In this thesis, the phenomenon of financial market integration is studied both empirically and theoretically. The empirical study confirms the trend of market integration, while the theoretical study provides significant financial implications on market integration. Firstly, a time series representing the developed financial markets' segmentation from 1973-2012 is studied. The time series reveals an obvious market integration trend. To further uncover features of this time series, we divide it into seven windows and generate seven visibility graphs. The measuring capabilities of visibility graphs provide a means to quantitatively analyze the original time series. It is found that important historical incidents that influence market integration coincide with variations in the measured graphical node degree. Through the measure of neighborhood span, the frequencies of historical incidents are disclosed. Moreover, it is also found that large "cycles" and significant noise in the time series are linked to large and small communities in the generated visibility graphs. For large cycles, how significantly historical incidents affected market integration is distinguished by the density and compactness of the corresponding communities. Secondly, a novel model of sequential arbitrage networks is developed to theoretically investigate the process of financial market integration. At every trade moment, an arbitrage network is built by considering exchanges as nodes and arbitrageurs' trades as edges. Taking each exchange's asset pricing as the state of a node and regarding price convergence as a sign of financial market integration, the state movement and the market integration are studied through a number of sequential trade moments. Two types of arbitrage networks are considered: one-trade mode and multi-trade mode. For the sequential one-trade arbitrage network, the connectivity of the network is further studied. The results support the Law of One Price hypothesis and are consistent with the trend that financial markets all over the world become more and more tightly connected with the deepening of financial globalization. For the sequential multi-trade arbitrage network, some sufficient conditions for the integration and low bounds for the speed of the integration are given. The results indicate that the financial market integration would be achieved as long as the arbitrageurs liquidate assets across all segmented markets constantly and the arbitrageurs' desire for profits has positive effects on the integration. Finally, profit distributions of the one-trade mode and the multi-trade mode in arbitrage networks are studied. It turns out that both investors in traded exchanges and arbitrageurs benefit from the trades. Moreover, the multi-trade mode brings competition among arbitrageurs and asset liquidity among exchanges, both of which make the profit distribution between the investors and the arbitrageurs differ from that of the one-trade mode. The competition helps the investors take a larger share of the total profit, while the asset liquidity works in an opposite way. How strong their influences are relies on the arbitrageurs' selections of the traded exchanges. In brief, this thesis complements the existing literatures of market integration and also provides meaningful insights for all investors and policy makers. Also, the model of arbitrage networks could be used to predict how market integration would be affected by financial market regulations and rules and how investors and arbitrageurs would benefit from market integration. In addition, we bring concepts from graph theory and switched systems as new perspectives for analyzing arbitrage networks.