This thesis consists of three essays that study three interdependent topics: microstructure foundation of volatility clustering, inefficiency of information diffusion and jump detection in high frequency financial time series data. Volatility clustering, with autocorrelations of the hyperbolic decay rate, is unquestionably one of the most important stylized facts of financial time series. The first essay forms Chapter 1 which presents a market microstructure model that is able to generate volatility clustering with hyperbolic autocorrelations through traders with multiple trading frequencies using Bayesian information updating in an incomplete market. The model illustrates that signal extraction, which is induced by multiple trading frequency, can increase the persistence of the volatility of returns. Furthermore, it is shown that the local temporal memory of the underlying time series of returns and their volatility varies greatly with the number of traders in the market. The second essay, Chapter 2, presents a market microstructure model showing that an increasing number of information hierarchies among informed competitive traders leads to a slower information diffusion rate and informational inefficiency. The model illustrates that informed traders may prefer trading with each other rather than with noise traders in the presence of the information hierarchies. Furthermore, it is shown that momentum can be generated from the trend following behavior pattern of noise traders. I propose a new nonparametric test based on wavelets to detect jump arrivals in high frequency financial time series data, in the third essay, Chapter 3. It is demonstrated that the test is robust for different specifications of price processes and the presence of market microstructure noise and it has good size and power. Further, I examine the multi-scale jump dynamics in U.S. equity markets and the findings are as follows. First, the jump dynamics of equities are entirely different across different time scales. Second, although arrival densities of positive jumps and negative jumps are symmetric across different time scales, the magnitude of jumps is distributed asymmetrically at high frequencies. Third, only twenty percent of jumps occur in the trading session from 9:30AM to 4:00PM, suggesting that jumps are largely determined by news rather than liquidity shocks.