to the total financial system risk. That risk measure can be decomposed into idiosyncratic and spillover risk contribution to further study the risk characteristics of each firm. Using equity return data from 1965 to 2018, I find two important results. First, the spillover risk can account for approximately 80% of the aggregate financial system risk, which emphasizes the importance of contagion risk as a major amplification mechanism of shocks during a systemic event. Second, a portfolio of the top 20 most systemically important financial institutions (SIFIs), ranked by SI index, earns a significantly lower risk-adjusted return than their counterparts. This substantial equity funding cost advantage of approximately 4% per year on average implies that the ex-ante implicit government guarantee for the “too-important-to-fail” is priced by the market. In chapter 2, I develop a theoretical model that features two benefits of securitization. First, banks can reduce idiosyncratic risks and enhance risk-absorbing capacity by converting a fraction of their risky investments into securitized assets. Second, securitized assets require less regulatory capital, helping banks obtain a higher leverage without breaking the regulation. This chapter studies effects of the two motives above, namely risk-transferring and regulatory arbitrage, on bank portfolio choices. My analytical results predict that banks would securitize safer loans and retain only higher-risk, higher-return assets that justify their regulatory capital cost. In chapter 3, I analyze new data points in the recently revamped HMDA data to examine mortgage securitization decision choices and motives of all non-exempt banks in the US. Combining with the bank-level data from Call Reports, I find that capital-constrained banks retain riskier loans and involve more in the securitization market to optimize return on capital and keep regulatory ratios in control. On the other hand, risk-constrained banks use securitization mainly for the purpose of risk and liquidity improvement. When putting together, risk transferring seems to dominate regulatory arbitrage as the main reason banks engage in securitization. Chapter 4 serves as a complementary case study to Chapter 3, in which I investigate the mortgage loan approval and securitization decision of PNC Bank. There are three interesting findings: First, the bank uses third-party automated underwriting systems to originate over 90% of its conforming residential mortgage loans and then sell more than 70% of them. Second, the bank retains safer loans on balance sheet, which emphasizes the role of securitization as a risk-transferring mechanism. Third, compared to a non-depository financial institution (shadow bank), a traditional commercial bank like PNC behaves differently and shows a clear presence of active securitization management. With a stable deposit funding channel, PNC is able to originate jumbo loans at a higher approval rate, retain more loans on balance sheet, and selectively choose to sell off riskier loans.